1. Spontaneous release of hormones from the hypothalamus and pituitary that stimulate thyroid hormone production

2. Feedback inhibition of production and release of these hypothalamic and pituitary hormones

3. Rates of intracellular conversion of T₄ to T₃

4. Metabolic destruction of both of these forms of the hormone⁶

Import and Export

Thyroid hormone entry into cells is mediated by specific transporters (see Fig. 5-2).^11–13 While early models suggested that lipophilic thyroid hormone molecules enter cells by diffusion across the plasma membrane, and this may occur to some extent, it is now clear that T₄ and T₃ entry mostly involves facilitated transport, a form of passive diffusion that requires membrane transport proteins.

Several proteins involved in thyroid hormone import and efflux across the plasma membrane of target cells have been identified. However, the quantitative contributions of these and possibly other transporters to overall thyroid hormone entry and export have not been established, and concentrations of various transporters in different tissues may vary. Potential transporters include members of several different multigene families: monocarboxylate transporters (MCTs), organic anion transporters (OATP), l-amino acid transporters, multidrug resistance–associated proteins, fatty acid translocase, and Na⁺/taurocholate-cotransporting polypeptide.

Presently, the best-characterized thyroid hormone transporter is the X-linked MCT8 isoform, which is important for thyroid hormone import into the brain. Expression of the MCT8 gene in cultured cells and frog oocytes enhances thyroid hormone transport into cells. Inactivating human MCT8 mutations are associated with severe neurologic defects in affected males, suggesting that MCT8-dependent thyroid hormone transport into neurons is essential for proper brain development. There are also changes in circulating thyroid hormones (low T₄ and high T₃). Analysis of MCT8 mutants in cultured cells reveals that the mutations abolish T₃ transport, likely explaining elevated circulating T₃ levels. Targeted deletion of the mouse MCT8 gene leads to alterations in thyroid hormone levels that are similar to those of affected humans, although mice are devoid of the neurologic defects. Here, liver thyroid hormone levels are normal, suggesting that other transporters are more important in this tissue.

Of other transporters, recent observations show that a closely related protein, MCT10, can also mediate T₄ and T₃ import into cells. MCT10 is a more effective transporter than MCT8 and displays a preference for T₃. It is widely expressed and particularly abundant in the liver, intestine, kidney, and placenta. Finally, OATP1C1-expressing cells exhibit preferential import of T₄ and rT₃. The protein is distributed widely in the brain, with locations consistent with a specific role in thyroid hormone transport across the blood-brain barrier. It is not clear whether there are other high-affinity thyroid hormone transporters that are yet to be identified or whether apparently low-affinity transporters may play a physiologically important role in some tissues.

Less is known about mechanisms involved in thyroid hormone export from cells. It is clear that MCT8 and MCT10 enhance both thyroid hormone uptake and export, but it is not clear whether these proteins are physiologically relevant hormone exporters or whether other, undescribed, exporters may alsoplay a role.

In some tissues—brown fat in the mouse is an example—hormone response is dependent on T₃ that is generated from intracellular T₄. The implication here is that free T₄ mostly penetrates these cells, where it constitutes a reservoir for T₃ generation, and that tissue uptake of T₃ is much lower. Regulated conversion of T₄ to T₃ provides an important control point; induction of the type 2 deiodinase via induction of second messenger pathways regulates T₃ levels and therefore the extent of the thyroid hormone response (also discussed later).

Accumulation of Thyroid Hormone in Cells

While free thyroid hormones are present at low concentrations in the circulation, levels of hormone can be much higher in target tissues. Although the extent to which this pool of hormone is active is not known, this observation suggests that there are mechanisms which permit thyroid hormone accumulation in target cells.

T₃ binds with nanomolar affinity to an intracellular protein that was originally called cytoplasmic T₃-binding protein (CTBP).¹⁴ This protein is found at high abundance in the kidney and, to a lesser extent, in liver. Because CTBP was later found to be homologous to mu-crystallin, a protein abundant in the lens of the kangaroo eye, the gene is now often called CRYM. Overexpression of CRYM in stable cell lines increases maximal cellular T₃ binding capacity and decreases T₃ efflux rates. However, these effects are coupled to reduced transcriptional responses to T₃, suggesting that CRYM sequesters T₃ in an inactive cytoplasmic complex, away from nuclear TRs.

CRYM is important for thyroid hormone action in vivo. Mice with a targeted CRYM gene deletion appear normal but exhibit moderate decreases in circulating T₄ (25%) and T₃ (13%) levels and perhaps more importantly, also exhibit extremely rapid rates of T₃ entry into and escape from target tissues, presumably because there is no CRYM to sequester cytoplasmic hormone from export mechanisms. Human patients with a natural CRYM mutation that blocks its ability to bind T₃ (K314T) are deaf, a known consequence of defective thyroid hormone signaling during development. Interestingly, CRYM requires NADPH for dimerization and T₃ binding (Fig. 5-3). As NADPH levels are reflective of levels of cellular reducing power and anabolic capacity, it is conceivable that CRYM may sequester or release T₃ in response to alterations in cellular metabolic status. This notion is attractive but not proven.

There could be other T₃-binding proteins in addition to CRYM and the nuclear TRs (which are described later). Radioisotope-labeled T₃ interacts with proteins in the endoplasmic reticulum, mitochondrion, and nuclear envelope.² These putative T₃-binding proteins have not been identified, and the functional significance of these interactions is unknown.

Deiodinases in Thyroid Hormone Activation and Inactivation

The thyroid gland produces mostly thyroxine. Limited amounts of T₃ and rT₃ are secreted by the gland, but more than 80% of both these forms of thyroid hormone are produced in the periphery via actions of specific deiodinases, D1, D2, and D3.15,16 D1 and D2 remove the iodine group from the 5′ position of the outer thyronine ring. As T₃ binds to TRs with higher affinity than T₄, the actions of D1 and D2 deiodinases serve to increase thyroid hormone activity through generation of T₃, and this represents a step-up in activity. Conversely, the action of D3 decreases hormone activity and represents a step down by removing the iodine group from the 5 position of the inner thyronine ring and converting T₄ and T₃ to rT₃ and 3,3′-T₂, respectively, neither of which interacts substantially with nuclear TRs at physiologic concentrations. Changes in deiodinase expression and activity are important control points in thyroid hormone signaling; they influence the production and availability of the biologically active form of thyroid hormone, T₃.

D2 activity generates most plasma T₃ in euthyroid (defined as levels of thyroid hormone within normal range) conditions via conversion of T₄ to T₃ in peripheral tissues. Comparisons of rates of T₃ production by D2 and D1 indicate that D2 is probably more important for T₄ to T₃ conversion at physiologic T₄ concentrations; it displays a higher K_m value for substrate than D1 and is likely to be sensitive to alterations in T₄ concentrations and more active than D1 at physiologic hormone concentrations. In addition, environmental stimuli regulate D2 activity catalyzing local T₃ production from T₄ in certain tissues. For example, signals such as cold exposure increase D2 expression in brown fat (discussed earlier), and the resulting increase in intracellular T₃ promotes induction of uncoupling protein with consequent heat generation. These lines of evidence suggest that D2 is important in generating T₃ that activates TRs. Accordingly, D2 is located in the perinuclear region where it can deliver T₃ efficiently to TRs.

D1 acts in a similar way to D2, but its functions are not as clear as with D2, because it displays a relatively high K_m value for T₄ to T₃ conversion (i.e., the enzyme binds the T₄ substrate with low affinity). D1 is highly expressed in liver and kidney, and its expression is induced by thyroid hormone. While previous models suggested that D1 is important for regulation of circulating T₃, more recent models suggest that D1 may only contribute substantially to plasma T₃ in conditions of thyroid hormone excess. D1 may also be involved in 5′ deiodination of other forms of thyroid hormone, including rT₃.

D3 is important for clearance of plasma T₃.¹⁷ It is highly expressed in human placenta, where it is thought to be important for protection of the developing fetus from effects of maternal thyroid hormones. D3 is also highly expressed in the brain and skin. Increased D3 levels may account for reductions of circulating thyroid hormone levels that are often observed in critically ill patients and states of inflammation. Finally, D3 is often overexpressed in vascular tumors, and this in turn can result in marked reductions in circulating thyroid hormone levels.

Alternate Metabolic Fates of Thyroid Hormone: Inactive and Active Derivatives

In addition to 5′ deiodination, T₄ undergoes other modifications which are generally believed to inactivate the hormone.¹⁸ About 45% of circulating T₄ is converted to rT₃ and 35% to T₃. T₃ and rT₃ undergo further deiodinations to create T₄ derivatives with all possible combinations of iodinated and deiodinated inner and outer rings (3,3′-T₂, 3,5-T₂, 3′,5-T₂, 3′-T₁, 3-T1 and T₀), none of which bind significantly to nuclear TRs. T₄ and T₃ can also be inactivated by glucuronidation in liver, followed by secretion into the bile or by sulfation in liver and kidney and excretion in the urine.

