A designed antagonist of the thyroid hormone receptor.

We synthesized an analogue of the thyromimetic GC-1 bearing the same hydrophobic appendage as the estrogen receptor antagonist ICI-164,384. While having reduced affinity for the thyroid hormone receptors compared to GC-1, it behaves in a manner consistent with a competitive antagonist in a transactivation assay.

Selective modulation of thyroid hormone receptor action.

Thyroid hormones have some actions that might be useful therapeutically, but others that are deleterious. Potential therapeutically useful actions include those to induce weight loss and lower plasma cholesterol levels. Potential deleterious actions are those on the heart to induce tachycardia and arrhythmia, on bone to decrease mineral density, and on muscle to induce wasting. There have been successes in selectively modulating the actions of other classes of hormones through various means, including the use of pharmaceuticals that have enhanced affinities for certain receptor isoforms. Thus, there is reason to pursue selective modulation of thyroid hormone receptor (TR) function, and several agents have been shown to have some beta-selective, hepatic selective and/or cardiac sparring activities, although development of these was largely not based on detailed understanding of mechanisms for the specificity. The possibility of selectively targeting the TRbeta was suggested by the findings that there are alpha- and beta-TR forms and that the TRalpha-forms may preferentially regulate the heart rate, whereas many other actions of these hormones are mediated by the TRbeta. We determined X-ray crystal structures of the TRalpha and TRbeta ligand-binding domains (LBDs) complexed with the thyroid hormone analog 3,5,3'-triiodithyroacetic acid (Triac). The data suggested that a single amino acid difference in the ligand-binding cavities of the two receptors could affect hydrogen bonding in the receptor region, where the ligand's 1-position substituent fits and might be exploited to generate beta-selective ligands. The compound GC-1, with oxoacetate in the 1-position instead of acetate as in Triac, exhibited TRbeta-selective binding and actions in cultured cells. An X-ray crystal structure of the GC-1-TRbeta LBD complex suggests that the oxoacetate does participate in a network of hydrogen bonding in the TR LBD polar pocket. GC-1 displayed actions in tadpoles that were TRbeta-selective. When administered to mice, GC-1 was as effective in lowering plasma cholesterol levels as T(3), and was more effective than T(3) in lowering plasma triglyceride levels. At these doses, GC-1 did not increase the heart rate. GC-1 was also less active than T(3) in modulating activities of several other cardiac parameters, and especially a cardiac pacemaker channel such as HCN-2, which may participate in regulation of the heart rate. GC-1 showed intermediate activity in suppressing plasma thyroid stimulating hormone (TSH) levels. The tissue/plasma ratio for GC-1 in heart was also less than for the liver. These data suggest that compounds can be generated that are TR-selective and that compounds with this property and/or that exhibit selective uptake, might have clinical utility as selective TR modulators.

Hormone selectivity in thyroid hormone receptors.

Separate genes encode thyroid hormone receptor subtypes TRalpha (NR1A1) and TRbeta (NR1A2). Products from each of these contribute to hormone action, but the subtypes differ in tissue distribution and physiological response. Compounds that discriminate between these subtypes in vivo may be useful in treating important medical problems such as obesity and hypercholesterolemia. We previously determined the crystal structure of the rat (r) TRalpha ligand-binding domain (LBD). In the present study, we determined the crystal structure of the rTRalpha LBD in a complex with an additional ligand, Triac (3,5, 3'-triiodothyroacetic acid), and two crystal structures of the human (h) TRbeta receptor LBD in a complex with either Triac or a TRbeta-selective compound, GC-1 [3,5-dimethyl-4-(4'-hydroy-3'-isopropylbenzyl)-phenoxy acetic acid]. The rTRalpha and hTRbeta LBDs show close structural similarity. However, the hTRbeta structures extend into the DNA-binding domain and allow definition of a structural "hinge" region of only three amino acids. The two TR subtypes differ in the loop between helices 1 and 3, which could affect both ligand recognition and the effects of ligand in binding coactivators and corepressors. The two subtypes also differ in a single amino acid residue in the hormone-binding pocket, Asn (TRbeta) for Ser (TRalpha). Studies here with TRs in which the subtype-specific residue is exchanged suggest that most of the selectivity in binding derives from this amino acid difference. The flexibility of the polar region in the TRbeta receptor, combined with differential recognition of the chemical group at the 1-carbon position, seems to stabilize the complex with GC-1 and contribute to its beta-selectivity. These results suggest a strategy for development of subtype-specific compounds involving modifications of the ligand at the 1-position.

