Defective Release of Corepressor by Hinge Mutants of the Thyroid Hormone Receptor Found in Patients with Resistance to Thyroid Hormone*

On positive thyroid hormone response elements (pTREs), thyroid hormone receptor (TR) binding to DNA in the absence of ligand (thyroid hormone, T3) decreases transcription (silencing). Silencing is due to a family of recently described nuclear corepressor proteins (NCoR and SMRT) which bind to the CoR box in the hinge region of TR. Ligand-dependent activation of TR is associated with displacement of corepressors and recruitment of coactivating proteins. Resistance to thyroid hormone (RTH) is due to mutations in the β isoform of the thyroid hormone receptor (TR-β). To date, three RTH mutations reportedly with near-normal T3binding (A234T, R243Q, and R243W) have been described in or near the CoR box. To determine the mechanism of RTH caused by these mutants, the interaction of wild type (wt) and mutant TRs with the corepressor, NCoR, and the coactivator, SRC-1, was tested in gel-shift assays. As expected, NCoR bound wt TR in the absence of T3 and dissociated from TR with increasing T3 concentration. SRC-1 failed to bind wt TR in the absence of T3, but bound to TR with increasing avidity as T3 concentrations rose. At no T3 concentration did both NCoR and SRC-1 bind to wt TR, indicating that their binding to TR was mutually exclusive. Hinge mutants bound NCoR normally in the absence of T3; however, dissociation of NCoR and recruitment of SRC-1 was markedly impaired except at very high T3 concentrations. Importantly, hinge mutant TRs when complexed to DNA bound T3 poorly despite their near-normal T3 binding in solution. These binding studies correlated with functional assays showing defective transactivation of pTREs by hinge mutants except at high T3concentrations. Thus, we describe a novel mechanism of RTH whereby TR hinge mutants selectively affect T3 binding when complexed to DNA, and prevent NCoR dissociation from TR. Our data also suggest that solution T3 binding by RTH mutants may not accurately reflect physiologically relevant T3 binding by TR when bound to DNA.

On positive thyroid hormone response elements (pTREs), thyroid hormone receptor (TR) binding to DNA in the absence of ligand (thyroid hormone, T 3 ) decreases transcription (silencing). Silencing is due to a family of recently described nuclear corepressor proteins (NCoR and SMRT) which bind to the CoR box in the hinge region of TR. Ligand-dependent activation of TR is associated with displacement of corepressors and recruitment of coactivating proteins. Resistance to thyroid hormone (RTH) is due to mutations in the ␤ isoform of the thyroid hormone receptor (TR-␤). To date, three RTH mutations reportedly with near-normal T 3 binding (A234T, R243Q, and R243W) have been described in or near the CoR box. To determine the mechanism of RTH caused by these mutants, the interaction of wild type (wt) and mutant TRs with the corepressor, NCoR, and the coactivator, SRC-1, was tested in gel-shift assays. As expected, NCoR bound wt TR in the absence of T 3 and dissociated from TR with increasing T 3 concentration. SRC-1 failed to bind wt TR in the absence of T 3 , but bound to TR with increasing avidity as T 3 concentrations rose. At no T 3 concentration did both NCoR and SRC-1 bind to wt TR, indicating that their binding to TR was mutually exclusive. Hinge mutants bound NCoR normally in the absence of T 3 ; however, dissociation of NCoR and recruitment of SRC-1 was markedly impaired except at very high T 3 concentrations. Importantly, hinge mutant TRs when complexed to DNA bound T 3 poorly despite their near-normal T 3

binding in solution.
