Chromatin remodeling by the thyroid hormone receptor in regulation of the thyroid-stimulating hormone alpha-subunit promoter.

The chromatin architecture of a promoter is an important determinant of its transcriptional response. For most target genes, the thyroid hormone receptor (TR) activates gene expression in response to thyroid hormone (T(3)). In contrast, the thyroid-stimulating hormone alpha-subunit (TSH alpha) gene promoter is down-regulated by TR in the presence of T(3). Here we utilize the capacity for the Xenopus oocyte to chromatinize exogenous nuclear- injected DNA to analyze the chromatin architecture of the TSH alpha promoter and how this changes upon TR-mediated regulation. Interestingly, in the oocyte, the TSH alpha promoter was positively regulated by T(3). In the inactive state, the promoter contained six loosely positioned nucleosomes. The addition of TR/retinoid X receptor together had no effect on the chromatin structure, but the inclusion of T(3) induced strong positioning of a dinucleosome in the TSH alpha proximal promoter that was bordered by regions that were hypersensitive to cleavage by methidiumpropyl EDTA. We identified a novel thyroid response element that coincided with the proximal hypersensitive region. Furthermore, we examined the consequences of mutations in TR that impaired coactivator recruitment. In a comparison with the Xenopus TR beta A promoter, we found that the effects of these mutations on transactivation and chromatin remodeling were significantly more severe on the TSH alpha promoter.

The molecular mechanism of nuclear hormone receptor-mediated gene regulation and the importance of chromatin structure to this process have been intensively investigated over the past decade. Although the packaging of DNA into dense chromatin is a barrier to transcription factor access, multiple mechanisms exist to overcome this obstacle and thereby facilitate regulation of gene expression (1)(2)(3). Regulation of the acetylation state of core nucleosomal histone proteins influences their interaction with DNA and, subsequently, the nucleosomal packing density and transcription factor accessibility to chromatin. Transcriptional repression by DNA-bound thyroid hormone receptor (TR) 1 in the absence of hormone (T 3 ) involves the recruitment of histone deacetylase-containing complexes that facilitate the formation of repressive chromatin structure. The addition of T 3 causes the release of the deacetylase complexes and stimulates transcriptional activation by the recruitment of coactivators that include acetyltransferase components (1)(2)(3). The acetyltransferases and deacetylases are numerous, but occur in discrete subcomplexes that may exhibit cell type and promoter context dependence (4).
The role of ATP-dependent mechanisms such as SWI/SNF, Mi2/NURD, and ISWI (5) in nuclear receptor-mediated regulation has also been demonstrated. Studies using the glucocorticoid receptor on the mouse mammary tumor virus promoter, which has been shown to have positioned nucleosomes (6), have demonstrated a requirement for SWI-SNF complexes and their ligand-induced targeting to the promoter to activate gene expression (7)(8)(9)(10)(11). More recently, it has been demonstrated that glucocorticoid receptor activation induces nucleosome translational positioning on the mouse mammary tumor virus promoter (12). From these studies and the additional observations that additional cofactors such as the DRIP/ARC complex require a chromatin environment in which to exert their effect on gene expression (13,14), it is clear that chromatin architecture plays a key role in nuclear receptor-mediator gene regulation.
In contrast to the majority of TR-regulated genes, for which T 3 induces up-regulation of promoter activity, the thyroid-stimulating hormone ␣-subunit (TSH␣) promoter is regulated by a negative feedback loop in which unliganded TR activates TSH␣ expression and the addition of T 3 results in repression (15). This negative regulation in response to T 3 presents a conundrum when considering TR action in the context of the mechanisms described above. This raises the question as to what are the mechanistic determinants of positive versus negative transcriptional responses to T 3 .
A recent report detailed a novel mechanism whereby recruitment of deacetylases by unliganded TR is associated, paradoxically, with histone acetylation and activation of transcription, but whereby subsequent overexpression of deacetylase reverses this effect, as does the addition of T 3 (16). It was proposed that the mechanism involves active exchange of repressors and activators between TR and intrinsic promoter regulatory factors. However, the role of chromatin structure and how it is altered in TR-mediated regulation of the TSH␣ promoter have not been investigated.
