The Role of Selenium in Thyroid Hormone Action*

Home , Deiodinase, Iodothyronine deiodinase

0163-769X/92/1302-0207$03.00/0 Endocrine Reviews Vol. 13, No. 2 Copyright 0 1992 by The Endocrine Society Printed in U.S.A.

MARLA J. BERRY AND P. REED LARSEN Thyroid Division and Howard Hughes Medical Institute, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts 02115

I. Introduction molecular biological techniques. The latter permitted II. Biochemical Properties and Physiological Roles of Type biochemical analyses of the wild type selenocysteine- I and II Deiodinases containing deiodinase and a mutant enzyme containing III. Purification and Characterization of Type I Deiodinase cysteine, providing insight into the importance of this IV. Cloning of the cDNA for the Type I Deiodinase rare amino acid in the active site of the enzyme. Evidence V. Type I DeiodinaseContains Selenocysteine,Encoded by a TGA Codon is reviewed which indicated that the type II deiodinase VI. Nutritional Studies Suggesta Role for Selenium in Thy- is not a selenoprotein. The discovery that selenocysteine roid Hormone Metabolism is encoded by a UGA codon, normally a stop codon, led VII. SelenocysteineConfers Specific Properties on the Deiod- to the question: How is UGA encoding selenocysteine inaseEnzyme distinguished from UGA specifying termination? Studies VIII. EvidenceThat Type II DeiodinaseIs Not a Selenoprotein addressing this recognition process are presented. Fi- IX. Why is SelenocysteinePresent in Proteins? nally, the presence of selenium in the type I deiodinase X. New Insights into Protein Synthesis in Eukaryotes has important clinical implications for treatment of com- XI. Human Iodothyronine Deiodinaseand Clinical Implica- bined selenium and iodine deficiency, and these are ad- tions dressed. I. Introduction Thyroid hormone is essential for many biological processes, including normal development, growth, and me- ECENT progress in the field of thyroid hormone tabolism. It has been recognized for a number of years R:deiodination has led to a greater understanding, that most, if not all of the biological effects of thyroid not only of this critical process, but also of mechanisms hormone can be attributed to T3. The thyroid gland is more basic to protein synthesis in general. The cloning the sole source of thyroid hormones, and when iodine is of the type I deiodinase, demonstration that this enzyme sufficient, its main secretory product is T.+ Since the is a selenoprotein, and subsequent identification of se- identification of TB in human plasma by Gross and Pitt- lenocysteine insertion sequences in the noncoding por- Rivers in 1952 (3) and the demonstration of its greater tion of the gene have provided considerable insight into biological potency compared to Tq, the importance of the mechanism of incorporation of this unusual amino iodothyronine deiodination in thyroid hormone action acid into proteins. The purpose of this review is to discuss has received increasing attention. The subsequent dis- some of these advances and their implications. We begin covery that thioureylene drugs impaired both the phys- by providing a brief background of the deiodination iological effects of Tq and its metabolism was a major process and the discovery of the enzymes involved, fol- advance toward systematic evaluation of the role of lowed by a discussion of some of the extensive charac- deiodination in thyroid hormone function (4). Propyl- terization studies that have been performed on these thiouracil (PTU) was shown to block Tq to T3 conversion enzymes over the years. Comprehensive reviews of these and decrease Tq stimulation of hepatic a-glycerophos- earlier studies have been published previously (1,2). We phate dehydrogenase (GPD) (5), lending support to the next describe the expression cloning of the type I deiod- thinking, generally accepted at that time, that T3 was inase and its identification as a selenoenzyme, evidence the biologically active thyroid hormone. However, the of which was provided from nutritional, biochemical, and situation was complicated by the subsequent demonstration that PTU impaired certain T4-mediated processes, Address requests for reprints to: Marla J Berry, Ph.D., Thyroid while having minimal impact on others. In agreement Division, Brigham and Women’s Hospital, Harvard Medical School, 75 with the studies above, PTU reduced Tq stimulation of Francis Street, Boston, Massachusetts 02115. * This work was supported in part by NIH Grant DK-36256. GPD to about one-third of the level in untreated animals, 207

Downloaded from edrv.endojournals.org on June 22, 2006 208 BERRY AND LARSEN Vol. 13, No. 2 and serum T3 levels were decreased to a similar extent. in PTU sensitivity and tissue distribution discussed However, in the same animals, PTU had little or no above, these enzymes are distinguishable by their enzyme effect on either Tq stimulation of growth or its suppres- kinetics and substrate specificity, and more importantly sion of TSH secretion (6), both of which are pituitary- by their physiological roles and responses to thyroid specific processes. status (13, 14, 16, 20, 22). Both enzymes catalyze 5’- These results suggested one of two possible explana- deiodination and can use either T4 or reverse TS (rT3) as tions. Either Tq was biologically potent in pituitary but substrate, and both require reduced thiol as cofactor. not in other tissues, or TB was the active hormone, and However, type I deiodinase exhibits a strong preference Tq to TB conversion in the pituitary was not blocked by for rTg as substrate over T4 and follows a ping-pong PTU. Several lines of evidence eventually proved the reaction mechanism with thiol as second substrate, latter explanation to be correct. The first of these was whereas the type II deiodinase displays a slight prefer- the demonstration that suppression of TSH secretion ence for T4 over rT3, following a sequential reaction correlated both quantitatively and temporally with TB mechanism. The identity of physiologically important binding to pituitary nuclear receptors, not with serum T3 thiol cofactors are not known, but in vitro studies have (7). After injection of labeled Tq, virtually all of the shown that naturally occurring thiols, including gluta- labeled iodothyronine bound to nuclear receptors was TZ thione, dihydrolipoamide, glutaredoxin, and thioredoxin, (7, 8). Silva and Larsen (7) also showed that, after and the synthetic dithiol, dithiothreitol (DTT), can func- administration of either Tq or Ts to hypothyroid rats, tion in this capacity (23-27). The inhibition of type I the quantities of nuclear T3 were identical, and TSH was deiodinase by PTU is competitive with thiol cosubstrate suppressed to a similar extent. At about the same time, and uncompetitive with iodothyronine, due to the reac- it was shown that in different tissues, the contributions tion of this inhibitor with enzyme-substrate complex, but of serum Tq and Ta to nuclear TB differed (8, 9). This not with free enzyme. approach used [lz51]T4 and [13’I]T3 to follow the fate of The physiological role of type I deiodination is to the two iodothyronines in vivo. These studies showed provide a circulating source of T3 to the peripheral tis- that in pituitary a significant fraction of nuclear T3 was sues, derived from T4 produced by the thyroid gland. derived from serum Tq by local or intracellular deiodi- Type I deiodinase activity in liver and kidney increases nation, the remainder being contributed by serum T3, in hyperthyroid and decreases in hypothyroid animals whereas in liver, most of the nuclear TB derived from (28). In the thyroid cell, both TSH and the circulating serum T3. The nuclear thyroid hormone receptors in thyroid-stimulating immunoglobulin present in patients pituitary were shown to be approximately 80% saturated, with Graves’ disease increase type I deiodinase activity whereas in liver, receptors were only approximately 50% (29-35). The former is particularly prominent in iodine- saturated, the difference being accounted for by intra- deficient rats, in which enhanced thyroidal Tq to TS cellular T4 to T3 deiodination in the pituitary (8). The conversion may even contribute to an enhanced effi- next major advance was the finding that iopanoic acid ciency of total body T4 to T3 conversion (36). Further- inhibition of T4 to T3 conversion in pituitary blocked Tq- more, the elevated serum TB may itself stimulate the mediated suppression of TSH secretion and induction of conversion of Tq to TS in hyperthyroidism due to Graves’ GH synthesis. Iopanoic acid had no effect on the ability disease and thus contribute to the markedly greater of T3 to induce these responses, confirming that conver- sensitivity of Tq to T3 conversion in hyperthyroid than sion to T3 is required for these biological effects (10, 11). euthyroid humans, to inhibition by PTU (37). And finally, demonstration of PTU-insensitive Tq to T3 Type II deiodinase functions to regulate intracellular conversion in pituitary fragments, homogenates, and mi- TS levels in those tissues where T3 is most critical, the crosomes clearly showed that the deiodination process in brain, pituitary, brown fat, and placenta. The type II this tissue differed from that in liver and kidney (12-14). enzyme allows the pituitary to monitor circulating Tq, PTU-insensitive deiodination was subsequently demon- thereby regulating TSH secretion, and thus thyroidal Tq strated in cerebral cortex (15, 16), brown adipose tissue synthesis. In brown fat, type II deiodinase is regulated (17, 18), and placenta (19), and the ensuing years pro- by a-1-adrenergic stimulation (17, 38). During cold duced numerous studies elucidating the differences be- stress, increased type II activity produces the elevated tween the PTU-sensitive (type I) and -insensitive (type intracellular T3 required for thermogenesis, which is II) deiodinase pathways and enzymes (20, 21). mediated by T3-induced synthesis of the uncoupling protein, thermogenin (39,40). The response of type II deiod- II. Biochemical Properties and Physiological inase to thyroid status is opposite that of type I (15, 21). Roles of Type I and II Deiodinases When T4 is limiting, its rate of deiodination to TB in type The properties of type I and II deiodinases are sum- II tissues is increased, allowing the intracellular environ- marized in Table 1. In addition to the critical differences ment in these tissues to remain relatively euthyroid when