Definition of the surface in the thyroid hormone receptor ligand binding domain for association as homodimers and heterodimers with retinoid X receptor.

Thyroid hormone receptors (TRs) bind as homodimers or heterodimers with retinoid X receptors (RXRs) to DNA elements with diverse orientations of AGGTCA half-sites. We performed a comprehensive x-ray crystal structure-guided mutation analysis of the TR ligand binding domain (TR LBD) surface to map the functional interface for TR homodimers and heterodimers with RXR in the absence and/or in the presence of DNA. We also identified the molecular contacts in TR LBDs crystallized as dimers. The results show that crystal dimer contacts differ from those found in the functional studies. We found that identical TR LBD residues found in helices 10 and 11 are involved in TR homodimerization and heterodimerization with RXR. Moreover, the same TR LBD surface is operative for dimerization with direct repeats spaced by 4 base pairs (DR-4) and with the inverted palindrome spaced by 6 base pairs (F2), but not with TREpal (unspaced palindrome), where homodimers appear to be simply two monomers binding independently to DNA. We also demonstrate that interactions between the TR and RXR DNA binding domains stabilize TR-RXR heterodimers on DR-4. The dimer interface can be functional in the cell, because disruption of key residues impairs transcriptional activity of TRs mediated through association with RXR LBD linked to GAL4 DNA-binding domain.

Molecular and structural biology of thyroid hormone receptors.

1. Thyroid hormone receptors (TR) are expressed from two separate genes (alpha and beta) and belong to the nuclear receptor superfamily, which also contains receptors for steroids, vitamins and prostaglandins. 2. Unliganded TR are bound to DNA thyroid hormone response elements (TRE) predominantly as homodimers, or as heterodimers with retinoid X-receptors (RXR), and are associated with a complex of proteins containing corepressor proteins. Ligand binding promotes corepressor dissociation and binding of a coactivator. 3. Recent studies from our group have focused on the acquisition and use of X-ray crystallographic structures of ligand-binding domains (LBD) of both the rat (r) TR alpha and the human (h) TR beta bound to several different ligands. We have also developed ligands that bind selectively to the TR beta, which may provide ways to explore the differential functions of TR alpha compared with TR beta isoforms. 4. The LBD is comprised mostly of alpha-helices. The ligand is completely buried in the receptor and forms part of its hydrophobic core. Kinetic studies suggest that the limiting step in formation of high-affinity ligand-receptor complexes is the rate of folding of the receptor around the ligand. Ligands can be fitted tightly in the ligand-binding pocket and small differences in this fitting may explain many structure-activity relationships. Interestingly, analysis of the structures of antagonists suggests that they have chemical groups, 'extensions', that could impair receptor folding around them and, thus, prevent the agonist-induced conformation changes in the receptor. 5. The TR structures allowed us to see that the mutations that occur in the syndrome of generalized resistance to thyroid hormone are located in the vicinity of the ligand-binding pocket. 6. X-ray structure of the TR has also been used to guide construction of mutations in the TR surface that block binding of various proteins important for receptor function. Studies with these TR mutants reveal that the interfaces for homo- and heterodimerization map to similar residues in helix 10 and 11 and also allow the definition of the surface for binding of coactivators, which appears to be general for nuclear receptors. Formation of this surface, which involves packing of helix 12 of the TR into a scaffold formed by helices 3 and 5, appears to be the major change in the receptor structure induced by hormone occupancy.

Structure and specificity of nuclear receptor-coactivator interactions.

Combinatorial regulation of transcription implies flexible yet precise assembly of multiprotein regulatory complexes in response to signals. Biochemical and crystallographic analyses revealed that hormone binding leads to the formation of a hydrophobic groove within the ligand binding domain (LBD) of the thyroid hormone receptor that interacts with an LxxLL motif-containing alpha-helix from GRIP1, a coactivator. Residues immediately adjacent to the motif modulate the affinity of the interaction; the motif and the adjacent sequences are employed to different extents in binding to different receptors. Such interactions of amphipathic alpha-helices with hydrophobic grooves define protein interfaces in other regulatory complexes as well. We suggest that these common structural elements impart flexibility to combinatorial regulation, whereas side chains at the interface impart specificity.