The Xenopus oocyte has been shown to provide a useful paradigm in which to study the determinants of chromatin assembly and TR-mediated alteration of the chromatin structure, particularly of the Xenopus TR␤A promoter (17)(18)(19)(20). The high capacity for chromatinization of DNA templates injected into oocytes makes this an ideal system in which to study transcription factor-mediated effects on promoter structure and activity. In this study, we sought to compare the effects of TR on the chromatin architecture of the TSH␣ promoter with those of the TR␤A promoter, which is ordinarily up-regulated by T 3 . In addition, mutations in TR identified in individuals with the clinical disorder of Resistance to Thyroid Hormone and shown to be deficient in their capacity to interact with coactivators (21) were used to investigate the role of coactivator recruitment in the modification of chromatin structure.

EXPERIMENTAL PROCEDURES
Reporter Plasmid Constructs and mRNA Synthesis-The reporter plasmid TR␤A-CAT, containing the Xenopus laevis TR␤A promoter linked to the chloramphenicol acetyltransferase reporter gene, has recently been described in detail (22). The reporter construct TSH␣-Luc contains the human TSH␣ proximal promoter (Ϫ846 to ϩ44) linked to the luciferase reporter gene (23), whereas the TRH-Luc reporter construct contains the human thyrotropin-releasing hormone promoter (Ϫ900 to ϩ55) also linked to luciferase (24). For mRNA synthesis, cDNA encoding human TR␤1 was cloned into pT7TS (25) between the X. laevis ␤-globin 5Ј-and 3Ј-untranslated regions. This template was linearized and then transcribed in vitro using T7 polymerase (Ambion Inc.). Mutations were introduced into TR␤1 by site-directed mutagenesis. cDNA encoding the human TR␤2 isoform was cloned into pSP64(A) (Promega), and the linearized template was transcribed in vitro using SP6 polymerase (Ambion Inc.). The X. laevis retinoid X receptor (RXR)-␣ construct has been described previously (17) and was transcribed in vitro using SP6 polymerase.
Xenopus Oocytes and Microinjection-Stage VI Xenopus oocytes were prepared as previously described (26) and stored in MBSH buffer (27) at 18°C for the duration of each experiment (up to 18 h). 0.5-5 ng of each mRNA was microinjected into the cytoplasm. 1 ng of reporter DNA template was injected into the nucleus. Oocytes were then incubated in the presence or absence of 3,3Ј,5-triiodo-L-thyronine (Sigma). Where specified, 33 nM trichostatin A or 50 g/ml cycloheximide was added to the medium. Typically, 20 oocytes were injected for each test sample.
Analysis of Transcription by Primer Extension-RNA was extracted essentially as previously described (17). For transcript quantitation, 3-5 oocyte eq of RNA was annealed with 0.2 pmol of 32 P-end-labeled primer in 30 mM Tris-Cl (pH 8.3), 45 mM KCl, 1.8 mM MgCl 2 , and 3 mM dithiothreitol. The primer used for TR␤A was Primer I, described previously (17). For both TSH␣-Luc and TRH-Luc, primer Luc64 was used, which corresponds to a region in the luciferase proximal gene (5Ј-TGGCGTCTTCCATTTTACCAACAG-3Ј). Endogenous histone H4 was also measured as an internal loading control using primer H4 (5Ј-GAGGCCGGAGATGCGCTTGAC-3Ј). Primer extension analysis of mRNA levels was performed as previously described (17). The specific signal for each reporter transcript was normalized against the histone H4 level.