These binding studies correlated with functional assays showing defective transactivation of pTREs by hinge mutants except at high T 3 concentrations. Thus, we describe a novel mechanism of RTH whereby TR hinge mutants selectively affect T 3 binding when complexed to DNA, and prevent NCoR dissociation from TR. Our data also suggest that solution T 3
Resistance to thyroid hormone (RTH) is the result of mutations in the carboxyl terminus of the ␤ thyroid hormone receptor (TR-␤) (30 -33). Individuals with the disorder require greater thyroid hormone (T 3 ) concentrations in order to achieve T 3 -dependent actions in tissues. RTH is a dominant disorder in which most individuals are heterozygous for a mutant TR-␤ allele. In a phenomenon called dominant negative activity, the mutant allele interferes with the activity of the normal allele (34 -37). RTH mutations congregate in two major "hot spots" in the ligand binding domain of TR-␤ (38,39). Recently, mutations have also been reported in the hinge region, suggesting a third hot spot for mutations within the TR-␤ locus (41)(42)(43).
To date, six naturally occurring RTH mutations have been described in or near the hinge region (amino acids 174 -237 (44) of TR: A234T, R243Q, R243W, V264D, T277A, and R282S (40,42,43). The former three mutations are located in or near the CoR box. Based on the location of corepressor binding, RTH mutations in or near the CoR box might be expected to alter corepressor binding and affect TR function. While the A234T mutation is reported to have mildly decreased T 3 binding, R243Q and R243W are reported to bind T 3 normally in solution * This work was supported by National Instiutes of Health Grants DK-02423 (to J. D. S.) and DK-43653 and DK-49126 (to F. E. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Site-directed mutagenesis (CLONTECH, Palo Alto, CA) was used to create hinge mutations in the context of the human TR-␤1 isoform (45). All mutations were confirmed by DNA sequencing of the TR-␤1 C terminus. The expression vector used for transfections in this study was pSG5, containing either wild type (wt) or mutant TR-␤ cDNAs as EcoRI fragments. The human TR-␤1 cDNA was used as the wt TR-␤1. Expression vector plasmid preparations used in this study were carefully quantitated by agarose gel electrophoresis. To confirm the integrity and quality of each expression vector plasmid DNA preparation, in vitro translation with [ 35 S]methionine was performed using T7 polymerase and the products analyzed by SDS-polyacrylamide gel electrophoresis.
The SRC-1 nuclear binding domain (NBD-1) fragment was generated by polymerase chain reaction of amino acids 594 -780 of an F-SRC-1 clone generously provided by William W. Chin. The generation of the binding domain of NCoR, NCoR-I, has been previously described (9).
Reporter constructs included two copies of idealized pTREs, direct repeat with a four-base pair interval (DRϩ4) (41) and inverted palindrome (chicken lysozyme F2) (46). Constructs contained the pTRE element fused upstream of a Ϫ109 thymidine kinase promoter and the luciferase gene to measure activity. The luciferase reporter gene was derived from pA3 and contained two transcriptional stop sequences upstream of the promoter to prevent read-through transcription. The reporter was documented to have no positive thyroid hormone response in the absence of response elements.
Transfections were performed in CV-1 cells which are relatively TR-deficient (47,48). Each transfection included a mutant or wt TR-␤ expression vector (500 ng) and a response element reporter construct (10 g). Cell cultures were transfected using a calcium-phosphate precipitation method and precipitate applied for 16 h in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, glutamine, and appropriate antimicrobials. The next day, media was removed and Dulbecco's modified Eagle's medium containing both anion exchange and charcoal stripped fetal bovine serum (10%) was added Ϯ T 3 . Concentrations of T 3 were used as noted. Data were pooled from at least three independent experiments and displayed as mean Ϯ S.E.
Gel-shift studies were performed using constructs of the wt and mutant TRs placed in a pGEM vector. To study corepressor binding, a construct was fashioned inserting the TR interacting domains of NCoR (9) into the pKCR2 expression vector. TRs and the binding domain of NCoR were in vitro translated using rabbit reticulocyte lysate (Promega, Madison, WI). The nuclear binding domain (NBD-1) of SRC-1 (21,49) was inserted into a procaryote glutationine S-transferase (GST) expression vector resulting in a GST fusion protein. TR-SRC interactions were evaluated accordingly. A polyclonal anti-GST antibody (Santa Cruz Biotechnology, Santa Cruz, CA) which specifically disrupted the TR/SRC⅐GST complex was used to more clearly identify TR-SRC-1 binding.