Downloaded from edrv.endojournals.org on June 22, 2006 May, 1992 SELENIUM AND THYROID HORMONE METABOLISM 209

TABLE 1. Properties of rat type I and II microsomal iodothyronine deiodinases

Type I Type II Substrate preference rTQ % T, > Tg T4 > rT1 Deiodination site Outer and inner ring Outer ring K, for rTg 1 x lo-70 3 x lo-$* K, for T, 2 x lo+ 1 x lo-= Ki for PTU[rTJ 1 x 10-6c -lop Ki for PTU[DTT] 5 x lo-” -10-3’ Ki for GTG 3 x 10-9d 3 x 10-7d Monomer size 29 kDa ? Active site Se (selenocysteine) S (cysteine)? Effect of T, administration Increase Decrease Physiological role Extracellular T, production Intracellular T, production

K, values are for inhibtion of rTa deiodination for type I and Tq deodination for type II assays, at the DTT concentrations indicated below. ’ Liver microsomes assayed at 3-5 mM DTT (2). * Brain microsomes assayed at 20 mM DTT (16). ’ Kidney microsomes assayed at 5 mM DTT, BAT, and brain microsomes assayed at 0.5, 1, and 2 mM DTT (20). d Liver microsomes assayed at 1 mM DTT, BAT microsomes assayed at 40 mM DTT (65). the rest of the body is hypothyroid. Under these circum- Subsequent affinity labeling studies with the substrate stances, however, the type II deiodinase may also con- analog, bromoacetyl-T4, showed that the amounts of tribute significantly to the pool of circulating Ts, as type deiodinase activity in liver and kidney microsomes after II enzyme levels are elevated and type I levels depressed various treatments (e.g. hypo- and hyperthyroidism) cor- (41). In U~‘UO,hyperthyroidism decreases levels of type II related with the amount of labeling of a 27-kDa protein deiodinase activity. In vitro administration of Tq to cul- (47). Kohrle et al. (48, 49) further established that the tured astrocytes produces rapid changes in the actin potency of competitive inhibitors of type I deiodination cytoskeleton and a parallel decrease in type II activity, paralleled their ability to inhibit labeling of the 27-kDa presumably mediated by alterations in the subcellular protein. All of these results pointed to the 27-kDa protein distribution of the type II deiodinase protein (42, 43). A as a component of the type I enzyme. The native molec- third deiodinating enzyme catalyzes inactivation of T.? ular mass of the type I deiodinase was subsequently by removal of a tyrosyl ring iodide, a process termed 5 determined by sedimentation analysis to be approxi- deiodination (15). This enzyme, designated type III, is mately 55 kDa (50). Despite the progress in purification present in brain and placenta. The response of type III of this enzyme and the ability to label it covalently, deiodinase to thyroid status parallels type I, resulting in sufficient sequence information to identify a complemen- increased T3 degradation when hormone is abundant and tary DNA (cDNA) for this protein was not forthcoming. slower degradation in the hypothyroid state, conserving Another approach to this problem, the use of antibodies active hormone in the critical tissues when it is limiting. against deiodinase preparations to screen cDNA librar- In addition to the tissue differences in iodothyronine ies, resulted in the identification of a cDNA-encoding metabolism, thyroid hormone transport also varies protein disulfide isomerase (PDI), presumably due to among tissues. In particular, exchange of Ts between contaminating anti-PDI antibodies (51). PDI was sub- serum and brain is significantly slower than exchange sequently shown to be unrelated to type I deiodinase with other tissues, thus intracellular Tq to TB conversion (52, 53). is critical for thyroid hormone action in brain (44). IV. Cloning of the cDNA for the Type I III. Purification and Characterization of Type I Deiodinase Deiodinase A quite different strategy, expression cloning, has During the past decade, a number of laboratories have proven successful in isolating clones for a number of devoted considerable efforts toward the purification and proteins that possess potent bioactivity but which, due biochemical characterization of the type I deiodinase. to properties such as low abundance, instability, or mem- Mol et al. (45, 46) succeeded in purifying rat liver type brane localization, are difficult to purify. When these I deiodinase approximately 2400-fold by affinity chro- problems were encountered with the type I deiodinase, matography on PTU-Sepharose. This resulted in a prep- the specificity, sensitivity, and simplicity of the deiodi- aration of approximately 50% purity exhibiting a subunit nase assay suggested the use of this strategy. In this molecular mass of approximately 25 kilodaltons (kDa). approach, a cDNA encoding a particular protein is iden-

Downloaded from edrv.endojournals.org on June 22, 2006 210 BERRY AND LARSEN Vol. 13, No. 2 tified in a pool, or library, on the basis of the activity it symmetrical ends, resulting in insertion in random ori- expresses. The pool is subdivided and the assay for entation. Transcription from either side of the cloning activity repeated until a single clone is obtained. The site will thus result in approximately 50% of the products assays can be performed after expression by transient being antisense, which would in turn decrease the effi- transfection of cells in culture, or more commonly, by ciency of expression by half. In practice, the impact may injection of RNA transcripts of the cDNA into Xenopus be greater, due to hybridization of antisense with sense oocytes. The oocyte system has the advantage of being transcripts, resulting in hybrid arrest. For these reasons, highly efficient at synthesizing proteins from microin- we used an orientation-specific strategy to construct our jetted RNAs but the disadvantage of being quite tedious cDNA libraries (Fig. 1). The choice of cloning vector is compared to transfection, as each cell must be injected also crucial. Vectors derived from bacteriophage-X allow individually. The first step in this approach is to dem- highly efficient cDNA insertion and replication. How- onstrate expression of activity from injected messenger ever, a cDNA of this size would be only 3-5% the size of RNA (mRNA). We and others showed that type I deiod- the vector, and large amounts of phage DNA would be inase activity was easily detectable in oocytes after injec- required to produce small quantities of RNA transcripts. tion of rat liver and kidney mRNA (53, 54). Lambda Zap11 has been engineered to allow simple in Before construction of a cDNA library, we wished to uiuo conversion to a small plasmid vector, Bluescript, identify the best tissue source of RNA for this enzyme. whereby the cDNA insert then represents 30-50% of the It had been shown some time ago that levels of renal and total DNA, and transcription from the plasmid is quite hepatic type I deiodinase activity were elevated in hy- efficient. We prepared several XZup libraries using cDNA perthyroid and depressed in hypothyroid animals. Using size-fractionated to the range of approximately 2 kb. Size oocyte expression of deiodinase activity to quantitate fractionation reduced the total amount of cDNA by about levels of deiodinase mRNA, we established that these lo-fold, thus increasing the screening efficiency by this changes in activity reflected changes in the amount of amount. Libraries were subdivided into pools, converted type I deiodinase mRNA (53). The high levels of deiodinase mRNA in hyperthyroid liver led us to choose this mRNA AAAA source for cDNA synthesis. Another requirement for successful expression cloning is that a functional protein reverse transcriptase must be encoded by a single mRNA, as the mRNAs for AAAA multimeric proteins would be separated in the subdivi- sion process. And, because expression screening relies on TTTT Xba I primer expression of a functional protein, a full-length cDNA DNA pal I must be present to detect activity. One way to circumvent DNA ligase these requirements is to use hybrid arrest of translation. T4 DNA pal AAAA This approach measures the ability of a cDNA to inhibit Xbal expression from a positive mRNA pool, by virtue of the TTTT El cDNA hybridizing to the functional mRNA and blocking EcoRl its ability to be translated. The cDNA does not have to adaptors be full length to inhibit translation, and inhibition of expression of one subunit in a multisubunit protein will block activity. Hybrid arrest screening was, in fact, used Xba I by St. Germain et al. (55) to identify a partial cDNA for I the type I deiodinase. We elected to use the direct expression approach, CyGq + because it ensures that if a positive clone is obtained, it will be full length and functional. We size-fractionated FIG. 1. Type I iodothyronine 5’-deiodinase cloning strategy. Isolate hyperthyroid rat liver mRNA before injection, which RNA from hyperthyroid rat liver. Oligo dT select poly(A)+ RNA. Synthesize orientation-specific cDNA. Size-fractionate on agarose to established a fairly discrete size for the active mRNA obtain 1.9-2.5 kb cDNA. Ligate into EcoRI + XbaI-digested XZAP species, approximately 2 kilobases (kb) (53). Although DNA. In vitro package DNA into X-phage particles. Titer library and the subunit composition of the type I enzyme was un- subdivide into 10 pools. Convert to plasmid by coinfection with pha- known at that time, this result pointed to a single mRNA gemid. Prepare plasmid DNA. Transcribe RNA in uitro. Inject into species, providing us the encouragement needed to at- Xenopus oocytes. Incubate oocytes 2-3 days, then homogenize. Assay for [““I]rT, deiodination. Subdivide active pool, prepare DNA, tran- tempt direct screening. Construction of a cDNA library scribe, inject, and assay. Repeat screening until a small pool is obtained. was the next step. Traditionally, cDNAs are inserted Plate to obtain individual colonies and repeat screening until a single into cloning vectors after ligation of linkers that produce positive clone is identified.

