Thyroid Hormone Resistance Syndrome Manifests as an Aberrant Interaction between Mutant T₃ Receptors and Transcriptional Corepressors

Abstract

Nuclear hormone receptors are hormone-regulated transcription factors that play critical roles in chordate development and homeostasis. Aberrant nuclear hormone receptors have been implicated as causal agents in a number of endocrine and neoplastic diseases. The syndrome of Resistance to Thyroid Hormone (RTH) is a human genetic disease characterized by an impaired physiological response to thyroid hormone. RTH is associated with diverse mutations in the thyroid hormone receptor β-gene. The resulting mutant receptors function as dominant negatives, interfering with the actions of normal thyroid hormone receptors coexpressed in the same cells. We report here that RTH receptors interact aberrantly with a newly recognized family of transcriptional corepressors variously denoted as nuclear receptor corepressor (N-CoR), retinoid X receptor interacting protein-13 (RIP-13), silencing mediator for retinoid and thyroid hormone receptors (SMRT), and thyroid hormone receptor-associating cofactor (TRAC). All RTH receptors tested exhibit an impaired ability to dissociate from corepressors in the presence of thyroid hormone. Two of the RTH mutations uncouple corepressor dissociation from hormone binding; two additional RTH mutants exhibit an unusually strong interaction with corepressor under all hormone conditions tested. Finally, artificial mutants that abolish corepressor binding abrogate the dominant negative activity of RTH mutants. We suggest that an altered corepressor interaction is likely to play a critical role in the dominant negative potency of RTH mutants and may contribute to the variable phenotype in this disorder.

INTRODUCTION

Nuclear hormone receptors are hormone-regulated transcription factors and include the thyroid hormone receptors (encoded by two different loci, T₃Rα and T₃Rβ), steroid receptors, and retinoid receptors (reviewed in Refs. ^1–6). All nuclear hormone receptors contain a variety of motifs involved in DNA binding, hormone binding, receptor dimerization, and interactions with the transcriptional machinery (Fig. 1). On binding to specific target DNA sequences, nuclear hormone receptors modulate the expression of adjacent target genes (1–6). The nature of the transcriptional response (activation or repression) is dictated by cell type, promoter context, and hormone status (1–10). In most contexts T₃Rs are transcriptional repressors in the absence of cognate hormone (T₃) and activators in its presence (2, 7, 8, 11–15).

Fig. 1. Schematic Representation of the Human T₃Rβ Protein and the Location of the RTH Mutations.

The T₃Rβ is presented schematically from the N (n) terminus to C (c) terminus, with a central zinc-finger domain (Zn) indicated. The locations of the RTH mutations analyzed in this manuscript are indicated below. The regions of the wt receptor implicated in DNA recognition, hormone binding, receptor dimerizaton, transcriptional (Tx) activation, and SMRT corepressor binding are depicted above. X indicates the location of the site-directed mutations introduced to disrupt SMRT association (T₃R^sad mutations; see text).

Some aspects of this transcriptional regulation are probably mediated by direct interactions between nuclear hormone receptors and the general transcriptional machinery (5). However, a series of specific ancillary proteins (denoted coactivators and corepressors) also appear to play critical roles in transcriptional control. For example, the transcriptional repression exhibited by T₃Rs in the absence of hormone appears to require an association of the receptor with a family of corepressor proteins (variously denoted TRACs, SMRT, RIP13, or N-CoR) (11–18). T₃R mutants unable to associate with these corepressors are unable to repress (11–15). Addition of T₃ hormone converts T₃Rs into transcriptional activators, a process that correlates with dissociation of the corepressors from the receptor and a corresponding recruitment of a novel set of proteins believed to function as coactivators (11–15, 19–25).

A variety of neoplastic and endocrine disorders are the consequence of aberrations in nuclear hormone receptor function (26–31). The syndrome of Resistance to Thyroid Hormone (RTH) is an autosomal dominant human endocrine disease (reviewed in Refs. ^32–34). Individuals with RTH exhibit a failure to respond to elevated circulating thyroid hormone. This disorder is associated with diverse mutations at the T₃Rβ locus, resulting in the synthesis of abnormal receptors that are impaired in hormone binding and in transcriptional activation (35–39). As a result, the RTH-T₃Rs appear to function as dominant negatives, and interfere with the actions of the normal T₃Rs synthesized from the unaffected T₃Rα- and β-alleles (40–50). More than 30 different T₃Rβ mutations have been associated with the RTH syndrome.