Mechanisms of thyroid hormone action: insights from X-ray crystallographic and functional studies.

This review summarizes the studies conducted in our laboratory on the mechanisms of thyroid hormone action over the past two decades. We have attempted to place our studies on thyroid hormone receptors (TRs) in perspective with the work conducted by other investigators that established their nuclear localization, DNA-binding properties, DNA response elements, and the role of other proteins involved in TR-mediated regulation of gene transcription. Recently, our crystallographic studies of the TR ligand binding domain (LBD) revealed that the ligand has a structural role in the folding of the receptor's hydrophobic core. The analysis of the structure led to biochemical and genetic studies that have defined the surfaces on the TR LBD required for dimerization and binding of coactivator proteins. Placement of the mutations found in patients with the syndrome of generalized resistance to thyroid hormone on the TR LBD revealed that they were restricted to amino acids in the vicinity of the binding pocket for thyroid hormone. The insights gained from the elucidation of the TR LBD structure will provide the basis for the design of compounds with selective agonistic or antagonistic activities.

X-ray crystallographic and functional studies of thyroid hormone receptor.

We have solved several X-ray crystallographic structures of TR ligand-binding domains (LBDs), including the rat (r) TR alpha and the human (h) TR beta bound to diverse ligands. The TR-LBD folding, comprised mostly of alpha-helices, is likely to be general for the superfamily. The ligand, buried in the receptor, forms part of its hydrophobic core. Tight fitting of ligand into the receptor explains its high affinity for the TR, although the structure suggests that ligands with even higher affinities might be generated. The kinetics of 3,5,3'-triiodo-L-thyronine (T3) and 3,5,3',5'-tetraiodo-L-thyronine (T4) binding suggest that folding around the ligand, rather than receptor opening, is rate-limiting for high affinity binding. TR beta mutations in patients with resistance to T3 cluster around the ligand; these different locations could differentially affect on other receptor functions and explain the syndrome's clinical diversity. Guided by the structure, mutations have been placed on the TR surface to define interactions with other proteins. They suggest that a similar surface in the LBD is utilized for homo- or heterodimerization on direct repeats and inverted palindromes but not on palindromes. Coactivator proteins that mediate TR transcriptional activation bind to a small surface comprised of residues on four helices with a well-defined hydrophobic cleft, which may be a target for pharmaceuticals. The coactivator-binding surface appears to form upon ligand-binding by the folding of helix 12 into the scaffold formed by helices 3, 4 and 5. The analysis of most currently used antagonists suggest that although they probably fit into the ligand-binding pocket, they possess a group that may alter proper folding of the receptor, with disruption of the coactivator-binding surface (the 'extension model').

A high-affinity subtype-selective agonist ligand for the thyroid hormone receptor.

BACKGROUND: Thyroid hormones regulate many different physiological processes in different tissues in vertebrates. Most of the actions of thyroid hormones are mediated by the thyroid hormone receptor (TR), which is a member of the nuclear receptor superfamily of ligand-activated transcription regulators. There are two different genes that encode two different TRs, TR alpha and TR beta, and these two TRs are often co-expressed at different levels in different tissues. Most thyroid hormones do not discriminate between the two TRs and bind both with similar affinities. RESULTS: We have designed and synthesized a thyroid hormone analog that has high affinity for the TRs and is selective in both binding and activation functions for TR beta over TR alpha. The compound, GC-1, was initially designed to solve synthetic problems that limit thyroid hormone analog preparation, and contains several structural changes with respect to the natural hormone 3,5,3'-triiodo-L-thyronine (T3). These changes include replacement of the three iodines with methyl and isopropyl groups, replacement of the biaryl ether linkage with a methylene linkage, and replacement of the amino-acid sidechain with an oxyacetic-acid sidechain. CONCLUSIONS: The results of this study show that GC-1 is a member of a new class of thyromimetic compounds that are more synthetically accessible than traditional thyromimetics and have potentially useful receptor binding and activation properties. The TR beta selectivity of GC-1 is particularly interesting and suggests that GC-1 might be a useful in vivo probe for studying the physiological roles of the different thyroid hormone receptor isoforms.

Hormone-dependent coactivator binding to a hydrophobic cleft on nuclear receptors.