Downloaded from edrv.endojournals.org on June 22, 2006 May, 1992 SELENIUM AND THYROID HORMONE METABOLISM 211 to plasmid, and tested for expression of deiodinase in cDNA (2106 nt) and the hybridizing mRNA (2.1 kb) oocytes. Finally, a signal of activity was detected, only made it likely that little if any 5’-untranslated sequence slightly above background, but reproducible. When this was missing. If this open reading frame represented the positive pool was further subdivided and screened, a entire coding sequence, then deletion of sequences be- higher signal was obtained, and repeating this process yond the TGA codon should have no effect on expression eventually resulted in the identification of a single posi- in oocytes. In fact, this deletion was inactive, establishing tive clone containing a 2.1-kb insert (56). The kinetic that sequences downstream of the UGA are necessary properties, substrate specificity, and PTU sensitivity of for activity (Fig. 3a). To determine the function of these the protein encoded by this cDNA confirmed its identity sequences, we inserted two nucleotides at a position just as type I deiodinase. Using a labeled RNA probe gener- beyond the UGA codon, generating a frame shift of the ated from the clone, we examined the tissue distribution sequences downstream of the insertion. The frame shift of deiodinase RNA in rat. Hybridizing RNA of about 2.1 mutant was also inactive, indicating that sequences kb was present in liver, kidney, thyroid, and pituitary, downstream of the UGA must be in the same reading tissues which express type I deiodinase. No hybridization frame as the upstream sequence, and therefore are likely was observed in RNA from brown adipose, lung, spleen, to be coding. Deletion of nt 745-1363 produced a protein or small intestine, tissues in which the enzyme is not with full activity, confirming that sequences downstream detectable. Northern analysis of RNA from hypothyroid, of the UGA are essential. euthyroid, and hyperthyroid rats showed that levels of An explanation for these puzzling results was sug- hybridizing RNA paralleled thyroid status, confirming gested from studies on mammalian glutathione peroxi- the expression studies in oocytes (Fig. 2). dase (57) and the bacterial enzymes formate dehydrogenase (58) and glycine reductase (59). In each of these V. Type I Deiodinase Contains Selenocysteine, enzymes, UGA encodes the amino acid selenocysteine, Encoded by a TGA Codon an analog of cysteine in which sulfur is replaced by selenium. If the UGA codon in the type I deiodinase were The sequence of the type I deiodinase cDNA showed also translated as selenocysteine, the protein would ter- an initiator methionine (ATG) codon at nucleotides (nt) minate at a UAG stop codon at nt 778-780, and its 7-9 and an open reading frame of 125 codons, terminat- predicted molecular mass would be approximately 29 ing with a TGA (UGA) codon at nt 382-384. This open kDa, in close agreement with the size of the affinity- reading frame predicted a protein of 125 amino acids, or labeled protein. To confirm that this was the case, we approximately 14 kDa, a surprising result, since aflinity- used site-directed mutagenesis to change the UGA codon labeling studies indicated an approximately 27-kDa sub- to a UAA stop codon. This mutation resulted in loss of unit size. The close agreement between the size of the deiodinase expression in oocytes, confirming that UGA was not read as a stop signal (Fig. 3b). We next changed KIDNEY UGA to UUA, encoding leucine, and this mutant was -Eu+ also inactive. Thus, leucine cannot substitute for selenocysteine in producing a functional enzyme. Finally, and of greatest interest, mutation of UGA to UGU, a cysteine codon, produced a functional deiodinase, albeit with altered properties compared to the wild type protein. Analysis of the protein products of the wild type and -18s mutant cDNAs by in vitro translation in reticulocyte lysates confirmed the results of the expression studies in oocytes. In vitro synthesized wild type deiodinase RNA produced a small amount of full-length approximately 29-kDa protein, with the majority of the translated product being approximately 14 kDa, consistent with termination at the UGA codon. We interpreted this as being FIG. 2. Effect of thyroid status on Type I 5’-deiodinase mRNA levels. due to inefficient translation of UGA as selenocysteine Rats were made hypothyroid by treatment for three weeks with 0.02% in the reticulocyte lysate. Translation of the UAA stop methimazole in drinking water. Hyperthyroidism was induced by ip codon mutant resulted in loss of the 29-kDa protein injection of 50 pg Ts daily for 3 days. Liver and kidney poly(A)+ RNA without affecting production of the 14-kDa protein. Both (5 rg) from hypothyroid (-), euthyroid (Eu), and hyperthyroid (+) rats was probed with type I deiodinase cRNA. The migration position of the leucine and cysteine codon mutants produced a 29- 18s RNA is shown. [Reprinted with permission from Berry et al.: kDa protein with high efficiency. These results con- Nature 353:273, 1991 (84).] firmed the translation of UGA in type I deiodinase as

Downloaded from edrv.endojournals.org on June 22, 2006 212 BERRY AND LARSEN Vol. 13, No. 2

Relative Deiodinase a. Deletion Mutants Activity

Pst I AC.21 Hind III Hind Ill i2QraksJEG 5’DIcDNA ‘I’ I I I ’ 100% 100% ATG TGA TAG 56 Pst I I I I 1 0 0 427 Act I I 0 0 745 1363 Hind III I 1 1 I I 1 100% 100% AG

b. UGA Codon Mutants mRNA Jranslatioq TGA UGA opal or Secys 100%

TAA UAA ochre 0 TTA UUA leucine 0

TGT UGU cysteine 10%

FIG. 3. Expression of type I 5’-deiodinase from wild type and mutant constructs. a, Deletions were constructed by restriction digestion at the indicated sites, followed by agarose gel purification of the desired fragments, religation, and mapping of the resulting constructs. RNAs were transcribed in vitro and 0.1-20 ng injected per oocyte. DNA transfections and deiodinase assays are as described previously (37). Oocyte activity of 100% is defined as deiodination of 30-40% 2 nM [Y]rTe/h with a homogenate of four oocytes injected with 0.1 ng deiodinase RNA per oocyte. JEG-3 cell sonicates were incubated with 25 mM dithiothreitol and 5 nM [““I]rT, for 1 h and lz511 was quantitated as described (37). All assays were in duplicate. ND, Not done. b, The TGA codon at nt 382 was replaced by the indicated codons using the P-select in vitro mutagenesis system of Promega (Madison, WI). Oligonucleotides corresponding to the desired changes were annealed to single-stranded DNA, and double-stranded DNA was synthesized. The entire coding regions of plasmids thus obtained were sequenced to confirm that these were the only mutations. Injections and assays were as described above. [Reprinted with permission from Berry et al.: Nature 353:273, 1991 (84).] selenocysteine, establishing this enzyme as a new mem- ficiency also resulted in depletion of thyroidal iodine and ber of the UGA-selenocysteine family. thyroid hormones, presumably due to increased synthesis and secretion of thyroglobulin by the thyroid, stimulated VI. Nutritional Studies Suggest a Role for by the elevated TSH levels. Selenium in Thyroid Hormone Metabolism Studies with [75]Se by Behne et al. (65) showed that, after severe selenium depletion, the distribution of label While these cloning studies were in progress, a role for in different tissues indicated that the brain and endo- selenium in thyroid hormone metabolism was emerging crine glands have priority on supplies of this element. from nutritional studies. Arthur, Beckett, and their col- This idea was supported by results of Arthur et al. (64), leagues (60) found that rats maintained on a selenium- showing that repletion of rats with a single injection of deficient diet for 4-6 weeks had elevated levels of serum 10 pg Se/kg body wt restored pituitary GH, plasma TSH, Tq and depressed serum TB compared to animals main- and thyroidal iodide and hormone levels, but did not tained on a normal selenium-sufficient diet. These alter- affect hepatic deiodinase or serum Tq. A higher dose, 200 ations in hormone levels were subsequently shown to be pg Se, restored serum hormone levels and hepatic deiod- accompanied by a marked decrease in hepatic deiodinase inase activity to normal. Beckett et al. (62) also observed activity. Time course studies showed these changes to be a decrease in brain type II deiodinase activity in sele- progressive from O-5 weeks of selenium deficiency (61), nium-deficient animals, and proposed that this protein leading these investigators to propose that selenium was might be a selenoenzyme. However, as discussed above, either a component of the type I deiodinase itself or of a changes in subcellular distribution of type II deiodinase necessary cofactor (62, 63). Plasma TSH levels were as a result of the elevated T4 concentrations associated elevated nearly 2-fold, and pituitary GH levels decreased with selenium deficiency might also explain the decrease by one-third in selenium deficiency, despite high levels in enzyme activity. Subsequent studies by our group and of serum Tq (64). These effects can be explained by the by others argue that the type II deiodinase is not a combinatorial effects of lower pituitary type II deiodinase selenoprotein (see below) (66, 67). activity due to high Tq, which would counterbalance the The initial [75]Se labeling studies described above iden- elevated levels of Tq, and the lower serum T3 leading to tified an approximately 27-kDa selenoprotein in rat liver, decreased saturation of nuclear receptors. Selenium de- kidney, and thyroid (68). Behne et al. (69) and Arthur et