Nonetheless, the precise mechanism(s) by which RTH-T₃Rs act as dominant negatives remains uncertain. In addition, a given RTH-T₃R mutation can have very different phenotypic consequences in different individuals or in different tissues in the same individual, suggesting a potential involvement of other factors modulating dominant negative function and the degree of resistance (32, 48, 51–54). We report here evidence that the RTH syndrome is associated with an aberrant interaction between the RTH-T₃Rs and the SMRT/TRAC corepressors and that this corepressor interaction is important in the ability of RTH-T₃Rs to act as dominant negative inhibitors. Furthermore, different RTH-T₃Rs exhibit distinct corepressor interactions. Conceivably, differential interactions with, expression expression of, or genetic polymorphisms within the corepressors may contribute to the highly variable presentation observed for RTH disease.

RESULTS

RTH-T₃Rβ Mutants Exhibit Aberrant Interactions with SMRT Corepressor

We wished to investigate whether corepressors play a role in the dominant negative effect that is postulated to be the molecular basis of RTH. We first tested the ability of wild type (wt) T₃Rβ and a panel of RTH-T₃Rβ mutants to associate with corepressor in vitro. We employed a glutathione-S-transferase (GST) fusion of SMRT, and ³⁵S-radiolabeled T₃Rs obtained by in vitro transcription and translation. All of our T₃Rβ alleles (wt or mutant) produced three main forms of translation product in vitro: a full-length molecule at approximately 55 kDa (p55) and two truncated forms at approximately 48 and 33 kDa (p48, p33) (Fig. 2). The truncated T₃R species are likely to the products of translational initiation on internal ATG codons and have been reported previously for both in vitro and in vivo expression systems (26, 27, 55).

Fig. 2. Binding of wt and RTH Mutant Receptors to SMRT Corepressor.

Wild type and four different RTH mutant T₃Rβs were tested for the ability to bind to an immobilized GST-SMRT fusion protein or to a nonrecombinant GST construct employed as a negative control, as indicated above each figure. Radiolabeled T₃Rs was obtained by in vitro translation and were incubated with the immobilized GST or GST-SMRT under the hormone conditions indicated (expressed as nm T₃ in the incubation buffer). The GST or GST-TRAC agarose was then extensively washed, and the radiolabeled receptor remaining bound to the matrix was eluted and was analyzed by SDS-PAGE and autoradiography. Input lanes display the in vitro translation products used for the binding experiments; all receptors were expressed at virtually identical levels and represent equivalent specific activities.

Wild-type T₃Rβ bound to the GST-SMRT construct in the absence of hormone, but not to nonrecombinant GST employed as a negative control (Fig. 2). Addition of increasing amounts of T₃ hormone strongly inhibited binding of the wtT₃Rβ to GST-SMRT, with a 50% inhibition observed at 56 nm hormone (Fig. 2, and quantified in Fig. 3 and Table 1). This interaction of wtT₃Rβ with SMRT closely resembles that observed previously for wtT₃Rα (11, 14, 15). Note that all three receptor translation products (p55, p48, and p33) were bound by the GST-SMRT construct and were released congruently by hormone, indicating that the use of alternative initiation sites did not alter the interaction of T₃Rβ with corepressor (Fig. 2).

Fig. 3. Quantification of wt and RTH-T₃Rβ Binding to SMRT.

The electrophoretograms shown in Fig. 3 were quantified by PhosphoImager (Molecular Dynamics, Sunnyvale, CA) analysis. The amount of T₃R bound to GST (squares) or to GST-SMRT (circles) at a given hormone concentration is expressed as a percentage of the total radiolabeled T₃R used in the binding reaction. Data for two different RTH-mutants, G345S (open symbols) and P453A (closed symbols), are shown in the third panel. Note the change in scale of the ordinate for the Δ432G mutant.

Table 1.

Clinical Phenotype, T₃ Binding Affinity, and SMRT Association Properties of wt and Mutant T₃Rβs

T3Rβ Allele	Index Clinical Phenotypea	Ka for T3b	[T3] for 50% Inhibition of SMRT Associationc
Wild type	Normal	2.2	56 ± 14
V264D	GRTH	<0.02	>1000
R320L	GRTH	0.21	245 ± 13
M334R	PRTH	0.11	690 ± 22
R338W	PRTH	0.21	290 ± 144
G344E	GRTH	N.D.	>1000
G345S	GRTH	<0.02	>1000
Δ430M	PRTH	N.D.	>1000d
Δ432G	PRTH	N.D.	>1000d
R438C	GRTH	0.67	103 ± 17
P453A	PRTH	0.38	>1000
P453H	GRTH	0.51	>1000

^aClinical phenotype refers to the index case from each kindred, with PRTH representing a diagnosis of pituitary resistance to thyroid hormone, and GRTH representing a diagnosis of generalized resistance to thyroid hormone (48). Note, however, that certain mutants (e.g. R320L and R338W) have been independently isolated from both PRTH and GRTH patients.