The ligand-binding domain of nuclear receptors contains a transcriptional activation function (AF-2) that mediates hormone-dependent binding of coactivator proteins. Scanning surface mutagenesis on the human thyroid hormone receptor was performed to define the site that binds the coactivators, glucocorticoid receptor-interacting protein 1 (GRIP1) and steroid receptor coactivator 1 (SRC-1). The residues involved encircle a small surface that contains a hydrophobic cleft. Ligand activation of transcription involves formation of this surface by folding the carboxyl-terminal alpha helix against a scaffold of three other helices. These features may represent general ones for nuclear receptors.

Mutations in the conserved C-terminal sequence in thyroid hormone receptor dissociate hormone-dependent activation from interference with AP-1 activity.

A short C-terminal sequence that is deleted in the v-ErbA oncoprotein and conserved in members of the nuclear receptor superfamily is required for normal biological function of its normal cellular counterpart, the thyroid hormone receptor alpha (T3R alpha). We carried out an extensive mutational analysis of this region based on the crystal structure of the hormone-bound ligand binding domain of T3R alpha. Mutagenesis of Leu398 or Glu401, which are surface exposed according to the crystal structure, completely blocks or significantly impairs T3-dependent transcriptional activation but does not affect or only partially diminishes interference with AP-1 activity. These are the first mutations that clearly dissociate these activities for T3R alpha. Substitution of Leu400, which is also surface exposed, does not affect interference with AP-1 activity and only partially diminishes T3-dependent transactivation. None of the mutations affect ligand-independent transactivation, consistent with previous findings that this activity is mediated by the N-terminal domain of T3R alpha. The loss of ligand-dependent transactivation for some mutants can largely be reversed in the presence of GRIP1, which acts as a strong ligand-dependent coactivator for wild-type T3R alpha. There is excellent correlation between T3-dependent in vitro association of GRIP1 with T3R alpha mutants and their ability to support T3-dependent transcriptional activation. Therefore, GRIP1, previously found to interact with the glucocorticoid, estrogen, and androgen receptors, may also have a role in T3R alpha-mediated ligand-dependent transcriptional activation. When fused to a heterologous DNA binding domain, that of the yeast transactivator GAL4, the conserved C terminus of T3R alpha functions as a strong ligand-independent activator in both mammalian and yeast cells. However, point mutations within this region have drastically different effects on these activities compared to their effect on the full-length T3R alpha. We conclude that the C-terminal conserved region contains a recognition surface for GRIP1 or a similar coactivator that facilitates its interaction with the basal transcriptional apparatus. While important for ligand-dependent transactivation, this interaction surface is not directly involved in transrepression of AP-1 activity.

A structural role for hormone in the thyroid hormone receptor.

The crystal structure of the rat alpha 1 thyroid hormone receptor ligand-binding domain bound with a thyroid hormone agonist reveals that ligand is completely buried within the domain as part of the hydrophobic core. In addition, the carboxy-terminal activation domain forms an amphipathic helix, with its hydrophobic face constituting part of the hormone binding cavity. These observations suggest a structural role for ligand, in establishing the active conformation of the receptor, that is likely to underlie hormonal regulation of gene expression for the nuclear receptors.

Expression of the rat alpha 1 thyroid hormone receptor ligand binding domain in Escherichia coli and the use of a ligand-induced conformation change as a method for its purification to homogeneity.

The rat alpha 1 thyroid hormone receptor (rTR alpha 1) mediates hormone-dependent gene regulation by utilizing several distinct structural domains, including those containing DNA and ligand binding sites. Binding of the hormone to the ligand binding domain (TR-LBD) induces conformational changes in the receptor that are involved in affecting the receptor's transcriptional regulatory and other functions. A 33-kDa protein fragment (Met122-Val410) of rTR alpha 1, which includes the entire TR-LBD, was expressed in Escherichia coli, yielding typically 1.5 mg of soluble TR-LBD/liter of bacteria. The protein was purified to > 99% homogeneity with a final yield of 24% by hydrophobic interaction, DEAE anionic exchange, and heparin cationic exchange chromatographic steps. The Kd of the purified TR-LBD for 3,3',5-triiodo-L-thyronine (T3) was 0.06 nM, identical to that for full-length rTR alpha 1. T3 analogs had affinities consistent with values obtained for full-length rTR alpha 1. In all three chromatography steps, TR-LBD prebound to [125I]T3 eluted earlier than the unliganded TR-LBD, like the full-length receptor. These studies indicate that the binding affinity and specificity of the TR-LBD are similar to those of the intact rTR alpha 1 and that the ligand-induced conformational changes occur in the LBD itself. These studies also provide methodology for obtaining milligram quantities of protein useful for biochemical and biophysical studies of the thyroid hormone receptor and its ligand-induced changes.