The binding affinities for wt and mutant TRs in solution were assessed with a filter binding assay utilizing 125 I-labeled T 3 (50). K a values were assigned after Scatchard analysis. T 3 binding on DNA was also assessed following a standard gel-shift protocol except that 125 Ilabeled T 3 replaced 32 P-labeled DRϩ4 for visualization. Fig. 1, three TR-␤ mutations were evaluated in this study (A234T, R243Q, and R243W). These mutations were introduced into a TR-␤1 cDNA in pGEM 3, using site-directed mutagenesis and transferred to the expression vector pSG5 for use in transient transfection studies. Mutant TR binding to T 3 in solution (Table I) is not significantly impaired with each mutant displaying only a 20 -40% reduction in T 3 binding. These results confirm those of other laboratories studying the T 3 binding properties of these mutant receptors (40,42,43).

TR-␤ Mutations Used in This Study-As shown in
Ligand-dependent Dissociation of Corepressor from Mutant TRs Is Impaired-We first evaluated mutant TR binding to the nuclear corepressor, NCoR, using a gel-shift assay. The three hinge mutants noted above (A234T, R243Q, and R243W), a severe GRTH mutant (⌬337T) (51), and an artificial hinge mutant known to not interfere with corepressor binding (E220R) (7) were tested in this assay. A portion of NCoR, containing the TR interacting domains but lacking the repressing domains, termed NCoR-I, was employed in the assay (9). In   vitro translated TRs formed a dimer (D) in the absence of NCoR-I on a DRϩ4 radiolabeled element. Addition of in vitro translated NCoR-I caused the dimer to be shifted to a new position and serial dilution of the amount of NCoR-I added to the binding reaction reduced the intensity of this complex. Ligand-independent binding of NCoR-I to each of the mutants was similar to wt TR (Fig. 2). Addition of RXR-␣ or RXR-␤ to   Fig. 2 is demonstrated. Reticulocyte lysate generated TRs were incubated with GST fusion protein generated SRC-1 (NBD-1) across a range of T 3 concentrations as noted: A, wt; B, A234T; C, R243Q; and D, R243W. The R243Q mutant is nearly normal in SRC-1 (NBD-1) recruitment; the A234T and R243W mutants require a T 3 concentration twice that for wt to recruit SRC-1. the binding reaction resulted in the formation of a strong heterodimeric complex, and a significant reduction in the TR⅐NCoR-I complex (data not shown).
In contrast, as shown in Fig. 3, T 3 dissociation of NCoR-I from certain mutant TRs was much different than from wt TR on the DRϩ4 element. With wt TR, 5 nM T 3 resulted in an almost complete dissociation of the TR/NCoR-I complex (panel A). The artificial hinge mutant with normal T 3 binding, E220R (panel B), had a similar NCoR-I dissociation pattern. The A234T mutant (panel C), however, required about 2-fold higher concentration of T 3 to dissociate NCoR-I than did wt TR. The R243Q and R243W mutants (panels D and E) required about 20-fold greater T 3 concentration relative to wt TR to dissociate this complex. Finally the ⌬337 mutant, which cannot bind T 3 , could not be dissociated from NCoR-I even at the 100 nM T 3 concentration. Similar observations were made on the chicken lysozyme TRE (Lys, Fig. 4) and palindromic TRE (data not shown). On the Lys element, a higher T 3 concentration was necessary to completely dissociate NCoR-I from wt TR (100 nM). Dissociation of NCoR-I from E220R and ⌬234T was similar to wt, while the R243Q and R243W mutant TRs required a 10-fold higher T 3 concentration (1000 nM) than did wt TR for dissociation. As expected, NCoR-I could not be dissociated from the ⌬337T mutant even at this very high T 3 concentration.