^bApparent association constants for T₃ (expressed as × 10¹⁰ m⁻¹) were previously summarized in Ref. 48 The G345S mutant bound T₃ but at levels too low to quantify, whereas no binding was detected (N.D.) for the G344E, Δ430M, and Δ432G mutants.

^cThe amount of hormone (nm) required to inhibit SMRT association by 50% was determined for each mutant using binding assays as in Fig. 3. Each value represents the average and range of at least two different determinations.

^dMutants that exhibited highly elevated levels of SMRT association relative to wt at all hormone concentrations tested.

The RTH-T₃Rβ mutants also associated with GST-SMRT in the absence of hormone, but in contrast to wtT₃Rβ, all the RTH-T₃Rβs tested exhibited aberrations in their ability to dissociate on T₃ treatment. The results for representative mutants are presented in detail in Fig. 2 and Fig. 3 and are summarized, along with data for additional mutants, in Table 1. Nine of the 11 RTH mutants analyzed here associated normally with SMRT in the absence of hormone (i.e. at levels comparable to wtT₃R). Four RTH mutants in this group of nine (R320L, M334R, R338W, and R438C) were eventually displaced from SMRT by T₃, but only at hormone levels significantly greater than those required for the wt receptor/SMRT dissociation. For the remaining five of these nine mutants (V264D, G344E, G345S, P453A, and P453H), little or no receptor/SMRT dissociation (less than 50%) was observed at up to 1 µM T₃; notably the P453A and p453H mutants fail to dissociate from SMRT even though these mutant receptors are able to bind thyroid hormone (see Discussion).

The last two of the 11 RTH mutants analyzed (Δ430 M and Δ432G) were not only impaired in their dissociation from SMRT in the presence of T₃, but also exhibited a dramatically enhanced SMRT association at all hormone concentrations (Fig. 2 and Fig. 3 and Table 1). These mutants bound to GST-SMRT in the absence or presence of T₃ at levels some 8- to 10-fold greater than that seen with wtT₃Rβ or the other RTH mutants (note the change in ordinate in Fig. 3); equal inputs of receptor were employed in all experiments. All mutant and wt receptors were transcribed and translated at equal efficiencies, utilizing identical conditions, and therefore represent equal specific activities. Additionally, the enhanced binding observed for mutants Δ430 M and Δ432G was consistently seen in multiple experiments utilizing different preparations of receptor and of GST-SMRT. Intriguingly, both of these mutants represent single amino acid deletions mapping to the same α-helix of the receptor and present clinically in a similar fashion (see Discussion).

We conclude that a common hallmark of the RTH receptors analyzed here is an aberrant interaction with corepressor. The phenotypes of these RTH mutants could be divided into three general categories: 1) normal levels of SMRT association, but requiring higher than normal levels of T₃ for dissociation, 2) normal levels of SMRT association, but with little or no T₃- mediated dissociation observed under the conditions employed, and 3) a dramatically elevated level of SMRT association at all hormone concentrations tested.

Altered Corepressor Association Correlates with the Dominant Negative Properties of the RTH Mutants

We next tested the transcriptional properties of representative RTH mutants from the categories defined above, using transient transfections of CV-1 cells. We initially introduced the various receptor mutants individually (Fig. 4A). In the absence of hormone, all five receptors tested functioned as repressors, inhibiting reporter gene expression some 50% relative to that seen in the absence of exogenous T₃R. As expected, addition of 100 nm T₃ converted the wtT₃Rβ from a repressor into a strong transcriptional activator (Fig. 4A). In contrast, the R320L mutant exhibited an impaired ability to activate reporter gene expression in the presence of 100 nm T₃, and the G345S, Δ432G, and P453A mutants functioned as constitutive repressors unable to induce significant activation of the reporter in response to T₃ (Fig. 4A).

Fig. 4. Transcriptional Properties of wt and RTH-T₃Rβs.