Downloaded from edrv.endojournals.org on June 22, 2006 May, 1992 SELENIUM AND THYROID HORMONE METABOLISM 213 al. (70) subsequently demonstrated that [7”]Se and the gold inhibition of the selenoenzyme, glutathione peroxi- substrate analog, [1251]BrAcT3, both label a 27-kDa pro- dase, to three thiol active-site enzymes showed that the tein in rat liver, providing compelling evidence that selenoenzyme was much more sensitive to competitive selenium was a component of the type I deiodinase. inhibition by gold (72). This inhibition is thought to be due to direct interaction of gold (Au’) with selenium VII. Selenocysteine Confers Specific Properties (Se-). We found the cysteine mutant deiodinase to be on the Deiodinase Enzyme approximately loo-fold less sensitive to competitive inhibition by gold thioglucose (GTG) than the wild type Although the oocytes allowed us to assay wild type and selenoenzyme. BrAcTs affinity labeling of the wild type mutant type I deiodinase proteins, this system was not and cysteine mutant enzymes with ‘2”I-labeled substrate amenable to production of large amounts of enzyme on analogs in the presence of these inhibitors correlates a routine basis. Further characterization of these en- closely with deiodinase activity (Fig. 4). Selenocysteine zymes was facilitated by first introducing the cDNAs and cysteine are structurally similar, differing solely by into a plasmid, CDM-8, which can be expressed in mam- the substitution of a selenium atom for a sulfur atom. malian cells, for example COS-7 monkey kidney cells. Selenium is directly below sulfur in the periodic table, The plasmid is then introduced into cells by transient thus the two atoms have the same valence. However, the transfection, achieved by adding precipitated DNA to selenol of selenocysteine has a pK of 5.2, whereas the the cell monolayer and chemically permeabilizing the thiol of cysteine has a pK of 8.3, thus cysteine is mostly membranes. Transient expression of transfected DNAs protonated at physiological pH, whereas selenocysteine is efficient, reproducible, and technically easy. Using this is ionized and therefore more reactive. As the selenium system, we were able to perform detailed biochemical to sulfur change is the only difference between the wild analyses of the wild type and cysteine mutant deiodinase type deiodinase and the cysteine mutant, the higher proteins (71). As mentioned above, the mutant in which reactivity of selenocysteine must account for the greater selenocysteine was replaced by cysteine was active, but sensitivity of the wild type enzyme to PTU and gold. We its properties, summarized in Table 2, were quite differ- conclude from these studies that the presence of seleno- ent from those of the wild type enzyme. The wild type cysteine determines many of the specific biochemical deiodinase exhibits an approximately loo-fold preference properties of the type I deiodinase. for rT.? over T4 as substrate. Substitution of cysteine Based on these results, we propose the scheme for type resulted in a lo-fold decrease in the inhibition constant I iodothyronine deiodination, competitive inhibition by (Ki) for T4 and a lo-fold increase in the Michaelis- gold, and uncompetitive inhibition by PTU shown in Fig. Menten constant (K,) for rT3, thus eliminating the 5. According to this scheme, the selenolate anion of preference for rTg characteristic of the wild type enzyme. selenocysteine is the iodide acceptor in the active center This increase in K, was interpreted in our initial studies of the enzyme. The enzyme reacts with Tq, removes the as a decrease in activity in the mutant, due to the 5’-iodide, and releases TB. The selenolyl iodide form of conditions of the assay, which employed tracer concen- the enzyme can then react with thiol cofactor, release trations (-2 nM) of rTg (Fig. 3). When assayed at higher free iodide, and regenerate free enzyme. PTU competes substrate concentrations, this difference in activity is with thiol for the selenolyl iodide form of the enzyme less apparent. The cysteine mutant exhibits an approxi- and forms the inactive mixed selenolyl-thiol complex. mately 300-fold decrease in sensitivity to both competi- The competitive inhibitor, gold, reacts with the seleno- tive inhibition by PTU us. thiol and uncompetitive in- late anion in the free enzyme to form the inactive selen- hibition us. iodothyronine. Previous studies comparing olyl-gold complex. TABLE 2. Properties of wild type and cysteine mutant type I deiodinases VIII. Evidence That Type II Deiodinase Is Not a Selenoprotein Wild type Cysteine mutant (Set) (CYS) The properties of the cysteine mutant bear some sim- K, for rT:, (FM) 0.25 2.7 ilarity to those of the type II deiodinase found in brain, K, for T, bid 11 1 pituitary, and brown adipose, suggesting that this en- K, for PTU[rTJ (PM) 0.2 55 zyme may contain cysteine instead of selenocysteine in K, for PTU[DTT] (FM) 0.12 14 the active site, and the reduced sensitivity to gold exhib- K, for GTG (nM) 6.6 600 Active site Se (selenocysteine) S (cysteine) ited by the cysteine mutant provided a way of addressing this question. Rats were administered 5 mg/lOO g body All reactions contained 10 UIM DTT. All values are for inhibition of 5’-monodeiodination of T, under conditions in which enzyme is rate wt GTG or thioglucose as a control, and deiodinase limiting (55). activities from various tissues were assayed (67). A single

Downloaded from edrv.endojournals.org on June 22, 2006 214 BERRY AND LARSEN Vol. 13, No. 2

FIG. 4. Affinity labeling of wild type and cysteine mutant deiodinase proteins produced by transient transfection of COS- - 68 kDa 7 cells. COS-7 cells were transfected with either wild type (selenocysteine) 5’- deiodinase cDNA or cysteine-mutant (cysteine) deiodinase cDNA in the vector, CDM-8, or with vector alone (CDM), - 30 as indicated at the bottom of the lanes. Labeling with [“‘I]BrAcT3 was for 10 min with the indicated additions. - IS

- 14

CDM SELENOCYSTEINE CYSTEINE injection of GTG inhibited liver type I deiodinase by steine. Type II deiodinase exhibits an approximately approximately 50% but had no effect on type II activity lOOO-fold lower K, for T4 and rT3 and an approximately in brown adipose tissue (BAT). Injection of the same loo-fold higher Ki for PTU than does the cysteine mu- dose for 2 days completely inhibited types I and II tant of the type I enzyme. Sedimentation studies by deiodinase, indicating that both enzymes were inhibita- Safran and Leonard (50) indicated a native molecular ble, but their sensitivities differed. Because in vitro stud- mass for the type II enzyme of approximately 199 kDa, ies are less susceptible to variations in tissue uptake or us. approximately 55 kDa for the type I enzyme. Further, metabolism of the inhibitor, we examined the effects of Northern analysis under low stringency conditions failed GTG on deiodinase activities in tissue sonicates or mi- to identify a type I cross-reactive mRNA in cold-stimu- crosomal protein. Type I deiodinase from liver was ap- lated BAT, indicating that the degree of sequence ho- proximately loo-fold more sensitive to GTG than was mology between the type I and II sequences was low (56). type II enzyme from BAT. Manipulation of reaction This hypothesis was further strengthened by results of conditions allowed us to assay both deiodinase activities Safran et al. (66), which showed that the type II deiodi- in the same tissue, pituitary. Type I and type II deiodinase in glial cells did not incorporate [75]Se under con- nases from pituitary display an approximately loo-fold ditions permitting labeling of the type I deiodinase in difference in sensitivity to GTG, giving strong support porcine kidney cells. Selenium depletion of the culture to the hypothesis that the type II enzyme does not media did not affect type II activity, whereas both type possess the unique selenocysteine active center. This is I deiodinase and glutathione peroxidase activity were not to suggest that the only difference between the type inhibited. Furthermore, in more recent studies it was I and II enzymes is the presence or absence of selenocy- shown that in hypothyroid selenium-deficient rats, brain type II deiodinase activity was not different from selenium-supplemented controls (73). This further supports “mT3 H 0 the above indirect evidence that the type II deiodinase does not contain selenocysteine in the active site.