Recruitment of SRC-1 by Mutant TRs Is Mildly Impaired Relative to wt TR-To test whether hinge mutants could normally recruit SRC-1, the NBD-1 of SRC-1 (21,49) was fused to the GST protein and expressed in bacteria. Purified GST and GST-SRC-1(NBD-1) were then employed in a gel-shift study of wt and mutant TR binding to a DRϩ4 probe (Fig. 5). In the absence of T 3 , SRC-1 did not form a complex with wt TR (panel A). A wt TR⅐SRC-1 complex was first noted at 2.5 nM and was maximal at 100 nM T 3 . These results were unexpected since T 3 is known to dissociate the TR homodimer and prevent its binding to DNA. The presence of a TR⅐SRC-1 complex indicates that the SRC-1(NBD-1) must, in some way, stabilize the binding of TR on DNA in the presence of T 3 . The three hinge mutants also recruited the NBD-1 of SRC-1 with increasing T 3 concentrations. The R243Q mutant (panel C) appeared qualitatively similar to wt TR in its ability to recruit SRC-1 over the T 3 concentrations tested. The A234T (panel B) and R243W (panel D) mutants required about 2-fold greater T 3 concentrations, relative to wt TR, to begin to recruit SRC-1(NBD-1). In Fig. 6, these experiments were repeated in the presence of RXR. SRC-1(NBD-1) was able to bind to both wt TR homodimers and wt TR/RXR heterodimers (panel A) as evidenced by the decrease in intensity of both complexes and the presence of two closely spaced more slowly migrating complexes on a lighter exposure of this autoradiogram (data not shown). Qualitatively, how-  P-labeled DRϩ4 gel shift was performed with reticulocyte lysate generated TRs incubated with TR binding domains of NCoR and SRC-1 across a range of T 3 concentrations: A, wt; B, R243Q; and C, R243W. An anti-GST antibody which specifically dissociates the TR⅐SRC-1 complex was added at each T 3 concentration. In the absence of ligand, the slower moving TR⅐NCoR complex is evident. At high T 3 concentrations, the faster moving TR⅐SRC-1 complex appears. Although there are T 3 concentrations at which both NCoR and SRC-1 bind TR, at no point is a band representing both cofactors bound to TR present. With the two hinge mutants shown, the TR⅐SRC-1 complex only appears at the highest T 3 concentrations when the TR⅐NCoR complex is dissociated. ever, the defect in SRC-1 association caused by the hinge mutants was similar to the data obtained without RXR. Thus, hinge mutants display a significantly greater impairment in T 3 -mediated dissociation from NCoR-I than T 3 -mediated association to the NBD-1 of SRC-1.
TR Binding to NCoR Prevents SRC-1 Recruitment-To determine whether NCoR binding interfered with SRC-1 recruitment, the gel-shift assays were repeated using TR interacting domains from both cofactors. In Fig. 7 gel-shifts obtained using the wt TR and two hinge mutants, R243Q and R243W are shown. Although the TR⅐NCoR-I and TR⅐SRC-1(NBD-1) complexes migrated at a similar position, closer inspection revealed that the TR⅐SRC-1(NBD-1) complex migrated slightly faster than the TR⅐NCoR-I complex. We used this difference in migration position and an antibody directed against the GST protein to distinguish the two complexes. In the absence of T 3 where only NCoR-I would be expected to bind to the wt TR, one distinct complex was formed and the location and intensity of the complex was not affected by addition of the GST antibody (panel A). Conversely at high T 3 concentration (100 nM) where only SRC-1(NBD-1) should bind to TR, a faster migrating complex was formed which was completely eliminated by the addition of the GST antibody. A gradual shift from the TR⅐NCoR-I to the TR⅐SRC-1(NBD-1) complex was noted over the T 3

concentrations tested (panel A).