The fact that most secreted T₄ is rapidly converted to apparently inactive forms of the hormone has led investigators to question whether some of these T₄ derivatives have unappreciated functions in activation of alternate receptors or as metabolic precursors of different active forms of the hormone. As mentioned earlier, one of the most prominent, rT₃, is a very weak thyroid hormone partial agonist that is unlikely to exert significant actions through the TRs. However, there are two cases in which alternate metabolic modification of T₄ does produce biologically active forms of hormone:

Basal Metabolic Rate

One of the most important effects of thyroid hormone involves changes in basal metabolic rate (BMR) in multiple tissues.21–24 Thyroid hormones stimulate oxygen consumption (indicative of enhanced metabolism) in multiple locations, including skeletal muscle, liver, kidney, and intestine. Increases in BMR are probably partly related to increased mitochondrial activity and number. Importantly, thyroid hormones induce expression of uncoupling proteins (UCPs), mitochondrial membrane proteins that dissipate the proton gradient in the absence of ATP synthesis, thereby converting potential energy to heat. This effect is probably a contributor to thyroid hormone–dependent increases in BMR.

Thyroid hormone does not always enhance BMR. For example, BMR is not enhanced by thyroid hormone in most regions of the brain, and the hormone actually suppresses metabolic activity in the pituitary. The receptors in these tissues are functional, suggesting that key mediators of thyroid hormone regulation of BMR are blocked or absent.

Tissue Effects

Thyroid hormone exerts specific effects on growth, development, and metabolism in a variety of tissues. Some of these effects are summarized briefly here:

Systematic Analysis of Thyroid Hormone Target Genes

Many thyroid hormone–responsive genes involved in the responses described earlier are known.² For example, in liver, thyroid hormone induces carnitine palmitoyl transferase 1a (that mediates the rate-limiting step in fatty acid oxidation); glucose-6-phosphatase and phosphoenolpyruvate carboxykinase (rate-limiting steps in gluconeogenesis); the NR coregulator PGC-1α, which is an important transcriptional coregulator that stimulates genes important for mitochondrial biogenesis, gluconeogenesis, and fat oxidation; CYP-7a1 (cholesterol-to–bile acid conversion); fatty acid synthetase; acetyl-CoA carboxylase; malic enzyme (increases fat synthesis); and spot 14, which is required for fat synthesis in some tissues and up-regulated in contexts in which carbohydrates are converted to fat. Many other hormone-responsive genes are probably still unknown.

The earliest systematic attempt to define thyroid hormone response was performed in the late 1970s. Use of two-dimensional (2D) gels to analyze extracts from radiolabeled methionine–pulsed cells revealed that about 1% of 1000 rat liver proteins change in response to altered thyroid hormone.36,37 There were inductions and repressions, and the studies began to define the “domain” of the thyroid hormone response. Recent advances in gene-profiling technology have permitted much more detailed descriptions of hormone-dependent alterations in gene profile. Presently there have been relatively few systematic genome-wide descriptions of thyroid hormone responses.^25,26,38,39 An early study of thyroid hormone action in mouse liver revealed that about 1% of mouse genes change in response to altered thyroid hormone (hypothyroid versus hyperthyroid),²⁵ confirming estimates from 2D gels. This study revealed target genes involved in carbohydrate, fat, and amino acid metabolism and unexpected thyroid hormone–regulated pathways; for example, genes involved in apoptosis are induced. The experiment also provided insights into patterns of thyroid hormone–responsive gene expression. Most regulated genes (>65%) in liver were repressed by thyroid hormone. This preponderance of negatively regulated genes is not seen in all tissues. Target genes are generally up-regulated by hormones in human muscle primary culture. Later studies of thyroid hormone patterns of gene expression in TR knockout mice provided insights into TR-specific effects and subtle differences in thyroid hormone–responsive gene expression and are discussed in more detail later in the chapter. Systematic screening of thyroid hormone–responsive genes should provide detailed descriptions of novel target genes and pathways in different tissues.

Thyroid Hormone–Regulated Micrornas

MicroRNAs (miRs) are small (18 to 25 nucleotides), noncoding RNAs that hybridize with coding mRNAs to inhibit translation and, in some cases, also promote mRNA degradation.⁴⁰ The enormous roles these miRs play in regulation is just beginning to be appreciated. MiRs are produced by cleavage and processing of large primary transcripts, which often code for proteins. Thus, many of the influences that regulate expression of parental primary transcripts also regulate expression of the associated miR. While effects of thyroid hormone on miR expression have not been studied extensively, one miR (miR-208) lies within a noncoding region of the thyroid hormone–responsive human αMHC transcript (and is therefore up-regulated by thyroid hormone itself). MiR-208 is implicated in up-regulation of stress-dependent cardiac responses. It is likely that other thyroid hormone–responsive miRs exist, and they may play important roles in hormone effects on protein translation.

Discovery of Thyroid Hormone Receptors and Their Mechanism of Action

TRs were identified in the 1970s.42,43 Following the discovery of several nuclear receptors, including the estrogen and glucocorticoid receptors, investigators in the thyroid hormone field utilized similar approaches to look for TRs. Specific binding of radiolabeled T₃ to liver cells was initially discovered by Oppenheimer and colleagues using intact cells and later characterized further by Samuels and colleagues, who found that the mechanism of TR actions differed mechanistically from the steroid receptors, which translocate into the nucleus on ligand binding. For TRs, hormone could bind directly to nuclei, and the location of receptors did not alter with T₃.

In the late 1970s and early 1980s, multiple aspects of the receptors were revealed.36,44 Analysis of properties of TRs revealed a single class of high-affinity binding sites with K_d values for T₃ in the 0.1 nM range. Specific hormone-binding sites were observed in extracts of several responsive tissues, including liver, anterior pituitary, brain, and heart. Estimates of the number of binding sites per cell suggested that the proteins are rare; highly responsive tissues only contained about 10,000 specific hormone-binding sites per cell. Evidence that the receptor preparations correspond to physiologic hormone targets came from observations that affinities for different thyroid hormones parallel their biological potencies (Triac>T₃>T₄>rT₃). Partial TR purification and photo-affinity labeling revealed two major nuclear hormone-binding proteins with molecular weights of 46 and 57 kD, which were later found to correspond to products of distinct genes—the TR α and β isoforms—described later in the chapter. TRs were also found to be associated with chromatin, and preferentially with active chromatin, suggesting that they influence gene expression through interactions with DNA.^45–48

Discovery That Thyroid Hormones Regulate Specific mRNAS and Gene Transcription

As molecular biology techniques became available, it was possible to isolate specific mRNAs from tissues where they were abundant and to translate them. By the 1970s, regulation of specific mRNAs had been demonstrated for estrogens, progestins, and glucocorticoids. Growth hormone (GH) mRNA was abundant in rat pituitary cells, and Baxter and colleagues were able to demonstrate that thyroid hormones could specifically regulate GH mRNA, conclusively demonstrating for the first time that thyroid hormones alter the levels of specific mRNAs. The same group cloned rat GH gene sequences, and these permitted demonstrations that thyroid hormones increased mRNA hybridizable to a rat GH cDNA probe. The cloned GH sequences were also used in so-called “polymerase run-on experiments” that showed that hormones stimulate transcription rather than induce some posttranscriptional modification of the mRNA. These studies resolved the controversies prevalent at the time of experiments by Tata and colleagues, discussed previously. Thus the fundamental concept of thyroid hormone action whereby the receptor regulates the transcription of specific genes was established.

Discovery That TRs Bind Preferentially to Active Chromatin and Specific DNA Sequences

Studies during the 1970s and 1980s also set the stage for understanding the nature of TR binding sites and their relationship to active chromatin. GH genomic DNA was used for interaction studies with partially purified TR preparations, demonstrating that TRs bound to specific DNA sequences. These experiments were forerunners of the now familiar concept that receptors act by binding to specific thyroid hormone response elements (TREs). That the receptors could affect DNA was further evidenced by the finding that they induce bending of rat or human TR DNA binding sites. Subcellular fractionation revealed that TRs were associated with chromatin, and preferentially with active chromatin, suggesting a relationship between the receptors and transcriptionally active chromatin.45–48 These studies set the stage for definition of thyroid hormone response elements (TREs).

Cloning of TRs

TRs were cloned in 1986.49,50 They were identified as cellular homologs of a retroviral oncogene (cancer-causing gene), v-erbA,⁵¹ which causes erythroblastoma in chickens. Isolation of these cDNAs occurred after those for other members of the NR family, the glucocorticoid and estrogen receptors, which led to the realization that v-erbA was derived from an NR. Both of the v-erbA homologs bound thyroid hormone with high affinity and appropriate specificity. The Vennstrom group clone was most closely related to v-erbA and was originally designated as c-erbA, but it is now called thyroid hormone receptor α (TRα). As compared to TRα, v-erbA contains 17 different amino acid substitutions and a C-terminal truncation, which are now known to prevent the viral oncogene from binding hormone.⁴⁹ The Evans group clone encoded a highly homologous but distinct 57-kD protein that was closely related to TRα and v-erbA but was the product of the THRB gene.⁵⁰ The characterization of TR genes (THRA and THRB), which encode the two major TR isoforms (TRα1 and TRβ1) and several variants that arise from differential splicing (described later), helped to reveal that TRs belong to the NR family, which includes receptors for the steroid hormones, vitamins A and D, and other small lipophilic molecules.⁴

Identification of these receptor cDNAs paved the way for many studies of receptor structure and function. Although it was known at the time that NRs were composed of modular domains, the receptor structures provided the primary sequences in these domains and the abilities to express them and study their functions. In addition, the availability of TR cDNAs paved the way for functional studies, analog development, and improvements in understanding TR mechanisms. While these studies are not described in historical context, there have been rapid advancements in understanding of TR action, including the definition that unliganded TRs are transcriptionally active, leading to a new paradigm for thyroid hormone action; definition of TRE sequences; identification of retinoid X receptor (RXR) heterodimer partners and coregulators; x-ray structural analysis of DBDs and LBDs, which demonstrated a new and unexpected concept that the receptor folds around the ligands; improved understanding of TR function through generation of TR gene knockout animals; and development of new thyroid hormone analogs with selective actions.