Analysis of Chromatin Supercoiling-The method used was essentially that described previously (19). Typically, five injected oocytes (1 ng of DNA each) were homogenized in 50 l of 0.25 M Tris-Cl (pH 7.5), followed by the addition of an equal volume of stop buffer (20 mM Tris-HCl (pH 7.5), 30 mM EDTA, 1% SDS, and 0.5 mg/ml proteinase K (Roche Molecular Biochemicals)), and incubated for at least 1 h at 37°C, followed by two phenol/chloroform extractions and then ethanol precipitation. The centrifuged pellet was redissolved in 10 l of Tris/ EDTA buffer containing 100 g/ml RNase A and incubated for 1 h at 37°C. DNA topoisomers were resolved on a 1.2% agarose gel in 1ϫ Tris phosphate/EDTA buffer in the presence of 90 g/ml chloroquine diphosphate (Sigma) for 16 h at 45 V. The gel was then washed for 1-2 h in water to remove chloroquine before performing Southern analysis using a 32 P-labeled random-primed probe with the original plasmid as template. Blots were scanned using a PhosphorImager. added to each oocyte homogenate and incubated at room temperature for 1 or 3 min. The reaction was stopped by the addition of 40 l of 50 mM bathophenanthroline disulfonate solution (Fluka 11890). 100 l of stop buffer (50 mM Tris-Cl (pH 7.5), 50 mM EDTA, and 2.5% SDS) and 10 l of ϳ15 mg/ml proteinase K solution were added, and the mixture was incubated for Ͼ8 h at 37°C, followed by two phenol/chloroform extractions and then one chloroform extraction. The DNA was precipitated by the addition of 0.1 volume of 3 M NaOAc and 0.7 volume of isopropyl alcohol and incubation on ice for Ͼ2 h, followed by microcentrifugation (15 min) and washing with 70% EtOH. The DNA was resuspended in 100 l of 10 mM Tris (pH 8) with 0.2 mg/ml RNase A and incubated at 37°C for 15 min. For both TSH␣-Luc and TR␤A, the DNA was then digested to completion with EcoRI in a 200-l final volume. 1 l of 0.5 M EDTA was added along with 2 l of 15 mg/ml proteinase K solution and incubated for 30 min at 37°C. The DNA was then extracted twice with phenol/chloroform and once with chloroform, followed by EtOH precipitation and washing with 70% EtOH. The pellet was redissolved in 10 l of Tris/EDTA buffer. The DNA was run on a 2% Tris acetate/EDTA gel at 35 V for 12 h and then transferred to a membrane. Southern probing of TSH␣ was performed using a 32 Plabeled random-primed HindIII/EcoRI fragment (615 base pair) from TSH␣-Luc spanning from position ϩ46 of the TSH␣ promoter into the luciferase gene. TR␤A was probed with a 266-base pair XbaI/EcoRI fragment from within the chloramphenicol acetyltransferase reporter gene. Blots were scanned using a PhosphorImager.
Western Blotting-Injected oocytes were homogenized in 10 l/oocyte 0.25 M Tris-Cl (pH 7.5), and the lysate was microcentrifuged for 15 min at 4°C. 0.5 oocyte eq was run on a 10% SDS-polyacrylamide gel and transferred to a membrane. Western analysis of TR␤ expression was performed using a polyclonal antibody directed against Xenopus TR␣, followed by a chemiluminescent secondary antibody (Amersham Pharmacia Biotech).
In Vivo DNase I Footprinting-For each set of conditions, 25 injected oocytes (1 ng of DNA each) were homogenized on ice in 800 l of DNase buffer (20 mM Tris-Cl (pH 7.6), 70 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, and 5% glycerol) and then separated into 5 ϫ 140-l aliquots at room temperature. A dilution series of DNase I (Worthington, DPRFS grade) was added to these aliquots (0.04, 0.2, 1, 6, and 20 units), followed by incubation at room temperature for 3 min. Reactions were stopped by the addition of 150 l of stop buffer (20 mM Tris-Cl (pH 7.4), 0.2 M NaCl, 2 mM EDTA, 2% SDS, and 0.2 mg/ml proteinase K) and incubated for 6 h at 37°C, followed by phenol/chloroform extraction and EtOH precipitation. Linear polymerase chain reaction was performed on the DNA using the 32 P-labeled Luc64 primer described above, and the products were resolved on a urea-polyacrylamide gel and then scanned using a PhosphorImager.

Classical "Negative" Promoters Can Be Up-regulated by T 3 in the Xenopus
Oocyte-The initial analysis of the transcriptional response of the TSH␣ promoter to T 3 in the Xenopus oocyte was performed as a comparative study alongside the well characterized positively regulated Xenopus TR␤A promoter (17)(18)(19)28). Following the microinjection paradigm described in Fig.  1a, we confirmed earlier observations that the Xenopus TR␤A promoter exhibits a high basal transcriptional activity that is repressed by unliganded TR/RXR to ϳ20% of the basal level, but is de-repressed and activated to twice the basal level in the presence of T 3 , giving a 9-fold range in activity (Fig. 1b). The TSH␣ promoter exhibited a much lower basal activity than Xenopus TR␤A but, in contrast to its normal in vivo response, was also repressed by unliganded TR/RXR (40% of the basal level) and stimulated ϳ13-fold in the presence of T 3 , permitting a 33-fold range in activity (Fig. 1b). To ascertain the generality of T 3 -induced activation of a promoter normally repressed by T 3 , we performed identical studies using the TRH promoter, which is also ordinarily down-regulated by T 3 in vivo. Again, we observed a low basal activity with a strong T 3 -dependent stimulation, giving a 33-fold range of activity (Fig. 1b). The positive T 3 response observed with these negative promoters is not a property of the reporter plasmid (pA3LUC) since we have shown these very same plasmid constructs to be down-regulated by T 3 when transfected into mammalian cells (21,29). Furthermore, we tested the keratin promoter K17, on the background of a chloramphenicol acetyltransferase reporter plasmid, which has also been shown to bind TR and to be downregulated by T 3 in mammalian cell culture (30), but found this too to be stimulated by T 3 in the Xenopus oocyte (data not shown). These observations provide strong evidence that the cellular biochemical environment of a promoter is critical in determining both the magnitude and the direction of response to regulatory factors.