Under no condition was a complex formed consistent with binding both NCoR-I and SRC-1(NBD-1) to wt TR. Rather, as the T 3 concentration rose, the shifted band migrated from the  9. A, on the DRϩ4 reporter, the R243 mutants suffer dimished ligand dependent stimulation at low and intermediate doses of T 3 . The A234T mutant suffers more mildly decreased activity. At high T 3 doses, all three hinge mutants tested have activity similar to wt. B, there is diminished relative activity of the TR mutants across T 3 concentrations on the Lys probe. As with DRϩ4, the degree of impairment is greater for the 243 mutants than for the A234T mutant. The ⌬337T mutant has no significant ligand-dependent activity at any T 3 concentration tested.
higher TR⅐NCoR-I position to the lower TR⅐SRC-1(NBD-1) position. This was particularly well seen at a T 3 concentration of 2.5 nM, where both NCoR-I and SRC-1(NBD-1) were capable of binding wt TR when tested individually. At this concentration and in the absence of the GST antibody, wt TR bound in a faster migrating complex consistent with its binding to SRC-1(NBD-1). Addition of the GST antibody shifted the position of the complex slightly to a location predicted for TR⅐NCoR-I binding. Since the GST antibody interferes with formation of the TR⅐SRC-1(NBD-1) complex, these data suggest that removal of SRC-1(NBD-1) from the TR allowed NCoR-I in solution to bind to the wt TR.
The net result of the impaired NCoR-I dissociation from hinge mutants is also shown in Fig. 7 (panels B and C). Only at 100 nM T 3 was NCoR-I sufficiently dissociated from the hinge mutants to permit significant SRC-1(NBD-1) recruitment, based on the position of the protein-DNA complex. The GST antibody blocks formation of this complex, confirming that at this T 3 concentration, the hinge mutants bound exclusively to SRC-1(NBD-1). At 10 nM T 3 , SRC-1(NBD-1) bound somewhat to the hinge mutants based on partial dissociation of the shifted complex by the GST antibody (R243Q Ͼ R243W). This result is consistent with the other data showing that the R243Q mutant recruited SRC-1(NBD-1) at lower T 3 concentrations than the R243W mutant. Similar data was obtained with the A234T mutant (not shown). Thus, an impairment in NCoR dissociation by T 3 , as observed with the hinge mutants, prevents SRC-1 recruitment.
Solution and DNA Complex Binding of Mutant TR Do Not Correlate-It is unclear why the hinge mutants displayed such significant impairment in dissociation from NCoR-I in the absence of significant defects in solution binding to T 3 . Moreover, the R243Q and R243W mutants required significantly greater T 3 concentrations to dissociate NCoR-I than did the A234T mutant despite the fact that these receptors had very similar T 3 binding profiles. To investigate the reason for this discrepancy, gel-shift assays were performed in parallel using either a 32 P-labeled DRϩ4 probe or 125 I-T 3 to identify the shifted complexes. Fig. 8 is such an experiment comparing wt and mutant TRs. In vitro translated RXR-␣ and the indicated TR-␤ were added to each column except columns 1 and 8 where an equal volume of unprogrammed lysate was utilized (UP). In columns 2-7, reticulocyte lysate generated TRs (wt, E220R, A234T, R243Q, R243W, and ⌬337T, respectively) were incubated with RXR-␣ and 32 P-labeled DRϩ4, resulting in heterodimeric and homodimeric complexes of equal intensity. In columns 9 -14 the same paradigm was followed except that the 32 P-labeled DRϩ4 pTRE was replaced by a nonradioactive DRϩ4 pTRE and 125 I-T 3 was added (10 nM final concentration).