Hormone Binding and Occupancy

Measurements of thyroid hormone affinity for its receptors reveal that T₃ binds to purified and recombinant TRs with high affinity. T₃ binds with about 10- to 15-fold greater affinity than T₄ (K_d values are around 2 nM for T₄ and 0.2 nM for T₃). These values appear higher than measured concentrations of circulating free hormones (0.02 nmol for T₄ and 0.06 nmol with T₃), implying that receptor occupancy could be minor in normal conditions. However, based on direct estimates of occupancy of nuclear TRs in different tissues after radiolabeled T₃ and T₄ injection into hypothyroid rats, about 25% to 35% of liver and kidney TRs and 58% of pituitary TRs become occupied with thyroid hormones in euthyroid conditions. Thus, there appear to be mechanisms to bring the hormone into the correct range to bind and activate receptors. One possibility was described earlier: intracellular generation of T₃ from T₄ leads to increased T₃ production in target cells. Facilitated transport alone would not be enough to concentrate T₃, but combinations of transport and cytoplasmic sequestration by CTBP/CRYM or similar proteins could play a role. Finally, TR complexes with stabilizing coactivators could display higher affinities for target hormones than isolated TRs tested in vitro or partially purified from cell extracts. In any case, it is clear that thyroid hormone concentrations are in the correct range to achieve reasonable receptor occupancy in target tissues under euthyroid conditions.

TR Action

Unlike classic models of hormone action in which unliganded receptor is inactive and hormone binding triggers its activity, TRs are transcriptionally active in the absence and in the presence of hormone (Fig. 5-5).⁵² Rather than activating the receptor, hormone binding alters the spectrum of its actions.

TRs bind constitutively to specific DNA sequences termed thyroid hormone response elements (TREs) in the proximal promoter of positively regulated target genes. As noted earlier, TRs function mostly via heterodimerization with another NR, RXR, which is a partner for many members of the NR family. RXR binds vitamin A derivatives and unsaturated fatty acids but is thought in many cases to be silent and unoccupied by ligands, when it forms complexes with TRs. Unliganded TRs repress transcription of positively regulated genes by recruiting corepressors such as the NR corepressor (N-CoR), silencing mediator of retinoid and thyroid hormone–responsive transcription (SMRT) and alien. Corepressors actively silence gene transcription by binding histone de-acetylases (mainly HDAC3) which condense local chromatin to prevent access of RNA polymerase II to target promoters.⁵³ Hormone reverses these inhibitory effects and further stimulates gene expression by promoting corepressor release and subsequent recruitment of several different coactivator complexes which enhance transcription in different ways: ATP-dependent remodeling of local chromatin, catalysis of specific histone modifications that mark the genes for transcriptional activity, and enhancement of recruitment and processivity of the basal transcription machinery.

The converse process can also occur.⁵⁴ Unliganded TRs stimulate gene transcription, and binding of T₃ overcomes this process, resulting in repression of gene transcription. Generally, hormone also suppresses transcription below levels stimulated by the unliganded TRs. One model to explain this phenomenon is that coactivators and corepressors exert opposite actions at this type of gene, and there are several examples where so-called NR coactivators repress gene transcription, and corepressors act as coactivators that bind to the “corepressor binding site” (Fig. 5-6). However, several mechanisms of NR repression have been described, so there are other possibilities. For example, ligand-dependent activation of peroxisome proliferator–activated receptors (PPARs) and liver X receptors permits covalent attachment of lysine residue in the LBD to a large 100-amino-acid protein called small ubiquitin-like modifier (SUMO), which arrests corepressor complexes at target promoters (Fig. 5-7).⁵⁴ This effect counters proinflammatory stimuli that promote corepressor release and gene induction and therefore acts as a repressive mechanism. Negative gene regulation may not involve direct receptor/DNA contacts.⁵⁵ Although promoters of some genes that are repressed in response to ligand contain variant TREs, negative regulation may also involve TR interactions with heterologous DNA-bound transcription factors (such as AP-1 and CREB) instead of promoter DNA itself, as has been demonstrated for other NRs. Further investigation will be needed to clarify mechanisms of transcriptional repression at individual genes.

Models of thyroid hormone response emphasize differences in TR complex formation in the absence of hormone and the presence of saturating levels of hormone. As discussed previously, physiologic levels of hormone do not saturate all receptors under these conditions,³⁹ so thyroid hormone–responsive genes will probably be occupied by a mix of liganded and unliganded TRs, with different associated factors. The consequences of partial occupancy of DNA-bound receptors with hormone are far from clear; there could be active competition between liganded and unliganded TRs for the same sites in the same cells, stochastic choices between formation of liganded and unliganded TR-associated complexes in different cells, or other differences.

Secondary Structure

TRs and other NRs are single polypeptide chains that adopt a common modular domain structure. Particular functions can be assigned to specific receptor domains (Fig. 5-8).^2,4 Use of basic recombinant DNA technology to create short cDNA fragments corresponding to each domain and targeted mutagenesis coupled with expression and functional analysis revealed the role of each domain.

The central DNA-binding domain (DBD) is highly conserved throughout the NR family. The DBD mediates receptor-DNA contact, contains a surface that mediates DNA-dependent heterodimer formation with RXR, and possibly has binding sites for other proteins with regulatory functions.

The amino acid sequences of the C-terminal ligand-binding domain (LBD) are relatively well conserved within the NR family, and the domain has a highly conserved common fold. The LBD is multifunctional and contains a ligand-binding cavity, a hormone-dependent transcription function (AF-2) that binds coactivators, an overlapping corepressor binding surface exposed in the absence of ligands, and a major surface that mediates homodimer formation and heterodimer formation (with RXRs). The LBD may also harbor surfaces that mediate formation of alternate oligomers (trimers, heterotrimers, and tetramers) and interactions with heat shock proteins.

The amino-terminal domain (NTD) is poorly conserved among NRs, suggesting that it is primarily responsible for many of the unique activities of particular receptor subtypes. Transfection analysis reveals that the NTD contains a complex transactivation function (AF-1) that binds coactivators and corepressors, complements AF-2 in some contexts, and acts independently in others. TR AF-1 is required for optimal hormone response at certain promoters and in certain cell types. Accordingly, this domain also harbors binding sites for corepressors and coactivators.

DBD Structure

The DBD contains two zinc “fingers,” protein domains of approximately 30 amino acids in length that require zinc ions for protein folding and stability. The zinc ion is coordinated by tetrahedral contacts with the side chains of four well conserved cysteine residues. Together, the two zinc finger motifs form a single functional DNA-binding domain.

The TR DBD x-ray structure was solved in complex with that of its heterodimer partner RXR on a thyroid hormone receptor response element (TRE), a direct repeat of the consensus sequence AGGTCA spaced by four bases (DR-4 element, see Fig. 5-8).^56,57 DNA contacts are mediated by a large α-helix in the first zinc finger, which docks into the major groove of DNA. This conserved region of the receptor contains the P-box, which dictates NR DNA element recognition specificity. Both the second zinc finger and a long carboxyl-terminal α-helix termed the C-terminal extension (CTE) also make DNA contacts. The second zinc finger interacts with the major groove, and the CTE interacts with the DNA minor groove and phosphate backbone. The organization of the CTE is unique to the TR, and these additional CTE-dependent DNA contacts may explain why TRs, unlike many other NRs, exhibit the capacity to bind to DNA as monomers (developed in following discussions).

RXR and TR DBDs interact with the DR-4 element in a head-to-tail manner, with RXR upstream. This polarity is dictated by the specific heterodimer contact surfaces, which involve multiple interactions between zinc finger regions of both receptors and between the outer face of the TR CTE and RXR. The 5′ location of RXR was already inferred by biochemical analysis⁵⁸; introduction of the glucocorticoid receptor P-box sequence into TR permitted the RXR-TR complex to recognize a hybrid response element composed of one classic TR half-site (AGGTCA) and an equivalent glucocorticoid receptor response element half-site (TCTTGT) but only when the GR binding half-site is placed downstream in the 3′ position. This reflects that the TR preferentially associates with the downstream position.

RXR/TR pairs bind preferentially to DR-4 elements, whereas other RXR/NR pairs recognize direct AGGTCA repeats with different spacings.⁵⁸ Comparison of RXR/TR structures with other RXR/nuclear receptor pairs reveals that the CTE sets this spacing preference.⁵⁷ All NR CTEs fold across the core of the DBD, but different CTEs adopt different trajectories and occlude receptor binding to the wrong element. For RXR-TRs, spacings of less than 4 bp would not be permitted because of steric clashes between the CTE and the RXR DBD. Conversely, introduction of extra bases between the half-sites will impose geometric constraints that prevent proper engagement of the RXR-TR DBD heterodimer surface.

LBD Structure

Although there are now many publicly available NR LBD x-ray structures, TR (rat TRα) was the first structure solved in complex with native hormone (it was reported simultaneously with the structure of the liganded retinoic acid receptor, RAR).59–61 Strikingly, the ligand was buried in the core of the LBD; this was contrary to contemporary models which assumed that ligand would interact with an allosteric site on the receptor surface. The ligand-binding pocket is small and well tailored to cognate agonists such as T₃ and Triac. This organization implies that admission of hormone into the core of the domain promotes its folding into an active conformation.