A, Transcriptional properties of the receptors when analyzed individually. CV-1 cells were transfected by lipofection with a DR4 luciferase reporter together with a pSG5 plasmid expressing either wt or mutant T₃Rβ, as indicated (none indicates an empty pSG5 vector). The cells were incubated in the absence (hatched bars) or presence (filled bars) of 100 nm T₃, and the cells were harvested. The resulting luciferase activity was determined and normalized relative to the activity of a pCH110-lac Z plasmid employed as an internal control. Data represent the average and range of duplicate experiments. B, Dominant negative activities of the different RTH-T₃Rβ mutants when cointroduced with wt receptor. CV-1 cells were transfected by lipofection with the DR4 luciferase reporter, together with different combinations of wt and RTH mutant receptors, as indicated below the panel. Either a 1:1 or 5:1 ratio of mutant to wt receptor were used, with the wt receptor at 100 ng per plate in all cases. Empty pSG5 was employed to keep the total amount of expression vector identical in all samples. The cells were incubated in the absence (hatched bars) or presence (filled bars) of 100 nm T₃, and the relative luciferase activity was determined as for panel A. Data represents the average and range of duplicate experiments. C, Effect of hormone on dominant negative properties of the R320L mutant. The same general experimental design as in panel B was used, but employing no exogenous receptor (open bars), wtT₃Rβ alone (filled bars) or a 5:1 ratio of R320L to wtT₃Rβ (horizontal striped bars) under a range of T₃ concentrations (indicated below the figure). A calcium phosphate transfection method was used. Data represent the average and range of duplicate experiments.

We next examined the abilities of the RTH mutants to act as dominant negatives when cointroduced together with the wt receptor (Fig. 4B). The G345S and Δ432G mutants were strong dominant negatives in this cotransfection assay and significantly inhibited the functions of the wt receptor in a dose-dependent manner. In contrast, the R320L mutant acted only as a weak inhibitor of wtT₃Rβ at 100 nm T₃. We retested the dominant negative properties of R320L at a range of hormone concentrations (Fig. 4C). Consistent with the effects of hormone on the SMRT/R320L-T₃R association in vitro, the R320L mutant functioned in cells as a strong dominant negative at low T₃ concentrations (i.e. at hormone levels that fail to displace SMRT) but lost dominant negative activity at higher T₃ concentrations (presumably reflecting the release of SMRT). We conclude that the RTH-T₃Rs that exhibited a strong constitutive association with SMRT in vitro manifested the strongest dominant negative phenotypes, whereas the mutant receptor that could be partially dissociated from SMRT by T₃ exhibited weaker, T₃-sensitive dominant negative properties.

Exogenously Introduced SMRT Derivatives Can Interfere with the Dominant Negative Phenotype of the RTH-T₃Rs

The N terminus of SMRT is required for transcriptional repression, whereas the C terminus contains the domains necessary for T₃R association (Fig. 5A and Refs. ^11–15). As a consequence, ectopic expression of an N-terminally truncated SMRT (ΔN-SMRT) interferes with the wtT₃R-mediated repression observed in the absence of hormone (Refs. ¹¹ and ¹⁴ and Fig. 5D), presumably by displacing endogenous full-length SMRT from the receptor (Fig. 5C). In contrast, expression of ΔN-SMRT has little or no effect on the transcriptional activation observed for wtT₃Rβ in the presence of T₃ (Fig. 5D and Ref. 14), conditions in which endogenous SMRT is not bound to the wt receptor (Fig. 5A). If the dominant negative actions of the RTH-T₃Rs reflect a hormone-resistant association with SMRT (Fig. 5B), cointroduction of a ΔN-SMRT deletion should counteract the dominant negative RTH phenotype and partially restore the thyroid hormone response (Fig. 5C).

Fig. 5. ΔN-SMRT Expression Reverses the Dominant Negative Actions of the RTH-T₃Rβs.

A, Model of the actions of SMRT corepressor on wtT₃R. SMRT is bound to receptor in the absence of T₃ and mediates transcriptional repression, but dissociates under physiological T₃ levels to permit transcriptional activation. The N-terminal (n) SMRT domain required for transcriptional repression (hatched) and the C-terminal (c) domain required for receptor association are depicted. B, Model of the actions of SMRT on RTH-T₃Rs. SMRT is bound to receptor in the absence of T₃, but remains bound under physiological T₃ conditions. As a result, the RTH-T₃R behaves as a constitutive repressor and interferes in a dominant fashion with the actions of the wtT₃R. C, Model of the actions of ΔN-SMRT derivatives on receptor-mediated repression. ΔN-SMRT retains the ability to bind receptor, but lacks the N-terminal domain required to silence transcription. As a result, the ectopic expression of ΔN-SMRT displaces endogenous full-length SMRT from T₃R and interferes with receptor-mediated repression. D, ΔN-SMRT interference with the dominant negative phenotype of the Δ432G T₃ mutant. CV-1 cells were transfected with the DR-4 luciferase reporter and with various combinations of pSG5 vector expressing wtT₃Rβ (0.5 µg per plate), the Δ432G mutant (2.5 µg per plate), or ΔN-SMRT (2.5 µg per plate), as indicated below the panel. A calcium phosphate precipitation procedure was employed. Empty pSG5 was introduced where necessary to keep the total amount of expression vector the same for all samples. The cells were incubated in the absence (hatched bars) or presence (filled bars) of 100 nm T₃, and the relative luciferase activity was determined as for Fig. 4. Data represent the average and range of duplicate experiments.