As found by others (52), the wt TR/RXR heterodimer was clearly labeled by 125 I T 3 (lane 9) and the homodimer was not labeled, despite equal amount of these complexes on the 32 Plabeled gel-shift. We interpret this result as indicating that occupancy of one TR molecule by radiolabeled T 3 marks the heterodimer which is not dissociated by T 3 , while T 3 bound to one TR molecule is sufficient to dissociate the homodimer and eliminate a labeled band at that position. Because of the migration position of free 125 I T 3 on the gel, we could not determine if the wt TR⅐NCoR-I complex was labeled. The heterodimeric complex containing the A234T mutant was labeled 3-fold less well by T 3 (as determined by densitometry) and heterodimers containing the R243Q, R243W, and ⌬337T mutants were not labeled. Clearly the ⌬337T mutant would not be expected to bind T 3 ; however, results with the hinge mutants were unexpected given their solution T 3 binding. These results indicate that solution and DNA-complex binding of T 3 by hinge mutants of the TR do not correlate.

Mutant TR Function Is Diminished at T 3 Concentrations Where NCoR Remains Bound and Is Not Reversed by Transfection of an NCoR
Inhibitor (NCoR-I)-As a model for thyroid hormone action, two copies of a positive thyroid hormone response element were fused upstream of a heterologous promoter luciferase construct (pTK109-Luc) for use in transient transfection assays of CV-1 cells. On the DRϩ4 element (Fig. 9,  panel A), activation by the wt TR was first noted at 1 nM T 3 and reached a maximum at 100 nM T 3 Activation of this element by the artificial hinge mutant (E220R) was similar to wt TR (data not shown). The natural hinge mutants were clearly defective in activation compared with wt TR. Activation by the A234T mutant was first achieved at 5 nM T 3 , while transactivation by the two 243 mutant TRs was first noted at 10 nM T 3 . At very high T 3 doses, however, transactivation by all mutants except ⌬337T equaled that of wt TR. Each hinge mutant had strong dominant negative activity against wt TR on the DRϩ4 element (data not shown). Qualitatively similar results were noted with the everted palindrome element derived from the chicken lysozyme gene promoter, Lys (Fig. 9, panel B).
Since transfection of an NCoR inhibitor (NCoR-I) expression vector can reverse TR-mediated repression on pTREs (9), we wanted to determine if cotransfection of this plasmid could correct the defect in transactivation observed with the hinge mutants. In Fig. 10A, cotransfection of NCoR-I completely reversed ligand-independent repression by the wt TR on the DRϩ4 element and yielded similar levels of reporter gene activation in the presence of T 3 . Since endogenous corepressors (NCoR and SMRT) and NCoR-I are released from wt TR at physiological T 3 concentrations, transcriptional repression by endogenous corepressors is selectively removed by NCoR-I without affecting T 3 -mediated coactivator recruitment and transactivation. In contrast, in Fig. 10B are data obtained with one of the hinge mutants, R243W. Note that NCoR-I cotransfection completely eliminated TR-mediated repression but did not allow for normal T 3 -mediated transactivation except at 100 nM T 3 . Importantly, these transfection studies are consistent with gel-shift studies showing that TR hinge mutants are markedly defective in their ability to dissociate NCoR-I and permit SRC-1 (NBD-1) recruitment. DISCUSSION Our study of hinge mutations of the TR has revealed several important properties of TR-mediated transactivation and shed new light on the RTH syndrome. First, hinge mutants of the TR display a defective release of NCoR in response to treatment with T 3 . Gel-shift studies demonstrate that both 243 mutants, and to a lesser extent the 234 mutant, remain bound to NCoR at concentrations of T 3 that completely dissociate NCoR from wt TR. Yoh et al. (13) first suggested that RTH mutants displayed aberrant interaction with nuclear corepressors. As expected, mutations which bound T 3 poorly failed to dissociate from corepressor except at very high T 3 concentrations. Our results with the ⌬337T mutation are in keeping with their findings. However, two mutants located in the AF-2 domain (P453A and P453H) appeared to have a much greater defect in T 3 -mediated dissociation from corepressor than would be expected due to their mild decrease (4 -5-fold) in T 3 affinity. The authors suggested that a conformation change in the extreme C terminus of TR caused by the mutation might explain this apparent discrepancy by uncoupling T 3 binding from corepressor dissociation.