Heterodimerization and deoxyribonucleic acid-binding properties of a retinoid X receptor-related factor.

The extent thyroid hormone receptors (TRs) bind to AGGTCA-related motifs as monomers and/or homodimers, and as heterodimers with retinoid X receptors (RXRs) depends on the number, spacing, and orientation of these half-sites. Here we show that recombinant RXR alpha affects TR binding to DNA in diverse ways; it enhances recombinant TR beta 1 binding to four-nucleotide-spaced direct repeat and palindromes but not to inverted palindromes. We also used an endogenous factor termed RXR alpha-RF that cross-reacted with antibodies to RXR alpha and copurified and formed heterodimers on DNA with rat liver TRs (mostly TR beta 1 isoform), supporting the fact that endogenous TRs are commonly heterodimers. RXR alpha-RF formed, like recombinant RXR alpha, heterodimers on DNA with vitamin D and retinoic acid but not estrogen receptors. RXR alpha-RF differed from recombinant RXR alpha in that it provoked enhancement of TR beta 1 binding to DNA irrespective of half-site architecture, was resistant to heating to 50 C, and did not form heterodimers with recombinant TR alpha 2 on four-nucleotide-spaced direct repeat. The overall enhancement of TR-DNA recognition by endogenous RXR alpha-RF, not found in studies with recombinant RXR alpha, might exemplify properties acquired in vivo by endogenous RXRs; this could promote wider DNA recognition by TRs and expand the thyroid hormone transcriptional influence in the cell.

Delineation of three different thyroid hormone-response elements in promoter of rat sarcoplasmic reticulum Ca2+ATPase gene. Demonstration that retinoid X receptor binds 5' to thyroid hormone receptor in response element 1.

Thyroid hormone (3,5,3'-triiodothyronine) positively regulates transcription of the sarcoplasmic reticulum Ca2+ATPase gene in rat heart, and sequences within 559 nucleotides upstream from the transcription start site confer thyroid hormone responsiveness upon a reporter gene. In the present study, three thyroid hormone-response elements (TREs) are identified between nucleotides -485 and -190. Each TRE is active in transient transfection assays and specifically binds 3,5,3'-triiodothyronine receptors (TRs) alpha 1 and beta 1 alone and in combination with retinoid X receptors (RXRs) alpha and beta. TRE 1 is a direct repeat of two half-sites separated by four nucleotides; TREs 2 and 3 are inverted palindromes of two half-sites separated by four and six nucleotides, respectively. Methylation interference analysis of TRE 1 showed binding of a TR alpha 1 monomer to the 3' half-site, whereas the heterodimer contacts both half-sites. Subsequent studies employed TR beta and RXR alpha mutants in which their P-boxes were replaced with the P-box of the glucocorticoid receptor. Bandshifts of wild type and mutant proteins with either wild type TRE 1 or a mutant version, in which the 5' half-site was converted to a glucocorticoid response element half-site, demonstrated preferential binding of RXR to the 5' half-site and of TR to the 3' half-site of TRE 1.

Roles of 3,5,3'-triiodothyronine and deoxyribonucleic acid binding on thyroid hormone receptor complex formation.