To demonstrate that the observed effect of T 3 on the TSH␣ promoter in the oocyte was directly mediated by TR rather than a secondary effect, TR/RXR mRNA and the TSH␣ reporter plasmid were injected into oocytes and incubated for 12 h in the absence of T 3 to permit full translation of the mRNA and the formation of a receptor-bound chromatinized TSH␣ promoter. Prior to the subsequent addition of T 3 , the oocytes were incubated for 4 h in the presence or absence of cycloheximide to eliminate the possibility of T 3 -induced transcription factors influencing promoter activity. As shown in Fig. 1c, cycloheximide had no effect on T 3 stimulation of TSH␣, and it did not affect the level of TR protein, as shown by Western analysis in Fig. 1d. This indicates that regulation of the TSH␣ promoter by liganded TR in the Xenopus oocyte is direct. This would be anticipated because the oocyte genome is tetraploid, and the capacity to generate adequate transcripts to cause secondary effects is very limited.
The Xenopus TR␤A, TSH␣, and TRH Promoters Exhibit a Differential Response to the Histone Deacetylase Inhibitor Trichostatin A-From the data in Fig. 1b and earlier work (17), it is apparent that in the case of the TR␤A promoter, repression of basal transcription by unliganded TR accounts for a large proportion (ϳ50%) of the observed transcriptional control in oocytes. As discussed earlier, many studies have linked repression with the recruitment of histone deacetylase activity (31). In Fig. 1e, we examined the relative contributions of acetylation on each of the TR␤A, TSH␣ and TRH promoters by incubating the oocytes in the presence or absence of the deacetylase inhibitor trichostatin A (TSA). We found that for the TR␤A promoter, maximal activation could be achieved by TSA alone, irrespective of the presence of TR, as seen previously (28), and that this activity was not further enhanced by the addition of T 3 . This indicates that the acetylation state of the TR␤A promoter is a key factor in its regulation. However, for both the TSH␣ and TRH promoters, TSA gave only partial activation, as did ligand-bound TR. Maximal activity was seen only with the combination of both T 3 and TSA. This indicates that mechanisms other than those involving histone acetylation, e.g. ATPdependent regulators, may play a relatively greater role on FIG. 1. Transcriptional regulation of negative and positive promoters by the thyroid hormone receptor in the Xenopus oocyte. a, schematic showing the paradigm for generation and regulation of a chromatin template in the Xenopus oocyte. b, transcriptional regulation of Xenopus TR␤A (xTR␤A), human TSH␣ (hTSH␣), and human TRH (hTRH) promoters by TR. Oocytes were either uninjected (Ϫ) or injected (ϩ) with 0.5 ng each of TR␤ and RXR␣ mRNAs. After a 4-h incubation, 1 ng of the appropriate reporter plasmid (pTR␤A, TSH␣-Luc, or TRH-Luc) was injected into each nucleus. The oocytes were cultured for a further 12 h in the presence (ϩ) or absence (Ϫ) of 100 nM T 3 , following which total RNA was extracted, and the message levels of each reporter as well as endogenous histone H4 were assayed by primer extension. The level of each reporter message (Txn) was normalized against that of histone H4 (H4). In each case, the transcriptional activity is reported relative (Rel.) to the basal activity for each promoter, i.e. in the absence of both T 3 and injected mRNA. c, effect of cycloheximide on TR-mediated activation of TSH␣. Oocytes were injected with mRNA and TSH␣-Luc DNA as described for b, but without the addition of T 3 . 12 h after the DNA injection, 50 g/ml cycloheximide (CHX) was added where appropriate, and the oocytes were incubated a further 4 h to permit cycloheximide-mediated inhibition of translation. T 3 was then added to activate TR-mediated transcription, followed by a further 4-h incubation, after which RNA was extracted and assayed for TSH␣-Luc activity. d, analysis of the level of TR␤ protein present after the treatment with cycloheximide in the experiment described for c. The lysate from the oocytes used for transcription analysis in c was resolved on an SDS-polyacrylamide gel and subjected to Western blot analysis using an antibody to TR. ns, nonspecific band. e, effect of TSA on promoter activity. Oocytes were treated as described for b, except that TSA (33 M) was added, as appropriate, immediately after DNA injection. TSH␣ and TRH than they do on TR␤A.