We tested this hypothesis using the Δ432G mutant (which exhibited both a very strong dominant negative phenotype in vivo and an enhanced association with SMRT in vitro). As before, introduction of the Δ432G mutant severely inhibited T₃-mediated gene activation by wtT₃Rβ (Fig. 5D). However, these dominant negative properties of the Δ432G mutant were strongly counteracted by the cointroduction of the ΔN-SMRT derivative, resulting in significant restoration of reporter gene activation by the wtTR₃ (Fig. 5D). These effects were proportional to the amount of ΔN-SMRT expression vector introduced and were not observed for an empty pSG5 vector (Fig. 5D and data not shown).

Artificial Mutants of RTH-T₃Rs That Are Impaired in SMRT Association Are Also Impaired in Dominant-Negative Function

Although several distinct receptor domains interact with corepressor, a conserved amino acid sequence in the receptor ligand-binding domain is particularly important for corepressor association (Fig. 1 and Refs. ^11–15). Receptors bearing mutations in this conserved sequence are impaired for corepressor association in vitro and for transcriptional repression in vivo, but retain the ability to activate transcription (Refs. ¹¹ and ¹⁴ and data not shown). To ask whether impairment of SMRT association results in impairment of the RTH dominant negative phenotype, we engineered analogous SMRT-association disruptive (sad) mutations into four of our representative RTH mutant clones (R320L, G345S, Δ432G, and P453A). When tested in vitro, these RTH-T₃R^sad mutants were all significantly reduced in their ability to bind to GST-SMRT relative to the parental receptors (Fig. 6A). The residual SMRT binding displayed by the T₃R^sad proteins (Fig. 6A) has been noted before and is likely mediated by additional points of contact between SMRT and receptor that are not altered in this mutagenesis scheme (14).

Fig. 6. Effects of Mutations That Inhibit SMRT Association on the Properties of RTH-T₃Rβs.

A, Inhibition of T₃Rβ binding to SMRT in vitro by SMRT association-disruptive (sad) mutations. The RTH (solid bars) or RTH^sad (speckled bars) T₃R derivatives were tested for the ability to bind in vitro to nonrecombinant GST (left panel) or GST-SMRT (right panel). The amount of radiolabeled T₃R bound was determined as in Fig. 2, quantified by Phospho-Imager analysis, and is presented in arbitrary units. B, Impairment of dominant negative activities of RTH-T₃Rβs by sad mutations. The dominant negative assay described in Fig. 4 was repeated, but employing the RTH-T₃R^sad mutants, as indicated below the panel. Unmodified Δ432G RTH-T₃Rβ (last lane) was also included as a control. Wild type T₃Rβ activity was defined as 100

In keeping with our hypothesis of an obligatory role for corepressor in RTH dominant negative function, the RTH-T₃R^sad mutants were all severely compromised in their ability to interfere with wtT₃R function (compare Fig. 6B to Fig. 4B). Little or no inhibition of T₃-mediated reporter gene activation was detected when T₃R^sad versions of R320L, G345S, Δ432G, or p453A were introduced at 1:1 or 5:1 ratios relative to wtT₃R (Fig. 6B). This contrasts with the strong inhibition of wtT₃Rβ activation observed for the cointroduction of the unmodified G345S, Δ432G, or P453A mutants (Fig. 6B, last lane, and Fig. 4B). Note that repression of the reporter gene in the absence of hormone, mediated by the unmodified wtT₃Rβ, is still observed in Fig. 6B, confirming that the defects in transcriptional repression are specific for the receptors bearing the sad mutations. The analogous sad mutant form of the wtT₃Rβ fails to repress, but retains the ability to activate, transcription when introduced individually into CV-1 cells (data not shown), indicating that the sad lesion does not abolish receptor expression, nuclear localization, or DNA binding.