TR LBD structure strongly resembles other NRs.4,62 The domain is almost entirely helical, composed of 12 α-helices. The α-helices fold into three layers: two outer layers comprise three distinct helices, the central layer comprises two helices, with the additional space occupied by hormone. Several short helical loops and β sheets link the individual α-helices. N-terminal LBD helix (H1) constitutes a backbone that links helical layers. C-terminal helix (H12) adopts different positions in the presence and absence of hormone and dictates cofactor interactions that influence receptor activity.

All NR LBDs possess the same fold, despite having only weak primary sequence similarity. TR LBDs share only 15% primary sequence identity with the progesterone receptor LBD, yet the two receptors exhibit a very similar structure, with only slight variations in the length of α-helices. Most structural variability occurs in the loops between conserved helices and in the extensions to C-terminal H12 (sometimes called the F-domain) that are present in other NRs but not the TRs. The overall conformation similarities of TRs and other NRs suggest that findings that apply to TRs will be general for other NRs and vice versa. Much has been learned about the function and interactions of NR homo-heterodimer surfaces, coregulator binding surfaces, and hormone effects on receptor structure with a combination of x-ray crystallography, approaches to assess LBD dynamics, and mutational analysis, to be discussed later.

Hinge Structure

The hinge domain (or D-domain) was first identified as a poorly conserved region that links the DBD and LBD. Original models suggested that it would behave as an unstructured peptide that facilitates rotation between the LBD and DBD in different contexts. However, TRβ LBD and DBD x-ray structures contain overlapping amino acids, and these reveal that residues that were originally assigned to the hinge actually comprise extensions of H1 of the LBD and the C-terminal portion of the DBD CTE.60,63 The true unstructured hinge is only 3 to 6 amino acids long. Presently it is not clear whether this short region is sufficient for true rotational flexibility between domains. Interestingly, the C-terminal helix of the DBD CTE region appears as a completely unstructured peptide in some LBD crystals but not others, and it has been suggested that this region of the receptor could either fold into a functional helix or unfold to improve rotational flexibility in different contexts.⁶³

NTD Structure

There are no published studies of TR NTD structure. NR N-terminal domains (NTDs) are important for transcriptional response and integration of NR ligand signals with second messenger pathways, but this region is poorly conserved throughout the receptor family and not at all defined at the level of tertiary structure.⁶⁴ Current models indicate that NR NTDs are composed of intrinsically unfolded domains; these types of proteins consist of multiple and potentially overlapping subregions that acquire active conformation on interaction with target cofactors.⁶⁴

Full-Length Receptors

No high-resolution structures of TRs or RXR-TR fragments that make up more than one domain have yet been resolved. There are published low-resolution solution x-ray structures of liganded TR DBD-LBD dimers, and an unusual tetrameric form of unliganded TR DBD-LBD.⁶⁵ However, these data provide sufficient resolution to determine likely shapes of the molecules, but not enough to define the trace of the protein backbone. Use of these low-resolution structures to dock high-resolution x-ray structural models of DBDs and LBDs suggests that in the dimer in solution, LBDs and DBDs are in close proximity to their counterparts in the neighboring subunit and that the DBDs are relatively distant from the LBDs, implying that the hinge is extended. In the solution tetramer, LBD pairs dock against each other with DBDs protruding in an X shape. Here, however, the DBDs are closer to the LBDs than they appear in the dimer, implying that there may be rearrangements in the hinge that are dependent on the oligomeric state of the LBD.

A recent structure of full-length liganded PPARγ in complex with liganded RXRα on a DNA element provides a guide to expected principles of organization of the full-length RXR-TR complex.⁶⁶ The PPARγ-RXRα heterodimer binds to a DR-1 element, with PPARγ in the 5′ position, unlike the TR which occupies the 3′ position (Fig. 5-9). LBDs and DBDs adopt the expected conformation and engage in the expected heterodimer contacts predicted from LBD and DBD structures and mutational analysis, described more fully for the TRs later. However, the PPAR CTE portion of the hinge makes unexpected DNA contacts 5′ to the AGGTCA motif. Moreover, the complex is nonsymmetrical, the PPARγ LBD is close to the RXR DBD, and this facilitates unexpected heterodimer contacts which involve LBD surfaces around helix 3 and the β-sheet region and the surface of the RXR DBD on the opposite side of the molecule from the DNA-binding surface. The RXRα hinge appears unstructured and flexible. This permits the RXR LBD to adopt a more distant position from the DBD than the PPAR LBD and raises the possibility that the RXR hinge is a potential source of rotational flexibility in response-element recognition. Finally, NTDs of both receptors are disordered, in keeping with the prediction that these domains can only adopt appropriate conformation on contact with partners. It will be interesting to determine the structures of full-length TR/RXR to learn how TRs adapt to different elements and whether there are unexpected heterodimer contacts between LBDs and DBDs of these receptors.

Variable TRE Sequence and Conformational Adaptation to DNA

Published liganded TR LBD x-ray structures are monomers.⁶⁷ However, many NR LBDs have been crystallized as homodimers and heterodimers; there is a surface that binds both homo- and heterodimer partners which is large and comprises residues from helices 9, 10, 11, 7, and 8.⁶⁸ Targeted mutagenesis of TRs and RXRs confirms that dimer and heterodimer formation is similar to this common interface, with a hotspot of particularly hydrophobic residues at the junction of H10 and H11 playing very important roles in binding.⁶⁸

Most TREs are composed of DR-4 elements, but there is significant variation in sequence and arrangement of TREs (Fig. 5-10) which results in TR isoform and oligomer-specific differences in DNA recognition. TRs and RXR-TRs bind to direct repeats with different spacing, inverted palindromes (IP), and palindromes (pal).^58,69 TRα- and TRβ-RXR heterodimers bind preferentially to DR-4 but also bind to IP-6 elements (where the DNA binding sites are in an inverted position with respect to each other, with a 6-nucleotide spacer) and palindromic (Pal) elements (where the DNA binding sites are arranged in a head-to-tail position with respect to each other, without spacer DNA). In contrast, TRβ homodimers bind strongly to IPs, weakly to consensus DR-4 elements, and not at all to Pal elements. TRα homodimers bind very weakly, if at all, to these TREs. In addition, TRβ homodimers bind strongly to a subset of variant DR-4 elements,⁷⁰ implying that additional sequence differences can facilitate homodimer interactions. RXR-TRs are the major TR species in living cells, but TRβ homodimers can clearly modulate transcription from some TREs in cell culture and yeast.⁶⁹ Moreover, blockade of RXR expression only inhibits T₃ response at a subset of target genes.⁷¹ Both observations suggest that alternate oligomeric forms of TR can mediate thyroid hormone responses. The precise contributions of RXR-TRs versus other TR oligomers to T₃ signals at different genes in physiologic conditions are not yet clear.

TRs and RXR-TRs must undergo significant conformational adaptation to bind different TREs.⁵⁸ RXR-TR heterodimer and TR-TR homodimer formation involves the same interface at the junction of H10-H11, irrespective of TRE organization. Consideration of relative orientations of LBD and DBD required for binding to each TRE indicates that with DR-4 elements, the upstream LBD must rotate with respect to the DBD to form the heterodimer surface in a head-to-head fashion (Fig. 5-11). Since RXR-TRs preferetially bind DR-4 elements, it is likely that this rotation is provided by the upstream RXR. The TR DBD and LBD do not need to rotate for inverted palindromes, where TR-TR homodimer formation is strongest. Thus, RXR is probably responsible for flexible response-element recognition, and this may contribute to its function as a near-universal heterodimer partner for NRs. The unstructured appearance of the RXR hinge domain observed in the PPAR-RXR full-length structure mentioned earlier suggests that this could be one candidate for the source of flexibility.

Alternate TREs

There are examples where TRs may interact with DNA in other ways.⁵⁸ Several response elements are composed of TRE half-sites that bind TR monomers. TRα and TRβ are unusual in the nuclear receptor family, with their high capacity to bind as monomers to AGGTCA half-sites in addition to more conventional interactions with DNA elements as dimers and heterodimers. In these cases, the recognition site is extended to include the core half-site and additional 5′ flanking sequences.

Natural TREs often contain multiple copies of half-sites. The spot 14 promoter contains two DR-4 elements, and the SERCa2 promoter contains a DR-4 element and two IP elements. In most cases, it is not known whether TRs recognize these elements as separate RXR-TR heterodimers or TR-TR homodimers or as larger assemblies, such as tetramers or heterotetramers. There is one clear example of the latter possibility. The rat growth-hormone promoter contains an unusual trimeric TRE that consists of a classic DR-4 element and a downstream half-site that forms a palindrome, the 3′ DR-4 half-site⁷² (Fig. 5-12). TRβ, but not TRα, forms homotrimers and heterotrimers with RXR that bind this element with high efficiency and cooperativity. Thus TRβ must contain additional subunit interaction surfaces, if it simultaneously engages in typical homo/heterodimer contacts and also binds another TRβ subunit in the 3′ position. TRβ0, a TR splice variant that lacks the NTD and is abundantly expressed in frogs, birds, and reptiles, binds this element with high affinity, suggesting that all surfaces are in the DBD-LBD region.