Identification of a Novel T 3 Response Element in the TSH␣ Promoter-We have confirmed a direct effect of TR on regulation of the TSH␣ promoter (Fig. 1c). However, to date, no definitive thyroid response elements (TREs) have been reported in this promoter, either in a chromatin context or on naked DNA. To investigate this issue in an in vivo configuration, we injected the TSH␣ promoter into oocytes in the presence or absence of TR/RXR, with or without T 3 , and performed in vivo DNase I footprinting on the chromatinized DNA. As illustrated in Fig. 2a, the presence of TR/RXR protected a region of chromatinized TSH␣ promoter between positions Ϫ200 and Ϫ240. Furthermore, this footprint was retained upon the addition of T 3 , as might be expected for ligand-bound receptor to activate transcription in a direct manner and in accordance with earlier observations on the TR␤A promoter (32). We analyzed the sequence of the TSH␣ promoter that was protected by TR/RXR and found that, within a 23-base pair stretch, it contained one perfect consensus half-site, AGGTCA (site A), and two degenerate half-sites (B and C). All three half-sites are arranged in a direct repeat orientation. Interestingly, the half-site spacing is unusual in that sites A and B are spaced by 5 base pairs, an arrangement more typical of a retinoic acid response element, whereas there is no spacing between sites B and C (33).
To demonstrate the functional relevance of this putative TRE, we mutated simultaneously the first of the two guanine residues (shown in boldface in Fig. 2a) in each half-site to adenine. Previous studies have shown that this residue is highly conserved in nuclear receptor-binding sites (33). We then examined the transcriptional activity of this promoter compared with that of the wild-type promoter (Fig. 2b) and found that the triple mutation lowered T 3 -induced activation to ϳ50% of the wild-type level, supporting the notion that this region represents a functional TRE. To further demonstrate the role of this putative TRE in receptor binding to the TSH␣ promoter, we performed competitive band shift analysis. Highly purified recombinant TR and RXR were bound to duplex oligonucleotides containing an authentic high affinity direct repeat TRE from the malic enzyme gene promoter (34). DNA comprising a 260-base pair fragment from the wild-type or mutant promoter including the putative novel TRE was used as an unlabeled competitor. Fig. 2c shows that the mutant promoter, TSH(⌬TRE), was notably impaired in its ability to disrupt the receptor-probe complex, even at a 450-fold mole excess (lanes 2 and 6 -8), whereas the wild-type fragment was an effective competitor at the high concentration (lanes 2 and 3-5). Quantitation of the receptor-probe complexes (Fig. 2d) reveal that, at the high concentration of competitor, wild-type TSH displaced 51% of the bound probe, whereas TSH(⌬TRE) displaced only 23%, supporting this region as one playing a role in TR binding.