DISCUSSION

The RTH Syndrome Is a Complex Genetic Disorder Characterized by a Diminished Physiological Response to Thyroid Hormone

More than 300 RTH kindreds have been identified, with most of the known genetic lesions mapping to two domains (codons 310 to 349 and 429 to 461) within the T₃Rβ locus (32–34, 38). These domains of the receptor play multiple roles in wtT₃Rβ function, including hormone binding, receptor dimerization, and interactions with the transcriptional machinery (1–6); as a result an elucidation of the precise molecular basis of the RTH phenotype has been complex. Significantly, the vast majority of RTH mutants display a diminished or complete loss of affinity for T₃ hormone. Apparently as a consequence, many RTH-T₃Rs can repress gene transcription, but are defective in gene activation in response to hormone (40–50). Thus, it has been proposed that RTH-T₃Rs function as dominant negative inhibitors, interfering in trans with the actions of the normal T₃Rα and -β (40–50).

However, many features of RTH remain poorly understood, and its precise molecular mechanism remains elusive. Various models by which RTH-receptors could function as dominant negatives have been proposed, including 1) the formation of inactive dimmers between RTH-T₃Rs and wtT₃Rs, 2) a competition between RTH and wt receptors for essential cofactors, or 3) a competition between RTH and wt receptors for DNA-binding sites. This complexity is paralleled by the clinical phenotype of RTH syndrome, which varies in its severity and characteristics from kindred to kindred. Thus, the known biochemical properties of the RTH receptors do not invariably predict the severity of the disease state, and individuals with the same genetic lesion can display different symptoms (32–34, 48, 49, 51, 52). It has been suggested that additional, independently inherited factors may be involved in RTH (32, 49, 53, 54).

We therefore asked whether the recently elucidated SMRT/N-CoR family of corepressors might be involved in the pathogenesis of RTH. In this manuscript, we present evidence that RTH-T₃Rβ mutants are indeed altered in their interactions with SMRT and that these alterations in corepressor association appear to play a role in the dominant negative properties of the RTH receptors and may therefore influence the clinical manifestations of the disease.

RTH Mutant Receptors Exhibit Defects in T₃-Mediated Dissociation from Corepressor

Wild type T₃Rβ binds to SMRT in the absence of T₃ but dissociates on binding hormone, a process paralleled by a conversion of the receptor from a transcriptional repressor to an activator (11–15). The RTH-T₃Rs analyzed here share this ability to associate with SMRT in the absence of T₃ but are notably impaired in the ability to dissociate from corepressor on addition of hormone. Four of the RTH mutants tested required higher than normal levels of hormone to dissociate from SMRT, whereas seven others failed to dissociate significantly from SMRT at even 1 µM hormone. Given the strong transcriptional silencing properties of SMRT (11, 14), the RTH-T₃R/SMRT complexes would be predicted to repress transcription under hormone conditions in which wtT₃Rβ activates, a prediction consistent with the dominant negative properties of the RTH mutants. Indeed, the RTH receptors that were constitutively bound to SMRT exhibited strong, hormone-refractory dominant negative properties in our transfection studies. In contrast, an RTH-receptor (R320L) that exhibited impaired, but detectable, release of corepressor at high hormone concentrations displayed a similarly hormone-labile dominant negative phenotype.

Two RTH Mutations Appear to Uncouple SMRT Dissociation from Hormone Binding

The failure of most of our RTH mutants to release from SMRT in the presence of T₃ could be attributed to the impaired affinities of the mutant receptors for this hormone. For example, the R338W mutant exhibits a 10-fold reduction in affinity for T₃ relative to wtT₃Rβ and requires some 6-fold more hormone for SMRT dissociation; similarly the G345S mutant is virtually unable to bind hormone in vitro and fails to significantly dissociate from SMRT at even 1 µM T₃. However, two striking exceptions were noted to this general correlation between affinity for T₃ and SMRT release. The P453A and P453H mutants are only mildly impaired in hormone binding in vitro (possessing even higher affinities for T₃ than the R338W mutant), yet fail to dissociate from SMRT at any hormone concentration tested. Protease sensitivity assays (56) confirm that the P453A and P453H mutant receptors are indeed occupied by hormone at these T₃ concentrations that fail to displace SMRT (data not shown). Notably, the P453A and P453H mutants represent different amino acid substitutions at the same codon, in a region of the receptor that is proposed to undergo a conformational change on binding hormone (57, 58). Codon 453 may therefore define a receptor domain critical for corepressor release after hormone binding, and lesions at this site may confer RTH by preventing SMRT dissociation in a manner independent of hormone occupancy.

Consistent with these concepts, mutants at codon 453 possess dominant negative properties that are relatively refractory to T₃ (48). This region of the receptor has been suggested to operate as a hinge, permitting a C-terminal tail of the receptor to rotate from an exposed position in the absence of hormone to a new position nested against the body of the receptor in the presence of hormone (57, 58). Our preliminary evidence, employing protease sensitivity as a probe of structure, supports the concept that this hinge function may be impaired in the P453A and P453H mutants (B. Lin, S. Yoh, and M. L. Privalsky, manuscript in preparation).