Thyroid hormone receptors (TRs) bind to thyroid hormone response elements (TREs) in the promoter region of target genes as monomers, homodimers, and heterodimers with nuclear proteins such as retinoid-X receptors (RXRs). Recently, we observed that T3 decreased TR homodimer, but not TR/RXR heterodimer, binding to TREs, suggesting that the latter complexes may be involved in transcriptional activation of target genes. However, little is known about TR complexes that form in solution. Thus far, there have been only limited studies comparing TR complex formation in solution and on DNA as well as examining the effects of T3 and the putative ligand for RXRs, 9-cis retinoic acid (9-cis RA), on TR complex formation. In this paper, we used a coimmunoprecipitation assay with anti-TR beta 1 antibody and the electrophoretic mobility shift assay under similar buffer and incubation conditions to demonstrate that in the absence of T3, TR beta 1 is present as a monomer in solution and binds primarily as a homodimer to the chicken lysozyme TRE, F2. In the presence of T3, TR beta 1 cannot form a homodimer on F2, but, instead, exists as a liganded monomer in solution. Kinetic studies demonstrated that T3 markedly increased the dissociation rate of TR homodimer from F2. Using similar methods, we observed TR beta 1/RXR alpha heterodimer formation in solution and 10-fold greater formation on F2. Neither T3 nor 9-cis RA significantly affected TR beta 1/RXR alpha heterodimer formation. Taken together, these results suggest that both T3 and TRE binding are important determinants of the formation of specific TR complexes in solution and on DNA.

Preliminary crystallographic studies of the ligand-binding domain of the thyroid hormone receptor complexed with triiodothyronine.

A truncated, recombinant form of the thyroid hormone receptor, including the hormone binding domain, has been co-crystallized with the hormone T3. The crystals are monoclinic, most likely space group P2, with two molecules per asymmetric unit and cell dimensions a = 63.6 A, b = 80.8 A, c = 100.9 A and beta = 92.1 degrees. The crystals diffract to only medium resolution and decay rapidly in the X-ray beam using laboratory sources. By contrast, high resolution, high-quality data are obtained using synchrotron radiation in conjunction with cryocrystallography.

Phosphorylation selectively increases triiodothyronine receptor homodimer binding to DNA.

Thyroid hormone receptors (TRs) are ligand-regulated transcription factors that bind to thyroid hormone response elements (TREs) as monomers and homodimers, and as heterodimers with nuclear proteins such as TR auxiliary proteins and retinoid X receptors. Recently, bacterially expressed human TR beta-1 (hTR beta-1) was shown to be phosphorylated in vitro by HeLa cytosolic extract. However, little is known about the consequences of phosphorylation on the nature of TR complexes. Therefore, we studied the effect of phosphorylation on TR binding of TREs. Bacterially expressed hTR beta-1 was phosphorylated in vitro with ATP by HeLa cytosolic extract. The ratio of phosphoserine to phosphothreonine was approximately 5:1. We then analyzed phosphorylated hTR beta-1 binding to several TREs by electrophoretic mobility shift assay. Phosphorylated hTR beta-1 bound better as a homodimer to the TREs than hTR beta-1 incubated with preheated cytosolic extract. Alkaline phosphatase treatment of the phosphorylated hTR beta-1 eliminated the enhanced homodimer binding to DNA. In contrast, phosphorylation did not affect TR/TR auxiliary protein or TR/retinoid X receptor heterodimer binding to DNA. Triiodothyronine decreased both phosphorylated and unphosphorylated hTR beta-1 homodimer binding to several TREs, and the addition of okadaic acid did not alter this triiodothyronine effect. These results indicate that phosphorylation, in addition to ligand binding, modulates TR dimer binding to TREs. As such, it is possible that phosphorylation may also participate in TR-mediated regulation of transcription.

Thyroid hormone alters in vitro DNA binding of monomers and dimers of thyroid hormone receptors.

T3 binds to intranuclear thyroid hormone receptors (TRs) on target DNA elements and exerts profound influences on gene expression by mechanisms not yet characterized. We used gel shift assays and cross-linking experiments to demonstrate that T3 greatly induced the monomeric binding of the hTR beta produced in Escherichia coli to DNA. T3 also increased the gel mobility of these monomer-DNA complexes suggesting they undergo a ligand-induced conformational change. This effect did not depend on the orientation and spacing of the half-site motifs within the DNA structure. In contrast, T3 had diverse effects on the dimeric interaction. T3 increased the dimeric interaction to the palindrome GGTCA.TGACC (an effect lost by spacing the half-sites with 3 base pairs) and decreased the dimeric interaction to the inverted palindrome containing the TGACC.GGTCA motif. Scatchard analyses indicated that the T3 enhancement on binding was due to an increase in the number of TR with high affinity DNA-binding activity and not by increasing the affinity of TR that could bind to DNA. The effects of various T3 analogs were directly related to their affinities for the TR. These ligand effects on in vitro TR-DNA binding may reflect mechanisms by which T3 regulates transcription in vivo.