There is significant variation in TRE half-site spacing and sequence in naturally occurring TREs that likely has consequences for TR activity.4,58 TRs preferentially recognize DR-4 and IP-6 elements but also bind similar elements with different spacings, including DR-3, DR-7, and IP-4. Actual half-site sequences diverge considerably from the classic AGGTCA half-site. In fact, negatively regulated genes sometimes contain weak TR binding sites that instead conform to the consensus TGGTTTGGGGTCCA.⁵⁵ These influences will affect the affinity of the receptor for the element but may also influence receptor activity in other ways. DNA contact alters DBD structure, and a small subset of glucocorticoid-receptor DBD mutations that mimic allosteric switches which occur on DNA binding affect activity of AF-1 and AF-2 and change the response element–specific behavior of the receptor. One of these, a lysine residue at the base of the first zinc finger domain that contacts DNA, alters the activity of all NRs (including TR); it changes NRs from ligand-mediated repressors to ligand-mediated activators at promoters with AP-1 sites, where the receptor acts in the absence of direct DNA contact. Thus the nature of the DNA sequence element could alter both TR conformation and activity.

LBD H12 Position and Coregulator Interactions

The major known effect of hormone binding is stabilization of H12 in an active position in which it packs against the body of the LBD to form a surface called activation function 2 (AF-2)52,61 (Fig. 5-13). Comparisons of liganded TR and other liganded NR structures reveal that H12 is docked against the LBD in a similar position in all “active” liganded structures. Conversely, examination of unliganded structures and structures of NRs in complex with antagonists or partial agonists reveal that H12 can adopt various “inactive” or “partly active” conformations, all highly distinct from the active position of H12. This suggests that H12 may be unstable in the absence of an activating ligand. The idea that ligand binding stabilizes H12 has been confirmed by dynamics studies in which PPARγ H12 mobility was measured directly.⁷³

H12 position influences coregulator interactions.⁵² In the active position, H12 docks against the body of the LBD to form a surface-exposed hydrophobic cleft composed of residues from helices (H) 3, 4, 5, and 12. Saturation mutagenesis reveals that this surface corresponds to the hormone-dependent activation function, AF-2. Mutations in the AF-2 surface abolish LBD hormone-dependent transcriptional activity and interactions with coactivator proteins. AF-2 is not needed for heterodimer formation or for hormone binding, implying that these mutations in the AF-2 surface do not generally disrupt NR structure. Targeted mutation of LBD surface residues and examination of activities of the receptors in the unliganded state reveals that the corepressor binding surface overlaps with the upper part of AF-2 but also extends outside AF-2 and includes the region of the body of the LBD over which H12 is positioned in the active state. Thus, H12 is probably displaced in the absence of hormone, exposing an extended hydrophobic surface capable of corepressor binding, while hormone promotes H12 packing over the lower part of this surface, simultaneously inhibiting corepressor binding and completing the AF-2 surface.

Identification of NR recognition motifs in coactivators and corepressors and their subsequent co-crystallization with receptor LBDs helped to confirm the basic formulation for hormone-dependent coregulator exchange, outlined previously. TR and NR coactivators contain a common recognition motif, short α-helical sequences of the consensus sequence Leu-X-X-Leu-Leu (NR boxes). Co-crystallization of TR and other NRs with representative NR box peptides (12 to 15 amino acids) confirmed that they bind to the AF-2 cleft defined by mutation studies. Corepressors contain a common NR recognition motif composed of consensus sequence Ile-X-X-Ile-Ile-X-X-Leu-Met (these are variously referred to as interaction domains [IDs] or corepressor NR boxes [CoRNR boxes]). These peptides make up a three-turn α-helix, longer than coactivator NR boxes. A co-crystal of the PPARα LBD with an antagonist and a representative ID motif confirm that the corepressor ID interacts with the extended surface defined by targeted mutagenesis.

TR Antagonists: The Extension Hypothesis

It has been possible to create TR antagonists based on x-ray structural analysis and knowledge of TR activation functions and interacting partners, outlined earlier.⁶¹ The “extension hypothesis” proposes that the design of agonist-like compounds with appropriately placed bulky extensions will prevent H12 from folding into the active conformation, thereby blocking coactivator binding (Fig. 5-14). Several competitive TR antagonists have been produced using the principles of this hypothesis. Moreover, later crystallization of estrogen-receptor LBDs in complex with the known antagonists tamoxifen and raloxifene revealed that both of these ligands possess extensions that displace H12.

Further chemical analysis of TR ligands with extensions revealed surprises; ligands with large extensions often behave as agonists or partial agonists. Chemical analysis revealed that the position of the extension is important, that the extension requires an inflexible linker for a bridge to the hormone-like moiety, and that the chemical nature of the extension is also important. Thus, more needs to be learned about rules of TR antagonism. The fact that some compounds with bulky extensions act as agonists suggests that the TR must adapt to bind large ligands and form a partially active conformation. The structural basis for this surprising effect is discussed later with two compounds, T₄ and GC-24.

Other Influences of H12 Rearrangement

Repositioning of NR H12 may influence NR function in other ways. The possibility that H12 position regulates hormone binding is discussed in the section on hormone binding and release. In addition, NR H12 may also play roles in LBD-LBD interactions. Unliganded RXR LBDs form tetramers which consist of pairs of LBD dimers linked by a large interface that overlaps lower parts of H3 and H11. In addition, H12 of one LBD subunit docks in the H3-H5 region (part of the AF-2 surface) of an LBD in the neighboring pair, enhancing tetramer interactions. Ligand promotes dissociation of LBD tetramers to form dimers and monomers, and this is probably initiated by hormone-dependent repositioning of H12, an event which would weaken the tetramer interaction.

Ligand Stabilizes LBD Structure

Hormone effects on TR LBD are probably more extensive than simple repositioning of H12.74–77 NR LBDs are somewhat mobile and unstable without ligand or association with cofactors, forming a so-called molten globular organization; hormone promotes widespread stabilization of domain structure.

The overall rearrangements that stabilize the domain may be functionally important. For TRs, homodimer interactions are inhibited by hormone in vitro, whereas RXR-TR LBD interactions are not. This implies that the homodimer and heterodimer interaction surfaces, though overlapping, are not identical, and that hormone promotes rearrangements in regions of the LBD surface that are specific for homodimer contact. Targeted mutagenesis suggests that conserved surface salt-bridge clusters within the TR H7-H8 region and H11 play a role in this effect.⁷⁷ For other NRs, ligand binding leads to rearrangements in the same region of the LBD surface that expose a short peptide sequence that is a target for covalent attachment of SUMO peptides,⁵⁴ an important event in some mechanisms of ligand-dependent repression, as described earlier. Finally, T₃ promotes increased packing of H1 against the remaining LBD core (H2-H12 region). Remarkably, this can even be recapitulated in cis, using separate fragments for the H2-H12 and H1 region. Since H1 links the LBD to DBD, this event could conceivably communicate information about hormone binding to the DBD.

Mechanisms of Ligand Binding and Release

The ways ligand enters and leaves the buried TR hormone-binding pocket remain unknown.78,79 Hormone is completely enclosed within the domain, and entry/exit routes are not obvious from consideration of the structure alone. One possibility is that the major ligand entry and exit route lies under H12. In this model, referred to as the mousetrap model, H12 displacement in the absence of hormone opens the ligand entry route. Here, hormone-induced folding of H12 over the entry site traps ligand until H12 is displaced and the ligand is released. Interestingly, T₄ displays a very high dissociation rate from TR LBDs, likely explaining low affinity for receptor. A crystal structure of the TR-LBD with T₄ revealed that subtle rearrangements in side chains in the hormone-binding pocket create extra space to permit the receptor to accommodate the bulky 5′ iodine group, but that H12 packing is relatively inefficient. Inefficient H12 packing could increase T₄ dissociation because it opens an exit route from the LBD.

There are other possible sites for hormone entry and exit. The H1-H3 loop is highly mobile and disordered in crystal structures and was proposed to be a distinct ligand entry route in the report of the first TR crystal structure. Use of molecular dynamics computer simulations, where information from x-ray structures is used to simulate molecular motions in macromolecules, revealed three possible exit routes. (These were under H12, through the H1-H3 loop, and via opening of the H8-H11 region near the dimer surface.) Further analysis indicated that ligand exit through H1-H3 appears most energetically favorable in aqueous solution. Accordingly, mutations which affect the degree of disorder within the H1-H3 loop region enhance ligand dissociation rates. Thus TRs may harbor more than one ligand entry/exit route.

Conformational Flexibility in The Hormone-Binding Pocket

A major surprise derived from recent x-ray and functional studies of TRs is that the TRβ LBD can adapt to bind a ligand that is much larger than classical thyroid hormones and still fold into an active conformation.⁷⁶ A structure of TRβ with a synthetic agonist (GC-24) that contains a bulky 3′ phenyl extension reveals a different mechanism for adaptation from that described for T₄ (Fig. 5-15). Here, helices 3 and 11 part to form an extension to the pocket.⁷⁶ This phenomenon may be useful for drug design; GC-24 is highly TRβ selective, suggesting that it is possible to capitalize on TR isoform-specific differences in flexibility. The surprising implications of this finding are that TRs could bind and be activated by larger ligands than expected, although it is not clear whether natural ligands with appropriately placed extensions exist.

The reasons that the H3-H11 opens to extend the conventional hormone-binding pocket is not clear; one hypothesis is that the GC-24 extension is accommodated within a region of receptor that can open up to allow ligand exit under H12.⁷⁶ There may be other regions of TR open to extend the pocket, as described earlier; some compounds with bulky 5′ extensions different from the GC-24 3′ extension can also act as agonists.