It should be noted that the hormone concentrations required for inhibition of T₃R/SMRT binding in our GST assay were higher than the association constants (K_a) for T₃ reported for the same receptors (Table 1). This phenomenon was consistently observed in multiple assays, and was also seen for wtT₃Rα,-β, and for retinoic acid receptors. Intriguingly, analogous discrepancies exist between the K_a values of these receptors and the significantly higher hormone concentrations required for transcriptional activation in transient transfection assays (e.g. Refs. ⁴⁷ and ⁴⁸). These differences may simply be technical in origin, i.e. due to sequestration of hormone by components of the transfection/binding assays (47). On the other hand, this phenomenon may reflect a physiologically relevant process. For example, SMRT association may alter the affinity of T₃Rs for hormone, analogous to the effects reported for heterodimer formation with retinoid X receptor (59, 60).

Two RTH Mutants Also Exhibit Dramatically Enhanced Levels of SMRT Association

Two of the nine RTH mutants analyzed (Δ430 M and Δ432G) were not only impaired in T₃-mediated dissociation from SMRT, but also exhibited a strongly elevated association with SMRT even in the absence of hormone. Consistent with this enhanced SMRT binding, the Δ430 M and Δ432G mutants exert dominant negative phenotypes that are among the strongest observed for any of RTH mutants tested; this is particularly evident with certain promoters and under high hormone concentrations (48). Notably, these two, independently derived mutations represent in-frame, single-codon deletions that map to an α-helical domain in the T₃Rβ C terminus (based on the crystallographic model of T₃Rα; Ref. 57). This helix appears to play important roles in the conformational changes associated with hormone binding, in transcriptional activation, and in dimer formation with other receptors, but has not been previously identified as a direct contact site for corepressor. It appears that by shortening or rotating this region by one amino acid residue, the binding of corepressor can be greatly enhanced, either by directly affecting an interaction surface or by a more indirect effect on the global conformation of the receptor C terminus.

Corepressor Association Appears to be Required for the Dominant Negative Actions of the RTH-T₃Rs

Correlating with their dominant negative properties, all 11 RTH mutants interacted with corepressor under hormone conditions in which the wtT₃R does not. Furthermore, mutations introduced into RTH-T₃Rs that disrupt SMRT association, or transfection of ΔN-SMRT derivatives that interfere with SMRT function, disrupted the dominant negative phenotype. These results strongly implicate corepressors in the ability of RTH-T₃Rs to act as dominant negative inhibitors of the T₃ response. In these properties, the RTH-T₃Rβ mutants closely parallel v-Erb A, an oncogenic allele of T₃Rα that both acts as a dominant negative allele and exhibits hormone-independent corepressor association (11, 14). The ability of v-Erb A to repress gene transcription and to function in oncogenesis closely correlates with SMRT association (11, 14). However, it should be noted that corepressors are a diverse family of protein factors, and the dominant negative properties observed for v-Erb A and for RTH-T₃Rs may be mediated by SMRT itself, N-CoR, or some, as yet unidentified, corepressor-like entity.

A Correlation May Exist between the SMRT Association Properties of RTH-T₃Rs in Vitro and the Clinical Manifestations of RTH

Clinically, RTH has been broadly divided into a pituitary form (PRTH) characterized by a predominant pituitary resistance but preserved peripheral response to T₃, vs. a generalized form (GRTH) characterized by T₃ resistance at the level of both pituitary and peripheral tissues (reviewed in Refs. ^32–34). Despite these different clinical manifestations, receptors isolated from PRTH and GRTH are often similar or identical to one another in their biochemical properties. It is therefore interesting that the RTH-T₃Rs that display a highly elevated SMRT association (Δ430M and Δ432G) in the current study were both isolated from patients presenting with PRTH. It is possible that this constitutively elevated SMRT association may play a role peculiar to the PRTH phenotype. However, the converse did not appear to be true: not all PRTH-T₃R mutants also displayed elevated SMRT association in vitro (e.g. R338W).

Notably, none of the experiments described here were performed in different tissues or in the genetic background of the original patients. Therefore, our results do not address whether alterations in corepressor, rather than in receptor, may also contribute to the RTH syndrome. It is tempting to speculate that differences in corepressor expression, perhaps coupled to genetic polymorphisms at the corepressor loci, might account for the variable resistance of the RTH syndrome in different individuals of the same kindred. Similarly, differences in corepressor expression in different tissues could contribute to the varying organ-specific effects of RTH. Clearly, however, RTH is a complex clinical disease and is associated not only with a failure to activate expression of certain genes in response to hormone, as tested here, but also as a failure to suppress expression of others, most notably that of pituitary TSH (32–34). Further work will be necessary to determine the contributions of corepressors or related factors to these different manifestations of this endocrine disorder.