IX. Why is Selenocysteine Present in Proteins? a---+ L PTU ;H3 DTTox DTTr,, +I@ The selenoenzymes glutathione peroxidase, formate FIG. 5. Proposed mechanism for type I iodothyronine deiodination and dehydrogenase, glycine reductase, and 5’-deiodinase are inhibition by PTU and gold. [Reprinted with permission from Berry et all oxidoreductases which undergo ping-pong reaction al.:J Biol Chem266:14155, 1991 (71).] mechanisms. It is possible that selenocysteine has been

Downloaded from edrv.endojournals.org on June 22, 2006 May, 1992 SELENIUM AND THYROID HORMONE METABOLISM 215 maintained in these proteins through evolution because the case for the type I deiodinase (84). Deletion of the of its high efficiency at catalyzing this type of reaction. sequences beyond nt 907 of the wild type deiodinase It has been speculated that early in evolution, TGA was cDNA resulted in complete loss of activity (Fig. 6a). This more frequently used as a selenocysteine codon, and that deletion begins approximately 125 nt beyond the end of this usage may have been selected against by the emer- the coding region, in the 3’-untranslated (3’ut) portion gence of oxygen in the atmosphere, such that it was only of the mRNA. Deletion of the same sequences in the maintained in anaerobic environments or in proteins cysteine mutant had no effect; this construct produced a where it is protected from oxidation (74, 75). Evolution fully active product. Sequences in the 3’ut region must of UGA from a sensecodon to a stop codon is supported therefore be present for selenocysteine, but not for cys- by the fact that UGA is more commonly used than UAG teine, incorporation. The deletions shown in Fig. 6a and UAA stop codons (74). The fact that UGA encodes identified the region between nt 1363-1615 as being selenocysteine in organisms ranging from Escherichia essential for selenocysteine insertion. These sequences coli to man also supports early usage of UGA as a are located greater than 1 kb downstream of the UGA selenocysteine codon. codon in the wild type mRNA but are still functional if A cysteine variant (UGC codon) of formate dehydro- the spacing is altered. Inversion of these sequencesabol- genaseoccurs in nature in methanobacteria. This protein ished their function, even though this mutation preserves is otherwise highly homologous to the selenocysteine- the secondary structure of the mRNA. The analogies to containing formate dehydrogenase in E. coli, indicating glutathione peroxidase already discussed led us to test a common origin (76). However, the cysteine form is less whether the 3’ut region of this mRNA could substitute active and is present in greater amounts, presumably to for the deiodinase 3’ut in supporting selenocysteine in- compensate for this decrease in activity (76). In vitro sertion. We constructed a hybrid between the coding mutagenesis of the E. coli formate dehydrogenase selen- region of the deiodinase clone and the 3’ut sequencesof ocysteine to cysteine produced an enzyme with a 300- the rat glutathione peroxidase clone (Fig. 6b). Since it is fold decrease in turnover (k,,,) (77). Our preliminary a seleno-enzyme, glutathione peroxidase should also pos- results indicate that this same substitution in the type I sess selenocysteine insertion sequences, and, in fact, it deiodinase results in a 100 to 300-fold decrease in k,,, does. The 3’ut from glutathione peroxidase could substi- (our unpublished results). Synthetic selenoproteins have tute for the deleted sequencesin the deiodinase mRNA, also contributed to our understanding of the unique restoring its capacity t’o direct selenocysteine insertion. properties of selenocysteine. Using automated peptide Deletion analyses of the hybrid construct identified a synthesis, an analog of metallothionein was produced in functional region of approximately 230 nt in the gluta- which the seven cysteines were replaced by selenocy- thione peroxidase 3’ut. Similarly, hybrids between the steines, resulting in altered copper binding (78). Chemi- coding region of the rat deiodinase and the 3’ut from a cal conversion of the active site cysteine in subtilisin to human deiodinase clone are functional (Fig. 6b). The selenocysteine changed the activity of this enzyme from essential regions in the deiodinase and glutathione per- that of a protease to an acyl transferase (79), which also oxidase 3’uts bear very little primary sequence similarity, can function as a glutathione peroxidase (80). suggesting that secondary structures may be involved in regulating selenocysteine insertion. X. New Insights into Protein Synthesis in Computer analyses of the 3’ut sequencesof the deiod- Eukaryotes inases and rat glutathione peroxidase predict the for- mation of stable RNA stem-loop structures. Support for Identification of selenocysteine in the active site of the the existence of such structures was obtained by analyz- type I deiodinase introduces an interesting physiological ing small, carefully chosen deletions in the rat deiodinase question: How does the cell distinguish a UGA codon for and glutathione peroxidase 3’ut sequences. Removal of selenocysteine from a UGA stop codon? This process has the the loop and the last 20% of the stem in the rat been shown to occur cotranslationally, as transfer RNAs deiodinase clone resulted in complete loss of deiodinase which recognize UGA and insert selenocysteine have activity (Fig. 6b). Even deletion of the loop alone in been identified in bacteria and animals (81), and an either the rat deiodinase (9 nt) or glutathione peroxidase elongation factor specific for this transfer RNA has been clone (8 nt) abolished activity. Confirmation that these identified in E. coli (82). However, specificity must be mutations abolished selenocysteine incorporation was conferred on the recognition process. Studies of bacterial obtained by in vitro transcription and translation of the formate dehydrogenase implicated a stem-loop or hairpin wild type and mutant constructs. The deiodinase and structure in the mRNA in the region of the UGA codon hybrid deiodinase-glutathione peroxidase constructs as being necessary and sufficient for translation of UGA which produced deiodinase activity also program trans- as selenocysteine (83). We have shown that this is not lation of an approximately 29-kDa protein, whereas the

Downloaded from edrv.endojournals.org on June 22, 2006 216 BERRY AND LARSEN Vol. 13, No. 2

a Delodlnase Actlvlty

4TG TBA TAG 2106 5’DI-Wild type 100% 7 382 778

907 5’Dl-907 0%

TGT 2106 5’DI-Cys 100% FIG. 6. Expression of 5’-deiodinase ac- 907 tivity from wild type and mutant 5’- 5’DI-Cys-907 100% deiodinase constructs. a, Deletion and inversion mutations of rat 5’-deiodinase cDNA 3’ut region. Constructs were as- 1580 sayed for production of 5’-deiodinase ac- 5’DI- 1580 0% tivity after transient transfection in 1270 I495 COS-7 cells. Reactions were performed 5’Dl Al278-1495 as described for JEG cell sonicates in the 22% legend for Fig. 3, except that rT3 was 907 1440 present at 300 nM. Deiodinase activity 5’DI A907- 1440 ,lOO% at the level of the wild type rat 5’-deiodinase construct is defined as 100% and 5DI A745-1363 745 1363 1615 Al615-2106 ,lOO% was equivalent to 5’-deiodination of 2 pmol rTs/min. mg protein for TGA-con- 1245 1615 taining constructs and 1 pmol rTo/min. 5’Dl I”“. I245- I61 5 0% mg protein for TGT-containing constructs. Reprinted from Berry et al. (84). b, Rat 5’-deiodinase constructs contain- b ing 3’ut sequences from rat or human Delodlnase Actlvlty Delodlnase Actlvlty 5’-deiodinase or rat glutathione peroxi- Rat WT 100% dase cDNAs. Constructs containing Rat Ml 0% 1539 1573 either rat or human 5 ‘-deiodinase or rat 906/1402 J glutathione peroxidase 3’ut sequences 90611402 1660 t adjacent to rat 5’-deiodinase coding sequences were assayed for production of 1560 0% 5’-deiodinase activity as above. Adapted 100% from Berry et aZ.:Nature 353:273, 1991 L H"ma" WT 906;\572 ; (84. I093