TRα AND TRβ

TRα1 and TRβ1 are similar in primary sequence (Fig. 5-17), structure, and mechanism of action, but they are not completely identical in function.^2,86–88 Phenotypes of human patients with TRβ mutations that cause resistance to thyroid hormone syndrome (RTH) and mouse TR knockout models reveal that TRα and TRβ play distinct roles in T₃ effects on different responses, genes, and tissues. These differential actions are often ascribed to differences in the receptor tissue distribution,²⁰ but this is not proven. That TRα and TRβ may act differently to regulate some genes, which is also likely, also needs to be considered. In this regard, TRβ displays stronger activity at alternate inverted palindromic TREs than TRα, and this correlates with the extent of TR-TR homodimer binding to these elements. Moreover, TRs display different ligand preferences with TRβ binding to Triac, with about two- to threefold higher affinity than TRα. It is not clear whether these differences are functionally important, but the findings are suggestive of different mechanisms.

Alternate TR Splice Products

Both TR genes produce alternate protein products by combinations of differential exon splicing and promoter usage (see Fig. 5-17). The major differentially spliced THRA product is called TRα2. Here, the first 370 amino acids are identical to TRα, but an alternate C-terminus alters LBD sequences encoding H11 and H12, resulting in a TR that does not bind hormone. It is proposed that TRα2 can associate with wild-type TRs and block their actions in response to ligand (referred to as dominant inhibitor activity), especially in tissues where it is present in large amounts (e.g., brain, testis, kidney, and brown fat). TRα2 dampens T₃ responses in transfection assays, and knockout mice that lack TRα2 display heightened thyroid hormone sensitivity in vivo. The mechanism of this dominant negative effect is unclear; many models of dominant negative activity require functional homodimer or heterodimer formation, but TRα2 appears to display reduced dimer formation through the altered LBD H10-H11 region that is reflected in reduced DNA binding of TRα2 relative to wild-type TRα.

There are minor TRα variants. TRα3 lacks the first 39 amino acids found in the unique α2 region. Its function is obscure. Two further isoforms, TRΔα1 and TRΔα2, are expressed in the intestine. These contain portions of the TR C-terminus and may act as dominant inhibitors of TRα1. Interestingly, the antisense strand of the THRA locus encodes another NR, called rev-erbA (from reverse ERBA), which plays important roles in diurnal rhythm. For some species, rev-erbA has been demonstrated to bind heme and respond to differences in heme redox state. There is no evidence that rev-erbA has any functional relationship with TRs.

TRβ1 splice variants are created by differential splicing of the THRB gene NTD exon. The major alternate hormone-binding form is called TRβ2 and is created as a result of transcriptional initiation at an isoform-specific promoter and alternative exon splicing, which creates a completely unique isoform-specific NTD. TRβ2 is predominantly expressed in pituitary and hypothalamus, although low levels of expression are detectable in liver, and this form of TRβ2 is believed to play an important role in negative regulation of thyroid hormone production through the hypothalamic-pituitary axis. TRβ2 is also expressed in the cochlea and retinal photoreceptor cells and is needed for appropriate development of color vision.

The distinct N-terminus of TRβ2 possesses a unique transactivation function. At positively regulated genes and reporters, TRβ2 fails to suppress basal transcriptional activity in the absence of hormone and instead activates gene transcription in the absence of exogenous ligand. The TRβ2 NTD is thought to mediate the recruitment of coactivators independently of the LBD and may also exert antirepressive activities by blocking actions of corepressor gene-silencing functions.

There are other TRβ1 splice variants with obscure function. TRβ0 (present in reptiles and birds) lacks the NTD, and TRβ3 contains a unique short 24-amino-acid NTD. In addition, there is a truncated C-terminal TRβ receptor, TRΔβ3, which lacks part of the LBD and does not bind to T₃.

Resistance to Thyroid Hormone

Inherited human TR mutations cause resistance to thyroid hormone syndrome (RTH).⁸⁹ Symptoms of RTH, coupled with genetic analysis of TR mutations, yielded some of the first insights into differential actions of TRβ and TRα isoforms.

Patients with RTH exhibit reduced sensitivity to thyroid hormone.⁸⁹ In most cases, circulating thyroid hormone levels are elevated to variable extents, but TSH levels (which should be suppressed by elevated hormone levels) are elevated (sometimes resulting in goiter) or normal. This is caused by failure of feedback inhibition of hormone synthesis through the hypothalamic-pituitary axis due to the reduced sensitivity of TRβ to thyroid hormone (see Fig. 5-1). Overall, the elevated thyroid hormone levels overcome the resistance and produce a mostly euthyroid state. (In some cases, a variant of the syndrome results in central but not peripheral resistance to thyroid hormone, where TSH levels are high, and there can be symptoms of frank hyperthyroidism.) However, there can also be a mix of hypo- and hyperthyroid phenotypes of variable clinical penetrance. Commonly there is tachycardia (hyperthyroid symptom) and attention deficit. To a lesser extent, there can be some mild mental retardation, skeletal malformations, and hearing abnormalities (hypothyroid manifestations). Most (>80%) RTH is caused by mutations in the LBD coding regions of the TRβ gene. Mutations in TRα have never been detected. The fact that there is elevated heart rate therefore suggested that normal TRα is responsible for this hyperthyroid effect and provided the first evidence for TR isoform–specific activities, later confirmed by gene knockout studies.

RTH can also be caused by mutations in other genes involved that play roles in thyroid hormone response. Cases have been described in which the target genes identified involve a thyroid hormone membrane transporter and enzymes required for transfer of an essential cofactor, selenium, to the active site of deiodinases.⁹⁰

Analysis of TR Isoform Action in Mice

Functions of TR isoforms have been investigated in mouse models by targeted deletion of TR genes and knock-in of TR alleles with dominant negative activities.18,88,91,92

TR Knockouts

Several mouse TR knockout strains are available. For TRα, mice with deletions that eliminate all TRα isoforms (TRα^0/0), TRα1, and TRα2 (TRα^−/−; this strain preserves expression of minor TRα isoforms), TRα1^−/− and TRα2^−/− are reported. For TRβ, strains that lack all TRβ isoforms (TRβ^−/−) or TRβ2 (TRβ2^−/−) are available. Finally, strains that lack both TR isoforms have been created by crossing of TRα^0/0 and TRβ^−/− strains (for example, TRα^0/0β^−/−). All TR knockout mice are viable and fertile. There are some complications in interpretation of the phenotypes of the mice. Targeted deletions often affect more than one TR splice variant, and in some cases, deletion of one TR isoform leads to compensatory biological responses in TR isoform expression. Additionally, different mouse genetic backgrounds influence the presentation of the phenotype. In spite of these qualifications, several clear lessons about TR activity and TR isoform-specific actions can be drawn from these studies:

Knock-INS

TR function has been probed with targeted (knock-in) mutagenesis. These experiments were first performed in order to create mouse models of RTH. These experiments were successful; a PV knock-in mutant (a highly dominant negative TR with altered H12 sequence) displayed several features of RTH, including elevated thyroid hormone with normal TSH, goiter, short stature, and tachycardia.

Mice strains that express analogous TRα mutants have also been created. The goal of these studies was to probe natural function of TRα with dominant negative versions of the protein. However, mouse strains bearing different mutations exhibit a range of phenotypes from lean and hypermetabolic to obese and insulin resistant. This suggests that TRα plays an important role in metabolic rate, but that different mutations affect these processes in different ways.

RTH

Analysis of RTH patient families revealed an autosomal dominant inheritance pattern which was explained by analysis of the effects of TRβ mutations on receptor function. RTH mutant TRs continue to bind corepressor at hormone concentrations in which it would normally be released. Projection of positions of RTH mutations onto the x-ray structural models of TR LBDs reveals that many map to the ligand-binding pocket, explaining why the mutations decrease T₃-binding affinity. Other mutations uncouple hormone binding from corepressor release. These tend to affect residues close to H12. The net result of increased corepressor binding is dominant negative activity; the mutant TR inhibits activity of the wild-type counterpart encoded by the other allele. This dominant negative activity explains the autosomal dominance of the inheritance pattern.

It is clear that dominant negative activity requires integrity of the dimer/heterodimer surface. Nevertheless, the molecular basis of this effect is not totally understood. Most TRs form heterodimers with RXRs, so there may be competition between heterodimers of wild-type RXR/mutant TRs and wild-type RXR/TRs for DNA binding sites. Alternatively, direct homodimer interactions between wild-type and mutant TRs may explain these effects.

A small number of RTH mutations in the LBD affect residues that appear far from the functional elements of this domain. These were studied to gain insights into coupling of hormone binding to response. Most mutations affect residues that are essential for TR LBD stability. Structural analysis reveals that these residues often participate in polar side-chain interactions that link helices. The mutations appear to destabilize the H1-H3 loop, a possible ligand escape route, and reduce ligand affinity by increasing dissociation rates.

While most patients exhibit generalized resistance that is manifested in the pituitary and periphery, others display so-called pituitary resistance—normal peripheral responses with a specific failure of feedback inhibition. The molecular basis for differences between pituitary and generalized resistance is not known. One possibility is that there are differential effects of mutations on activities of TRβ1 (widely expressed) versus TRβ2 (limited distribution, including the pituitary) whereby the TRβ1 is normal or near normal, and the TRβ2 is impaired, leading to decreased suppression of TSH release. Alternatively, since TR actions in the hypothalamic-pituitary axis often involve transcriptional repression, it is possible that pituitary resistance mutations preferentially affect ligand-dependent TR repression mechanisms.