We initially favored this hypothesis to explain our findings with the hinge mutants. These mutants displayed good solution T 3 binding (less than a 2-fold decrease), yet failed to dissociate from NCoR in gel-shift experiments and continued to silence gene transcription in transfection studies except at very high T 3 concentrations. Importantly, however, we determined the ability of these mutants to bind to T 3 when complexed to DNA. To our surprise, these mutants as RXR heterodimers bound radiolabeled T 3 poorly, if at all, in gel shift studies. Our results are limited to the heterodimer complex since TR homodimers were not labeled in this assay, perhaps due to the fact they would be dissociated when bound by T 3 , and free radiolabeled T 3 obscured the TR⅐NCoR-I complex, Regardless these data suggest that binding of T 3 in solution and on DNA do not always correlate as highlighted by mutations in the hinge region of TR.
Based on the crystal structure of TR-␣, hinge mutants might affect T 3 binding (53). These mutations are near helix two, which is believed to stabilize the polar pocket which binds T 3 . Since regions of the hinge domain are known to contact DNA (54,55) and to mediate ligand binding (56) it is possible that DNA-binding induces a conformation change in the mutant hinge domain so as either to prevent ligand access to the heterodimeric bound TR or prevent stabilization of ligandbinding on the TR. The net result is that corepressor continues to bind to TR and prevent transactivation despite the presence of T 3 in solution. It is unclear whether this discrepancy between solution and DNA-complex binding of T 3 is specific for hinge mutants or will also be observed with RTH mutants in other locations on the TR-␤. FIG. 11. Model of resistance caused by hinge mutants of the TR. When wt TR is bound to DNA, addition of T 3 results in dissociation of corepressor, permitting recruitment of coactivator and transcriptional activation. When hinge mutant TR is bound to DNA, T 3 cannot bind and NCoR is not dissociated. SRC-1 recruitment is prevented and DNA transcription remains repressed.
In evaluating TR cofactors and their role in RTH, the question of the relative importance of corepressors and coactivators in the genesis of RTH has arisen. For example, impaired liganddependent activity could be associated either with impaired (13) corepressor release or with defective coactivator recruitment (52,57). Our data suggest that impaired corepressor release is the primary defect in RTH patients with hinge mutations and probably for most other patients with RTH. This is based on two lines of evidence. 1) Most RTH mutations affect T 3 binding to TR in solution or on DNA, as is the case with hinge mutants. This would result in impaired release of corepressor at physiological T 3 concentrations and continued gene silencing. 2) Both NCoR and SRC-1 do not bind simultaneously to TR, suggesting at least two discreet steps are necessary for transactivation. Addition of T 3 must first cause dissociation of NCoR before recruitment of SRC-1 can take place. Either the binding locations of the two cofactors must overlap such that both cannot be present on TR or the TR conformation when NCoR is bound is such that SRC-1 binding is not favorable. This result is most clearly seen using the NCoR inhibitor in transfection studies. Transfection of NCoR-I competes with endogenous corepressors, resulting in a loss of ligand-independent repression on pTREs caused by the R243W mutation. Remarkably the mutant TR is still unable to activate transcription except at very high T 3 concentration. Since mutant TR binding to NCoR-I is also resistant to T 3 dissociation, SRC-1 or other coactivators are not recruited to the TR due to lack of binding to the corepressor occupied TR, yielding a lack of transactivation.
We propose the following model (Fig. 11) based on our results. wt TR, in the absence of ligand, binds corepressor and transcription is repressed. Addition of T 3 results in dissociation of corepressor, permitting recruitment of coactivators and transcriptional activation. Although, the hinge mutant TRs bind corepressor normally in the absence of ligand resulting in normal silencing activity, T 3 fails to dissociate the corepressor and thus coactivator recruitment is prevented. While T 3 binding by the hinge mutant TRs is normal or near-normal in solution, their T 3 binding is clearly defective when complexed to DNA. Our data indicate that DNA-binding modifies the ability of hinge mutants to bind T 3 , yielding this novel mechanism of resistance by this class of RTH mutants.