GPXMI go6:922’“~~ 1 , ‘04’, ,55 0%

GPX WT 42%

906/922 I

inactive loop and stem-loop mutants produce the 14-kDa XI. Human Iodothyronine Deiodinase and protein expected from termination at the UGA codon. Clinical Implications These results show that sequences located greater than 1 kb downstream of the UGA codon in the deiodinase The sensitivity of iodothyronine deiodination in man mRNA are necessary for its translation as selenocy- to PTU has been a subject of debate. PTU administration steine, and that these sequences function in an orienta- inhibits Tq to Ts conversion in hyperthyroid humans by tion-specific but spacing-independent manner. We have, approximately 50% (37) but has little effect in euthyroid therefore, termed these essential 3’ut regions selenocy- patients (85-87). This may be due to the enhanced type steine-insertion sequence motifs. How these selenocy- I deiodinase activity in hyperthyroid liver, kidney, and steine-insertion sequence motifs function remains to be thyroid (28,33), together with the specific immunoglob- determined. ulin stimulation of type I deiodinase through the TSH

Downloaded from edrv.endojournals.org on June 22, 2006 May, 1992 SELENIUM AND THYROID HORMONE METABOLISM 217 receptor (34). The differences in sensitivity to PTU of the importance of supplementation with iodine before the wild type and cysteine mutant rat type I deiodinases administration of selenium in patients deficient in both raised the question of whether selenocysteine was also of these elements. present in the human type I enzyme. Using the rat type A previous analysis in one individual (92) in whom I deiodinase as a probe, we recently isolated a partial type I deiodinase levels appeared to be reduced showed cDNA for the human liver type I deiodinase. A portion a circulating hormonal profile quite similar to that de- of this clone was then used as a probe to obtain the scribed above for experimental selenium deficiency in remaining coding region from a human kidney cDNA rats. In this patient, serum total and free Tq values were library, and splicing the two together resulted in a func- elevated, whereas serum total and free T3 results were tional clone capable of expressing deiodinase activity. persistently in the lower half of the normal range. De- Sequencing of the human clone confirmed the presence spite the elevation in serum-free Tq, which would be of an in-frame UGA codon corresponding to the position expected to decrease TSH synthesis and/or release (l), of the UGA selenocysteine codon in the rat cDNA, and serum TSH was 2.2-2.4 mu/liter and TRH response affinity labeling studies indicate that the protein product normal or enhanced. In another family studied by Maxon of the human cDNA is approximately 28 kDa, confirming et al. (93), an elevation in total and free T4 was associated translation of the UGA (88). Overall, the coding regions in six affected members with high normal total TB levels of the human and rat clones are about 88% homologous and normal TSH. Although this pattern might also be at the amino acid level, and the properties of the two seen in mild generalized thyroid hormone resistance, the enzymes reflect this homology. The sensitivity of the elevation in serum T, (mean 210 nM) was out of propor- transiently expressed human enzyme to PTU is about 8- tion to that of serum Ts (2.7 nM), suggesting an impair- fold greater than that of the rat enzyme, but other kinetic ment in type I deiodinase activity. The fact that in both properties of the two enzymes are quite similar. These selenium-deficient rats and in the patients mentioned, studies indicate that selenocysteine is also present in the serum Tq is elevated whereas serum TSH is normal, human type I deiodinase, and this finding has important implies a role for type I deiodinase in the regulation of clinical implications for the treatment of hyperthyroid- TSH. With the present data, it is not possible to deter- ism. mine whether this reduced pituitary feedback sensitivity In central Africa, myxoedematous cretinism is preva- to T4 can be explained simply on the basis of slightly lent in areas of severe iodine deficiency. These cretins subnormal circulating TB concentrations or of a previ- are characterized by hypothyroidism with onset before ously unrecognized role of type I deiodinase as a source or shortly after birth, absence of goiter, and progressive of locally produced TB in the hypothalamus or pituitary thyroid damage and destruction (89,90). The severity of (1). cretinism was found to be proportional to the degree of The convergence of biochemical, molecular biological, hypothyroidism, but also to be correlated with selenium and nutritional studies on selenium and thyroid hormone deficiency (90, 91). The hypothesis was put forth that metabolism has yielded considerable advancement in our defective glutathione peroxidase due to selenium defi- understanding of iodothyronine deiodination and, in ad- ciency resulted in lack of protection against peroxidative dition, is providing new insight into selenium metabolism damage induced by the high levels of H,Oz in the thyroid and regulation of protein translation, with implications cell. Glutathione peroxidase activity was found to be ranging from the clinical to basic cellular processes and decreased in selenium-deficient areas in Zaire and evolutionary biology. Ubangi, and the enzyme activity in cretins was half the level in normal subjects (91). Selenium supplementation Acknowledgments for 2 months corrected glutathione peroxidase levels in We thank our colleagues in the Thyroid Division, Laila Banu, Susan both normal subjects and cretins (90,91). However, this Mandel, Yoyi Chen, David Kieffer, John Harney, and Robert Warne. supplementation also produced decreases in serum T4 and TB and an increase in TSH (91). Selenium supple- References mentation in the cretins, already suffering thyroid dys- 1. Larsen PR, Silva JE, Kaplan MM 1981 Relationships between function, produced a further decrease in serum Tq. For circulating and intracellular thyroid hormones: physiological and this reason, the selenium supplementation trial was dis- clinical implications Endocr Rev 2:87 continued, and patients were supplemented with iodine, 2. Leonard JL, Visser TJ 1986 Biochemistry of deiodination. In: Hennemann G (ed) Thyroid Hormone Metabolism. W. B. Saun- which resulted in reversal of thyroid parameters. These ders, Philadelphia, p 189 results suggested that, by decreasing the rate of thyroid 3. Gross J, Pitt-Rivers R 1952 The identification of 3,5,3’-L-triiodo- hormone metabolism, selenium deficiency could protect thyronine in human plasma. Lancet 1:439 4. Morreale de Escobar G. Escobar de1 Rev F 1967 Extrathvroidal against some of the consequences of iodine deficiency. effects of some antithyroid drugs and their metabolic consequences. In view of these findings, Contempre et al. (91) stressed Recent Prog Horm Res 2.3:87

Downloaded from edrv.endojournals.org on June 22, 2006 218 BERRY AND LARSEN Vol. 13. No. 2