MATERIALS AND METHODS

Molecular Clones

Wild type T₃Rβ was the generous gift of R. M. Evans; the origins of the RTH mutant clones have been described previously (7, 48). All T₃Rβ alleles were introduced as BamHI to BamHI fragments into a pSG5 vector for expression in vitro and for transient transfections. The T₃R^sad mutants (equivalent to an A223G, H224G, T227A triple mutant, using the numbering system of Ref. 12) were created by a PCR protocol (61). Appropriate restriction fragments of the resulting PCR product was used to replace the corresponding sequences within the pSG5 clones of the RTH-T₃Rβ alleles. Mutations were confirmed by restriction digestion/DNA sequence analysis.

Assay of the SMRT/T₃Rs Interaction in Vitro

GST-SMRT (identical to the GST-TRAC-1 construct described previously) was isolated from transformed Escherichia coli and was immobilized by binding to glutathione agarose, as previously described (14). ³⁵S-Labeled receptors were synthesized by in vitro transcription from pSG5 templates using T7 RNA polymerase, coupled to in vitro translation using rabbit reticulocyte lysates (TNT kit, Promega, Madison, WI). Each receptor was subsequently mixed with approximately 2 µg immobilized GST-SMRT (bound to 20 µl glutathione-agarose) in 200–400 ml HEMG buffer (14) containing 10 mg/ml BSA and a protease inhibitor cocktail (Complete, Boehringer-Mannheim, Indianapolis, IN). T₃ hormone (or an equivalent volume of ethanol carrier) was included in the binding reactions where indicated. The binding reactions were incubated for 60–90 min at 4 C with gentle rocking, and the agarose matrix was subsequently washed with 4 × 1 ml changes of HEMG buffer. Bound proteins were eluted in 35 µl of 50 mm Tris-Cl (pH 7.6) containing 10 mm glutathione, resolved by SDS-PAGE, and visualized by autoradiography (14). Quantification of the binding experiments were performed with a Molecular Biosystems Storm phosphoimaging system (Sunnyvale, CA).

Transient Transfections

For calcium phosphate transfections, CV-1b cells (2 × 10⁵ per 60-mm plate) were propagated overnight at 37 C in a 5% CO₂ atmosphere in DMEM containing 10% heat-inactivated FBS and penicillin-streptomycin (GIBCO/BRL, Gaithersburg, MD). The cells were subsequently transferred to hormone-depleted medium and incubated an additional 6 h. Calcium phosphate/DNA precipitates, prepared by standard protocol (62), were then added, and the cells were incubated for an additional 16–18 h. Typically, each plate received 1 µg pCH110-lacZ (employed as an internal standard), 1 µg of a DR4-luciferase reporter (63), various combinations of empty pSG5, pSG5-T₃Rβ, or pSG5-ΔN-SMRT (as indicated), and sufficient pUC18 or pUC19 to bring the total DNA to 10 µg. After an overnight incubation with precipitate, the cells were washed and transferred to fresh medium lacking or containing T₃. The cells were harvested 30 h later and lysed in 100 µl Reporter Lysis Buffer (Promega) per plate. Luciferase activity was determined by mixing 30 µl of extract with 100 µl Promega luciferase reagent in an MGM luminometer (MGM Instruments, Cambridge, MA); β-galactosidase activity was determined by colorimetric assay (14, 64).

Lipofections were performed using 3 × 10⁴ to 5 × 10⁴ cells seeded per 1 cm-well in 24-well microtiter plates. Lipofectin reagent (GIBCO/BRL) was diluted 10-fold into serum-free medium, incubated 30–45 min at room temperature, then mixed with the DNA and incubated an additional 10 min. Typically, 300 ng pCH110, 100 ng DR4-luciferase reporter, and 100 ng pSG5-T₃R construct were employed per well, with sufficient pUC19 DNA to bring the total DNA concentration to 1 µg. The cells were overlayed with the DNA and Lipofectin mixture, incubated 6 h at 37 C, subsequently incubated 18 h with medium containing 20% FBS, and then incubated 24–48 h in medium containing 10% FBS with or without hormone. Cells were lysed and assayed for luciferase and β-galactosidase activity in a similar manner as that described for the calcium phosphate method.