Cancer

Cancer genome sequencing analysis has revealed that TR mutations occur frequently in cancer (including liver, kidney, thyroid gland, and breast).⁹³ Presently, the significance of these mutations in cancer etiology is not completely clear, but it is well established that mutated TRs can cause cancer; as described earlier, v-erbA is a non-hormone-binding form of TRα that actively represses transcription and causes erythroblastoma.⁵¹ Moreover, mouse models which contain a targeted mutation of TRβ (knock-in) that resembles the human PV RTH mutant develop thyroid cancer at increased rates.

TR mutants that emerge in cancer often resemble RTH mutants; the mutations preserve repression at the expense of hormone binding and transcriptional activation. This led to the suggestion that localized RTH may contribute to some forms of cancer. However, other TR cancer mutants appear to lack dominant negative activity, such as those that disrupt DNA binding, presumably eliminating TR activity at target genes. There is also a correlation in other cases with loss of nuclear localization; unlike TRα, which is predominantly nuclear, v-erbA displays nuclear and cytoplasmic localization. Moreover, TRα partly relocalizes to the cytoplasm in cancers. The connections between TR structure-function and oncogenesis require further investigation.

1. Sawin, CT. The heritage of the thyroid: A brief history. In: Braverman LE, Utiger RD, Ingbar SH, Werner SC, eds. Werner and Ingbar’s The Thyroid: A Fundamental and Clinical Text. ed 9. Philadelphia: Lippincott Williams & Wilkins; 2000:3–7.

2. Yen, PM. Physiological and molecular basis of thyroid hormone action. Physiol Rev. 2001;81:1097–1142.

3. Zhang, J, Lazar, MA. The mechanism of action of thyroid hormones. Annu Rev Physiol. 2000;62:439–466.

4. Laudet, V, Gronemeyer, H. The Nuclear Receptor Facts Book. London: Academic Press; 2002.

5. Flamant, F, et al. International Union of Pharmacology. LIX. The pharmacology and classification of the nuclear receptor superfamily: thyroid hormone receptors. Pharmacol Rev. 2006;58:705–711.

6. Leonard, JL, Koehrle, J. Intracellular pathways of iodothyronine metabolism. In: Braverman LE, Utiger RD, Ingbar SH, et al, eds. Werner and Ingbar’s the thyroid: a fundamental and clinical text. ed 9. Philadelphia: Lippincott Williams & Wilkins; 2000:136–173.

7. Panicker, V, et al. A common variation in deiodinase 1 gene DIO1 is associated with the relative levels of free thyroxine and triiodothyronine. J Clin Endocrinol Metab. 2008;93:3075–3081.

8. Panicker, V, et al. Genetic loci linked to pituitary-thyroid axis set points: a genome-wide scan of a large twin cohort. J Clin Endocrinol Metab. 2008;93:3519–3523.

9. Sorensen, HG, et al. Identification and consequences of polymorphisms in the thyroid hormone receptor alpha and beta genes. Thyroid. 2008;18:1087–1094.

10. Scanlan, TS, et al. 3-Iodothyronamine is an endogenous and rapid-acting derivative of thyroid hormone. Nat Med. 2004;10:638–642.

11. Visser, WE, Friesema, EC, Jansen, J, et al. Thyroid hormone transport in and out of cells. Trends Endocrinol Metab. 2008;19:50–56.

12. Friesema, EC, et al. Effective cellular uptake and efflux of thyroid hormone by human monocarboxylate transporter 10. Mol Endocrinol. 2008;22:1357–1369.

13. Dumitrescu, AM, Refetoff, S. Novel biological and clinical aspects of thyroid hormone metabolism. Endocr Dev. 2007;10:127–139.

14. Suzuki, S, Mori, J, Hashizume, K. mu-crystallin, a NADPH-dependent T(3)-binding protein in cytosol. Trends Endocrinol Metab. 2007;18:286–289.

15. Berkenstam, A, et al. The thyroid hormone mimetic compound KB2115 lowers plasma LDL cholesterol and stimulates bile acid synthesis without cardiac effects in humans. Proc Natl Acad Sci U S A. 2008;105:663–667.

16. Gereben, B, et al. Cellular and Molecular Basis of Deiodinase-Regulated Thyroid Hormone Signaling. Endocr Rev. 2008.

17. Huang, SA, Bianco, AC. Reawakened interest in type III iodothyronine deiodinase in critical illness and injury. Nat Clin Pract Endocrinol Metab. 2008;4:148–155.

18. Moreno, M, et al. Metabolic effects of thyroid hormone derivatives. Thyroid. 2008;18:239–253.

19. Sherman, SI, et al. Augmented hepatic and skeletal thyromimetic effects of tiratricol in comparison with levothyroxine. J Clin Endocrinol Metab. 1997;82:2153–2158.

20. Bookout, AL, et al. Anatomical profiling of nuclear receptor expression reveals a hierarchical transcriptional network. Cell. 2006;126:789–799.

21. Silva, JE. Thermogenic mechanisms and their hormonal regulation. Physiol Rev. 2006;86:435–464.

22. Silva, JE, Bianco, SD. Thyroid-adrenergic interactions: physiological and clinical implications. Thyroid. 2008;18:157–165.

23. Videla, LA, Fernandez, V, Tapia, G, et al. Thyroid hormone calorigenesis and mitochondrial redox signaling: upregulation of gene expression. Front Biosci. 2007;12:1220–1228.

24. Kim, B. Thyroid hormone as a determinant of energy expenditure and the basal metabolic rate. Thyroid. 2008;18:141–144.

25. Feng, X, Jiang, Y, Meltzer, P, et al. Thyroid hormone regulation of hepatic genes in vivo detected by complementary DNA microarray. Mol Endocrinol. 2000;14:947–955.

26. Weitzel, JM, et al. Hepatic gene expression patterns in thyroid hormone-treated hypothyroid rats. J Mol Endocrinol. 2003;31:291–303.

27. Obregon, MJ. Thyroid hormone and adipocyte differentiation. Thyroid. 2008;18:185–195.

28. Klein, I, Danzi, S. Thyroid disease and the heart. Circulation. 2007;116:1725–1735.

29. Wiersinga, WM. The role of thyroid hormone nuclear receptors in the heart: evidence from pharmacological approaches. Heart Fail Rev. 2008 Dec 19.

30. Crunkhorn, S, Patti, ME. Links between thyroid hormone action, oxidative metabolism, and diabetes risk? Thyroid. 2008;18:227–237.

31. Bassett, JH, Williams, GR. Critical role of the hypothalamic-pituitary-thyroid axis in bone. Bone. 2008;43:418–426.

32. Samuels, MH. Cognitive function in untreated hypothyroidism and hyperthyroidism. Curr Opin Endocrinol Diabetes Obes. 2008;15:429–433.

33. Williams, GR. Neurodevelopmental and neurophysiological actions of thyroid hormone. J Neuroendocrinol. 2008;20:784–794.

34. Plateroti, M, Kress, E, Mori, JI, et al. Thyroid hormone receptor alpha1 directly controls transcription of the beta-catenin gene in intestinal epithelial cells. Mol Cell Biol. 2006;26:3204–3214.

35. Doshi, DN, Blyumin, ML, Kimball, AB. Cutaneous manifestations of thyroid disease. Clin Dermatol. 2008;26:283–287.

36. Baxter, JD, et al. Thyroid hormone receptors and responses. Recent Prog Horm Res. 1979;35:97–153.

37. Ivarie, RD, Baxter, JD, Morris, JA. Interaction of thyroid and glucocorticoid hormones in rat pituitary tumor cells. Specificity and diversity of the responses analyzed by two-dimensional gel electrophoresis. J Biol Chem. 1981;256:4520–4528.

38. Yen, PM, Feng, X, Flamant, F, et al. Effects of ligand and thyroid hormone receptor isoforms on hepatic gene expression profiles of thyroid hormone receptor knockout mice. EMBO Rep. 2003;4:581–587.

39. Astapova, I, et al. The nuclear corepressor, NCoR, regulates thyroid hormone action in vivo. Proc Natl Acad Sci U S A. 2008;105:19544–19549.

40. van Rooij, E, et al. Control of stress-dependent cardiac growth and gene expression by a microRNA. Science. 2007;316:575–579.

41. Tata, JR, et al. The action of thyroid hormones at the cell level. Biochem J. 1963;86:408–428.

42. Oppenheimer, JH, Schwartz, HL, Surks, MI. Tissue differences in the concentration of triiodothyronine nuclear binding sites in the rat: liver, kidney, pituitary, heart, brain, spleen, and testis. Endocrinology. 1974;95:897–903.

43. Surks, MI, Koerner, DH, Oppenheimer, JH. In vitro binding of L-triiodothyronine to receptors in rat liver nuclei. Kinetics of binding, extraction properties, and lack of requirement for cytosol proteins. J Clin Invest. 1975;55:50–60.

44. Nelson, C, et al. Discrete cis-active genomic sequences dictate the pituitary cell type-specific expression of rat prolactin and growth hormone genes. Nature. 1986;322:557–562.

45. MacLeod, KN, Baxter, JD. DNA binding of thyroid hormone receptors. Biochem Biophys Res Commun. 1975;62:577–583.

46. Spindler, BJ, MacLeod, KM, Ring, J, et al. Thyroid hormone receptors. Binding characteristics and lack of hormonal dependency for nuclear localization. J Biol Chem. 1975;250:4113–4119.