5. Oppenheimer JH, Schwartz HL, Surks MI 1972 Propylthiouracil Subcellular localization of a rat liver enzyme converting thyroxine inhibits the conversion of L-thyroxine to L-triiodothyronine. An into tri-iodothyronine and possible involvement of essential thiol explanation of the antithyroxine effect of propylthiouracil and groups. Biochem J 157:479 evidence supporting the concept that triiodothyronine is the active 28. Kaplan MM 1979 Changes in the particulate subcellular compo- thyroid hormone. JClin Invest 51:2493 - nent of hepatic thyroxine-5’-monodeiodinase in hyperthyroid and 6. Frumess RD. Larsen PR 1975 Correlation of serum triiodothvro- hypothyroid rats. Endocrinology 105:548 nine (T3) and thyroxine (T,) with biologic effects of thyroid hor- 29. Laurberg P 1978 Selective inhibition of triiodothyronines from the mone replacement in propylthiouracil-treated rats. Metabolism perfused canine thyroid by propylthiouracil. Endocrinology 103:900 24:547 30. Ishii H, Inada M, Tanaka K, Mashio Y, Naito K, Nishikawa M, 7. Silva JE, Larsen PR 1977 Pituitary nuclear 3,5,3’-triiodothyronine Imura H 1981 Triiodothyronine generation from thyroxine in and thyrotropin secretion: an explanation for the effect of thyrox- human thyroid: enhanced conversion in Graves’ thyroid tissue. J ine. Science 198617 Clin Endocrinol Metab 52:1211 8. Silva JE, Dick TE, Larsen PR 1978 Contribution of local tissue 31. Erickson VJ. Cavalieri RR. Rosenberg LL 1982 Thvroxine-5’- thyroxine monodeiodination to the nuclear 3,5,3’-triiodothyronine deiodinase of rat thyroid, but not thacof liver, is dependent on in pituitary, liver, and kidney of euthyroid rats. Endocrinology thryotropin. Endocrinology 111:434 103:1196 32. Ishii H, Inada M, Tanaka K, Mashio Y, Naito K, Nishikawa M, 9. Silva JE, Larsen PR 1978 Contributions of plasma triiodothyronine Matsuzuka F, Kuma K, Imura H 1983 Induction of outer and inner and local thyroxine monodeiodination to triiodothyronine to nu- ring monodeiodinases in human thyroid gland by thyrotropin. J clear triiodothyronine receptor saturation in pituitary, liver, and Clin Endocrinol Metab 57:500 kidney of hypothyroid rats. J Clin Invest 61:1247 33. Toyoda N, Nishikawa M, Horimoto M, Yoshikawa N, Mori Y, 10. Larsen PR, Dick TE, Markovitz BP, Kaplan MM, Gard TG 1979 Yoshimura M, Masaki H, Tanaka K, Inada M 1990 Synergistic Inhibition of intrapituitary thyroxine to 3,5,3’-triiodothyronine effect of thyroid hormone and thyrotropin on iodothyronine 5’. conversion prevents the acute suppression of thyrotropin release deiodinase in FRTL-5 rat thyroid cells. Endocrinology 127:1199 in hypothyroid rats. J Clin Invest 64:117 34. Toyoda N, Nishikawa M, Horimoto M, Yoshikawa N, Mori Y, 11. Obregon MJ, Pascual A, Mall01 J, Morreale de Escobar G, Escobar Yoshimura M, Masaki H, Tanaka K, Inada M 1990 Graves’ im- Del Rey F 1980 Evidence against a major role of L-thyroxine at munoglobulin G stimulates iodotyronine 5’-deiodinating activity the pituitary level: studies in rats treated with iopanoic acid (Tel- in FRTL-5 rat thvroid cells. J Clin Endocrinol Metab 70:1506 epaque). Endocrinology 106:1827 35. Borges M, Ingbar SH, Silva JE 1990 Iodothyronine deiodinase 12. Cheron RG, Kaplan MM, Larsen PR 1979 Physiological and phar- activities in FRTL-5 cells: predominance of type I 5’-deiodinase. macological influences on thyroxine to 3,5,3’-triiodothyronine con- Endocrinology 126:3059 version and nuclear 3,5,3’-triiodothyronine binding in pituitary. J 36. Pazo-Moura CC, Moura EG, Dorris ML, Rehnmark S, Melendez Clin Invest 64:1402 L, Silva JE, Taurog A 1991 Effect of iodine deficiency and cold 13. Silva JE, Leonard JL, Crantz FR, Larsen PR 1982 Evidence for exposure on thyroxine 5’-deiodinase activity in various rat tissues. two tissue-specific pathways for in uiuo thyroxine 5’-deiodination Am J Physiol 260:E175 in the rat. J Clin Invest 69:1176 37. Abuid J, Larsen PR 1974 Triiodothyronine and thyroxine hyper- 14. Visser TJ, Kaplan MM, Leonard JL, Larsen PR 1983 Evidence for thyroidism. J Clin Invest 54:201 two pathways of iodothyronine 5’-deiodination in rat pituitary that 38. Silva JE, Larsen PR 1986 Hormonal regulation of iodothyronine differ in kinetics, propylthiouracil sensitivity, and response to 5’-deiodinase in rat brown adipose tissue. Am J Physiol 251:E639 hypothyroidism. J Clin Invest 71:992 39. Bianco AC, Silva JE 1987 Intracellular conversion of thyroxine to 15. Kaplan MM, Yaskoski KA 1980 Phenolic, tyrosyl ringdeiodination triiodothyronine is required for the optimal thermogenic function of iodothyronines in rat brain homogenates. J Clin Invest 66:551 of brown adipose tissue. J Clin Invest 79:296 16. Visser TJ, Leonard JL, Kaplan MM, Larsen PR 1982 Kinetic 40. Bianco AC, Silva JE 1987 Optimal response of key enzymes and evidence suggesting two mechanisms for iodothyronine 5’-deiodi- uncoupling protein to cold in BAT depends on local T3 generation. nation in rat cerebral cortex. Proc Nat1 Acad Sci USA 79:5080 Am J Physiol 253:E255 17. Silva JE, Larsen PR 1983 Adrenergic activation of triiodothyronine 41. Silva JE, Larsen PR 1985 Potential of brown adipose tissue type production in brown adipose tissue. Nature 305:712 II thyroxine 5’-deiodinase as a local and systemic source of tri- 18. Leonard JL, Mellen SA, Larsen PR 1982 Thyroxine 5’-deiodinase iodothyronine in rats. J Clin Invest 76:2296 activity in brown adipose tissue. Endocrinology 112:1153 42. Farwell AP, Leonard JL 1989 Identification of a 27-kDa protein 19. Kaplan MM, Shaw EA 1984 Type II iodothyronine 5’-deiodination with the properties of type II iodothyronine 5’-deiodinase in di- bv human and rat placenta in uitro. J Clin Endocrinol Metab butvrvl cvclic AMP-stimulated elial cells. J Biol Chem 264:20561 59:253 43. Far%1 AP, Lynch RM, Okulici WC, Comi AM, Leonard JL 1990 20. Silva JE, Mellen SA, Larsen PR 1987 Comparison of kidney and The actin cytoskeleton mediates the hormonally regulated trans- brown adipose tissue iodothyronine 5’-deiodinases. Endocrinology location of type II iodothyronine 5’-deiodinase in astrocytes. J Biol 121:650 Chem 265:18546 21. Leonard JL, Kaplan MM, Visser TJ, Silva JE, Larsen PR 1981 44. Silva JE, Matthews PS 1984 Production rates and turnover of Cerebral cortex responds rapidly to thyroid hormones. Science triiodothyronine in rat developing cerebral cortex and cerebellum: 214:571 responses to hypothyroidism. J Clin Invest 74:1035 22. Silva JE, Gordon MB, Crantz FR, Leonard JL, Larsen PR 1984 45. Mol JA, van den Berg TP, Visser TJ 1988 Partial purification of Qualitative and quantitative differences in the pathways of extra- the microsomal rat liver iodothyronine deiodinase I. Solubilization thyroidal triiodothyronine generation between euthyroid and hy- and ion-exchange chromatography. Mol Cell Endocrinol55:149 pothyroid rats. J Clin Invest 73:898 46. Mol JA, van den Berg TP, Visser TJ 1988 Partial purification of 23. Goswami A, Rosenberg IN 1983 Stimulation of iodothyronine outer the microsomal rat liver iodothyronine deiodinase II. Affinity chro- ring monodeiodinase by dihydrolipoamide. Endocrinology 112:1180 matography. Mol Cell Endocrinol55:159 24. Goswami A, Rosenberg IN 1985 Purification and characterization 47. Safran M, Kohrle J, Braverman LE, Leonard JL 1990 Effect of of a cytosolic protein enhancing GSH-dependent microsomal iodo- biological alterations of type I 5’.deiodinase activity on affinity- thyronine 5’-monodeiodination. J Biol Chem 26:6012 labeled membrane proteins in rat liver and kidney. Endocrinology 25. Goswami A, Rosenberg IN 1987 Thioredoxin stimulates enzymatic 126:826 outer ring monodeiodination of reverse triiodothyronine. Endocri- 48. Kohrle J, Rasmussen UB, Rokos H, Leonard JL, Hesch RD 1990 nology 121:1937 Selective affinity labelling of a 27-kDa integral membrane protein 26. Goswami A, Rosenberg IN 1988 Effects of glutathione on iodothy- in rat liver and kidney with N-bromoacetyl derivatives of L-thyronine 5’-deiodinase activity. Endocrinology 123:192 roxine and 3,5,3’-triiodo-L-thyronine. J Biol Chem 265:6146 27. Visser TJ, Van der Does-Tobe I, Dotter R, Hennemann G. 1976 49. Kohrle J, Rasmussen UB, Ekenbarger DM, Alex S, Rokos H, Hesch

Downloaded from edrv.endojournals.org on June 22, 2006 May, 1992 SELENIUM AND THYROID HORMONE METABOLISM 219