47. Latham, KR, Ring, JC, Baxter, JD. Solubilized nuclear “receptors” for thyroid hormones. Physical characteristics and binding properties, evidence for multiple forms. J Biol Chem. 1976;251:7388–7397.

48. MacLeod, KM, Baxter, JD. Chromatin receptors for thyroid hormones. Interactions of the solubilized proteins with DNA. J Biol Chem. 1976;251:7380–7387.

49. Sap, J, et al. The c-erb-A protein is a high-affinity receptor for thyroid hormone. Nature. 1986;324:635–640.

50. Weinberger, C, et al. The c-erb-A gene encodes a thyroid hormone receptor. Nature. 1986;324:641–646.

51. Privalsky, ML. v-erb A, nuclear hormone receptors, and oncogenesis. Biochim Biophys Acta. 1992;1114:51–62.

52. Glass, CK, Rosenfeld, MG. The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev. 2000;14:121–141.

53. Lazar, MA. Nuclear receptor corepressors. Nucl Recept Signal. 2003;1:e001.

54. Ricote, M, Glass, CK. PPARs and molecular mechanisms of transrepression. Biochim Biophys Acta. 2007;1771:926–935.

55. Shibusawa, N, Hollenberg, AN, Wondisford, FE. Thyroid hormone receptor DNA binding is required for both positive and negative gene regulation. J Biol Chem. 2003;278:732–738.

56. Rastinejad, F, Perlmann, T, Evans, RM, et al. Structural determinants of nuclear receptor assembly on DNA direct repeats. Nature. 1995;375:203–211.

57. Khorasanizadeh, S, Rastinejad, F. Nuclear-receptor interactions on DNA-response elements. Trends Biochem Sci. 2001;26:384–390.

58. Desvergne, B. How do thyroid hormone receptors bind to structurally diverse response elements? Mol Cell Endocrinol. 1994;100:125–131.

59. Wagner, RL, et al. A structural role for hormone in the thyroid hormone receptor. Nature. 1995;378:690–697.

60. Wagner, RL, et al. Hormone selectivity in thyroid hormone receptors. Mol Endocrinol. 2001;15:398–410.

61. Webb, P, et al. Design of thyroid hormone receptor antagonists from first principles. J Steroid Biochem Mol Biol. 2002;83:59–73.

62. Weatherman, RV, Fletterick, RJ, Scanlan, TS. Nuclear-receptor ligands and ligand-binding domains. Annu Rev Biochem. 1999;68:559–581.

63. Nascimento, AS, et al. Structural rearrangements in the thyroid hormone receptor hinge domain and their putative role in the receptor function. J Mol Biol. 2006;360:586–598.

64. Kumar, R, Thompson, EB. Transactivation functions of the N-terminal domains of nuclear hormone receptors: protein folding and coactivator interactions. Mol Endocrinol. 2003;17:1–10.

65. Figueira, AC, et al. Low-resolution structures of thyroid hormone receptor dimers and tetramers in solution. Biochemistry. 2007;46:1273–1283.

66. Chandra, V, et al. Structure of the intact PPAR-gamma-RXR-nuclear receptor complex on DNA. Nature. 2008;456:350–356.

67. Ribeiro, RC, et al. X-ray crystallographic and functional studies of thyroid hormone receptor. J Steroid Biochem Mol Biol. 1998;65:133–141.

68. Ribeiro, RC, et al. Definition of the surface in the thyroid hormone receptor ligand binding domain for association as homodimers and heterodimers with retinoid X receptor. J Biol Chem. 2001;276:14987–14995.

69. Velasco, LF, et al. Thyroid hormone response element organization dictates the composition of active receptor. J Biol Chem. 2007;282:12458–12466.

70. Wu, Y, Xu, B, Koenig, RJ. Thyroid hormone response element sequence and the recruitment of retinoid X receptors for thyroid hormone responsiveness. J Biol Chem. 2001;276:3929–3936.

71. Diallo, EM, Wilhelm, KG, Jr., Thompson, DL, et al. Variable RXR requirements for thyroid hormone responsiveness of endogenous genes. Mol Cell Endocrinol. 2007;264:149–156.

72. Mengeling, BJ, Lee, S, Privalsky, ML. Coactivator recruitment is enhanced by thyroid hormone receptor trimers. Mol Cell Endocrinol. 2008;280:47–62.

73. Kallenberger, BC, Love, JD, Chatterjee, VK, et al. A dynamic mechanism of nuclear receptor activation and its perturbation in a human disease. Nat Struct Biol. 2003;10:136–140.

74. Pissios, P, Tzameli, I, Kushner, P, et al. Dynamic stabilization of nuclear receptor ligand binding domains by hormone or corepressor binding. Mol Cell. 2000;6:245–253.

75. Pissios, P, Tzameli, I, Moore, DD. New insights into receptor ligand binding domains from a novel assembly assay. J Steroid Biochem Mol Biol. 2001;76:3–7.

76. Togashi, M, et al. Conformational adaptation of nuclear receptor ligand binding domains to agonists: potential for novel approaches to ligand design. J Steroid Biochem Mol Biol. 2005;93:127–137.

77. Togashi, M, Nguyen, P, Fletterick, R, et al. Rearrangements in thyroid hormone receptor charge clusters that stabilize bound 3,5′,5-triiodo-L-thyronine and inhibit homodimer formation. J Biol Chem. 2005;280:25665–25673.

78. Martinez, L, et al. Molecular dynamics simulations reveal multiple pathways of ligand dissociation from thyroid hormone receptors. Biophys J. 2005;89:2011–2023.

79. Martinez, L, Webb, P, Polikarpov, I, et al. Molecular dynamics simulations of ligand dissociation from thyroid hormone receptors: evidence of the likeliest escape pathway and its implications for the design of novel ligands. J Med Chem. 2006;49:23–26.

80. Moore, JM, et al. Quantitative proteomics of the thyroid hormone receptor-coregulator interactions. J Biol Chem. 2004;279:27584–27590.

81. Webb, P, et al. The nuclear receptor corepressor (N-CoR) contains three isoleucine motifs (I/LXXII) that serve as receptor interaction domains (IDs). Mol Endocrinol. 2000;14:1976–1985.

82. Lonard, DM, O’Malley, BW. Nuclear receptor coregulators: judges, juries, and executioners of cellular regulation. Mol Cell. 2007;27:691–700.

83. Privalsky, ML. The role of corepressors in transcriptional regulation by nuclear hormone receptors. Annu Rev Physiol. 2004;66:315–360.

84. White, R, et al. Role of RIP140 in metabolic tissues: connections to disease. FEBS Lett. 2008;582:39–45.

85. Meng, X, et al. E1A and a nuclear receptor corepressor splice variant (N-CoRI) are thyroid hormone receptor coactivators that bind in the corepressor mode. Proc Natl Acad Sci U S A. 2005;102:6267–6272.

86. Lazar, MA. Thyroid hormone receptors: multiple forms, multiple possibilities. Endocr Rev. 1993;14:184–193.

87. O’Shea, PJ, Williams, GR. Insight into the physiological actions of thyroid hormone receptors from genetically modified mice. J Endocrinol. 2002;175:553–570.

88. Forrest, D, Vennstrom, B. Functions of thyroid hormone receptors in mice. Thyroid. 2000;10:41–52.

89. Yen, PM. Molecular basis of resistance to thyroid hormone. Trends Endocrinol Metab. 2003;14:327–333.

90. Refetoff, S, Dumitrescu, AM. Syndromes of reduced sensitivity to thyroid hormone: genetic defects in hormone receptors, cell transporters and deiodination. Best Pract Res Clin Endocrinol Metab. 2007;21:277–305.

91. Liu, YY, et al. A mutant thyroid hormone receptor alpha antagonizes peroxisome proliferator-activated receptor alpha signaling in vivo and impairs fatty acid oxidation. Endocrinology. 2007;148:1206–1217.

92. Sjogren, M, et al. Hypermetabolism in mice caused by the central action of an unliganded thyroid hormone receptor alpha1. EMBO J. 2007;26:4535–4545.

93. Yen, PM, Cheng, SY. Germline and somatic thyroid hormone receptor mutations in man. J Endocrinol Invest. 2003;26:780–787.

94. Davis, PJ, Leonard, JL, Davis, FB. Mechanisms of nongenomic actions of thyroid hormone. Front Neuroendocrinol. 2008;29:211–218.

95. Furuya, F, Lu, C, Guigon, CJ, et al. Nongenomic activation of phosphatidylinositol 3-kinase signaling by thyroid hormone receptors. Steroids. 2009;74:628–634.

96. Webb, P. Selective activators of thyroid hormone receptors. Expert Opin Investig Drugs. 2004;13:489–500.

97. Biondi, B, Cooper, DS. The clinical significance of subclinical thyroid dysfunction. Endocr Rev. 2008;29:76–131.

98. The coronary drug project. Findings leading to further modifications of its protocol with respect to dextrothyroxine. The coronary drug project research group. JAMA. 1972;220:996–1008.

99. Baxter, JD, et al. Selective modulation of thyroid hormone receptor action. J Steroid Biochem Mol Biol. 2001;76:31–42.

100. Baxter, JD, Webb, P, Grover, G, et al. Selective activation of thyroid hormone signaling pathways by GC-1: a new approach to controlling cholesterol and body weight. Trends Endocrinol Metab. 2004;15:154–157.

101. Scanlan, TS. Sobetirome: a case history of bench-to-clinic drug discovery and development. Heart Fail Rev. 2008 Nov 11.