RD, Leonard JL 1990 Affinity labelling of rat liver and kidney type confers the biochemical properties of the type I iodothyronine I 5’-deiodinase. J Biol Chem 265:6155 deiodinase. J Biol Chem 266:14155 50. Safran M, Leonard JL 1991 Comparison of the physicochemical 72. Chaudiere J, Tappel AL 1984 Interaction of gold(I) with the active properties of type I and type II iodothyronine 5’-deiodinase. J Biol site of selenium glutathione peroxidase. J Inorg Biochem 20:313 Chem 2663233 73. Chanoine JP, Safran M. Farwell AP, Arthur JR, Beckett GJ, 51. Boado RJ. Camnbell DA. Chonra IJ 1988 Nucleotide seauence of Leonard JL, Effects of selenium deficiency on brain type II iodo: rat liver iodothyronine 5’-mdnodeiodinase (5’MD): its’ identity thvronine 5’-deiodinase in rats. Program of the 65th Annual Meet- with the protein disulfide isomerase. Biochem Biophys Res Com- ing of the American Thyroid AssocTation, Boston, 1991 (Abstract mun 155:1297 45) 52. Schoenmakers CHH, Pigmans IGAJ, Hawkins HC, Freedman RB, 74. Mizutani T, Kurata H, Yamada K 1991 Study of mammalian Visser TJ 1989 Rat liver type I iodothyronine deiodinase is not selenocvstevl-tRNA svnthesis with l’%elHSe-.. . Fed Eur Biochem identical to protein disulfide isomerase. Biochem Biophys Res Sot 28&59 - - Commun 165857 75. Bock A, Forchhammer K, Heider J, Leinfelder W, Sawers G, 53. Berrv MJ. Kates AL. Larsen PR 1990 Thvroid hormone regulates Veprek B, Zinoni F 1991 Selenocysteine: the 21st amino acid. Mol type-1 deiodinase messenger RNA in rai liver. Mol Endocrinol Microbial 5:515 76. Stadtman TC 1990 Selenium Biochemistry. Annu Rev Biochem 4:743 59111 54. St. Germain DL, Morganelli CM 1989 Expression of type I iodo- 77. Axley MJ, Bock A, Stadtman TC 1991 Catalytic properties of an thyronine 5’-deiodinase in Xenopus laevis oocytes. J Biol Chem Escherichia coli formate dehydrogenase mutant in which sulfur 264:3054 replaces selenium. Proc Nat1 AcadSci USA 88:8450 55. St. Germain DL, Dittrich W, Morganelli CM, Cryns V 1990 Mo- 78. Oikawa T, Nobuyoshi E, Tanaka H, Soda K 1991 Metalloseleno- lecular cloning by hybrid arrest of translation in Xenopus hevis nein. the selenium analosue of metallothionein: svnthesis and oocytes. J Biol Chem 265:20087 characterization of its complex with copper ions. P&c Nat1 Acad 56. Bern, MJ. Banu L. Larsen PR 1991 Tvpe I iodothvronine deiodi- Sci USA 88:3057 nase-is a selenocysteine-containing enzyme. Nature 349:438 79. Wu ZP, Hilvert D 1989 Conversion of a protease into an acyl

57. Chambers I.I Framnton. J.I Goldfarb P. Affara N. McBain W. Har- transferase: selenosubtilisin. J Am Chem Sot 111:4513 rison PR 1986 The structure of the mouse glutathione peroxidase 80. Wu ZP, Hilvert D 1990 Selenosubtilisin as a glutathione peroxidase gene: the selenocysteine in the active site is encoded by the “ter- mimic. J Am Chem Sot 112:5647 mination” codon, TGA. EMBO J 5:1221 81. Lee BJ, Worland PJ, Davis JN, Stadtman TC, Hatfield DL 1989 58. Zinoni F, Birkmann A, Stadtman T, Bock A 1986 Nucleotide Identification of a selenocysteyl-tRNAS” in mammalian cells that sequence and expression of the selenocysteine-containing polypep- recognizes the nonsense codon, UGA. J Biol Chem 264:9724 tide of formate dehydrogenase (formate-hydrogenase-lyase-linked) 82. Forchhammer K, Leinfelder W, Bock A 1989 Identification of a from Escherichia coli. Proc Nat1 Acad Sci USA 83:4650 novel translation factor necessary for the incorporation of seleno- 59. Garcia GE, Stadtman TC 1991 Selenoprotein A component of the cysteine into protein. Nature 342:453 glycine reductase complex from Clostridium purinolyticum: nu- 83. Zinoni F, Heider J, Bock A 1990 Features of the formate dehydrog- cleotide sequence of the gene shows that selenocysteine is encoded enase mRNA necessary for decoding of the UGA codon as seleno- by UGA. J Bacterial 173:2093 cysteine. Proc Nat1 Acad Sci USA 87:4660 Beckett GJ, Beddows SE, Morrice PC, Nicol F, Arthur JR 1987 84. Berry MJ, Banu L, Chen Y, Mandel SJ, Kieffer JD, Harney JW, 60. Larsen PR 1991 Recognition of UGA as a selenocysteine codon in Inhibition of hepatic deiodination of thyroxine is caused by sele- type I deiodinase requires sequences in the 3’ untranslated region. nium deficiency in rats Biochem J 248443 Nature 353:273 61. Arthur JR, Nicol F, Hutchinson AR, Beckett GJ 1990 The effects 85. Saberi M, Sterling FH, Utiger RD 1975 Reduction in extrathyroidal of selenium depletion and repletion on the metabolism of thyroid triiodothyronine production by propylthiouracil in man. J Clin hormones in the rat. J Inorg Biochem 39:lOl Invest 55:218 62. Beckett GJ, MacDougall DA, Nicol F, Arthur JR, 1989 Inhibition 86. Geffner DL, Azukizawa M, Hershman JM 1975 Propylthiouracil of type I and type II iodothyronine deiodinase activity in rat liver, blocks extrathyroidal conversion of thyroxine to triiodothyronine kidney and brain produced by selenium deficiency. Biochem J and augments thyrotropin secretion in man. J Clin Invest 55:224 2592387 87. LoPresti JS, Eigen A, Kaptein E, Anderson KP, Spencer CA, 63. Beckett GJ, Nicol F, Proudfoot D, Dyson K, Loucaides G, Arthur Nicoloff JT 1989 Alteration in 3,3’,5’-triiodothyronine metabolism JR 1990 The changes in hepatic enzyme expression caused by in response to propylthiouracil, dexamethasone, and thyroxine selenium deficiency and hypothyroidism in rats are produced by administration in man. J Clin Invest 84:1650 independent mechanisms. Biochem J 266:743 88. Mandel SJ, Berry MJ, Kieffer JD, Harney JW, Warne RL, Larsen 64. Arthur JR. Nicol F. Rae PWH. Beckett GJ 1990 Effects of selenium PR, Cloning and in vitro expression of human selenoprotein, Type deficiency’ on the’ thyroid gland and on plasma and pituitary I iodothyronine deiodinase. J Clin Endocrinol Metab, in press thyrotrophin and growth hormone concentrations in the rat. Clin 89. Goyens P, Golstein J, Nsombola B, Vis H, Dumont JE 1987 Chem Enzymol Commun 3:209 Selenium deficiency as a possible factor in the pathogenesis of 65. Behne D, Hilmert H, Scheid S, Gessner H, Elger W 1988 Evidence myxoedematous endemic cretinism. Acta Endocrinol (Copenh) for specific selenium target tissues and new biologically important 114:497 selenoproteins. Biochem Biophys Acta 966:1251 90. Vanderpas JB, Contempre B, Duale NL, Goossens W, Bebe N, Safran M, Farwell AP, Leonard JL 1991 Evidence that type II 5’- Thorpe R. Ntambue K. Dumont J, Thillv CH. Diplock AT 1990 66. Iodine and selenium deficiency associated-with cretinism in north- deiodinase is not a selenoprotein. J Biol Chem 26613477 ern Zaire. Am J Clin Nutr 52:1087 67. Berry MJ, Kieffer JD, Larsen PR 1991 Evidence that cysteine, not 91. Contempre B, Dumont JE, Bebe N, Thilly CH, Diplock AT, selenocysteine, is in the catalytic site of type II iodothyronine Vanderpas J 1991 Effect of selenium supplementation in hypothy- deiodinase. Endocrinology 129:550 roid subjects of an iodine and selenium deficient area: the possible 68. Behne D, Scheid S, Kyriakopoulos A, Hilmert H 1990 Subcellular danger of indiscriminate supplementation of iodine-deficient sub- distribution of selenoproteins in the liver of the rat. Biochem jects with selenium. J Clin Endocrinol Metab 73:213 Biophys Acta 1033:219 92. Kleinhaus N, Faber J, Kahana L, Schneer J, Scheinfeld M 1988 69. Behne D. Kvriakonoulos A. Meinhold H. Kohrle J 1990 Identifi- Euthyroid hyperthyroxinemia due to a generalized 5’-deiodinase cation of Type I ibdothyronine 5’-deiodinase as a selenoenzyme. defect. J Clin Endocrinol Metab 66:684 Biochem Biophys Res Commun 173:1143 93. Maxon HR, Burman KD, Premachandra BN, Chen IW, Burger A, 70. Arthur JR, Nicol F, Beckett GJ 1990 Hepatic iodothyronine 5’- Levy P, Georges LP 1982 Familial elevations of total and free deiodinase: the role of selenium. Biochem J 272:537 thyroxine in healthy, euthyroid subjects without detectable binding 71. Berry MJ, Kieffer JD, Harney JW, Larsen PR 1991 Selenocysteine protein abnormalities. Acta Endocrinol (Copenh) 100:224

Downloaded from edrv.endojournals.org on June 22, 2006