T 3 Induces Alteration in the Chromatin Architecture of the TSH␣ Promoter-We examined the receptor-mediated effects of T 3 on the chromatin architecture of the TSH␣ promoter under both repressed and active transcriptional states. In Fig.  3a, we utilized the susceptibility of chromatinized DNA to chemical cleavage by MPE to map nucleosome positions and to examine the T 3 -induced changes. In the absence of receptor (lanes 1 and 2), a diffuse banding pattern was observed with no apparent nucleosome positioning. However, this was not due to a simple lack of chromatinization, as Fig. 2c shows that topoisomers were still generated, either in the absence or presence of TR, indicating effective chromatinization of exogenous TSH␣ promoter in the Xenopus oocyte. The presence of additional diffuse bands in the middle of nucleosomes A, C, and D in lanes 1-4 suggests that the positioning preference for those nucleosomes in the basal state is relatively weak and that the nucleosomes exist in more than one position. In the presence of unliganded TR (Fig. 3a, lanes 3 and 4), no significant change in

FIG. 2. Identification and characterization of a novel thyroid hormone response element in the TSH␣ promoter. a, in vivo
DNase I footprint of the TSH␣ promoter by TR. Oocytes were left uninjected or were injected with 5 ng each of TR and RXR mRNAs, followed 4 h later by 1 ng of TSH␣-Luc DNA, and incubated for 12 h in the presence or absence of 100 nM T 3 . Oocytes were then collected and subjected to DNase I treatment as described under "Experimental Procedures." In addition, naked DNA (Naked) was also treated with DNase I. A sequencing ladder of TSH␣-Luc was run on the same gel to determine nucleotide positions (not shown). Solid and dashed arrows denote the respective positions of perfect and degenerate consensus recognition half-sites for TR binding that lie within the footprinted region of the TSH␣ promoter. Only the antisense strand of the TSH␣ promoter is shown. The boldface G nucleotides in half-sites A-C were those mutated to adenine in the mutant TSH␣ promoter TSH(⌬TRE). b, effect of mutation of the putative thyroid response element in the TSH␣ promoter. Each guanine residue highlighted in a was mutated to adenine to generate TSH(⌬TRE), a mutant promoter devoid of the putative thyroid response element identified in a. Oocytes were left uninjected or were injected with 0.5 ng each of TR␤ and RXR␣ mRNAs, followed 4 h later by 1 ng of wild-type TSH␣-Luc (TSH(wt)) or the TSH(⌬TRE) mutant, and incubated for 12 h in the presence or absence of 100 nM T 3 . Total RNA was then extracted and analyzed for promoter activity by primer extension. Shown is a sample data set below a graph representing the mean Ϯ S.E. of three replicate analyses. c, gel shift analysis of the ability of the wild-type TSH␣ and TSH(⌬TRE) promoters to compete with an authentic thyroid response element for binding to TR/RXR. Equal amounts of purified recombinant TR and RXR were incubated with 20 fmol of a labeled duplex oligonucleotide containing the DRϩ4 thyroid response element from the malic enzyme gene promoter. In addition, increasing amounts of an unlabeled 250-base pair fragment of the wild-type TSH␣ or TSH(⌬TRE) promoter, encompassing the putative TRE, used as an unlabeled competitor for DNA binding, were added; and the reactions were run on native polyacrylamide gel. Lane 1 contains only labeled probe. Lane 2 contains only TR/RXR and labeled probe. Lanes 3-5 and 6 -8 contain TR/RXR, probe, and increasing levels of the respective unlabeled competitors. The arrowhead denotes the specific TR/RXR-probe complex. d, quantitation of the raw data shown in c. The intensity of the TR/RXR-probe complex in each lane was quantitated using a PhosphorImager and expressed as a percentage of the labeled input probe. This is plotted against the mole ratio of unlabeled competitor to labeled probe. Txn, reporter message. chromatin structure was detected when compared with lanes 1 and 2, indicating that the relatively small repression of transcriptional activity imparted by unliganded receptor occurs without major changes in chromatin structure. This lack of chromatin structural change in the presence of unliganded TR/RXR is supported by the lack of change to promoter supercoiling shown in Fig. 3c, indicating that the overall nucleosome density is not altered.
However, upon the addition of ligand, a dramatic remodeling of the chromatin architecture was observed (Fig. 3a, lanes 5  and 6). First, the generation of a strongly positioned dinucleosome (C ϩ D) occurred between positions Ϫ220 and Ϫ570. With the exception of the band at position Ϫ390, which presumably represents the linker region between the two nucleosomes, all other bands seen in lanes 1-4 in this region disappeared, indicating that this dinucleosome is acutely positioned only in the transcriptionally active state. Second, nucleosome A was also repositioned upon activation, as demonstrated by the disappearance of the mid-nucleosomal band at position ϩ40 in the extreme 3Ј end of this TSH␣ promoter fragment that encompasses the transcription start site. Third, there was a dramatic manifestation of MPE hypersensitivity at positions Ϫ220 and Ϫ570 (either side of the positioned C ϩ D dinucleosome) and, to a lesser degree, at position Ϫ60 in the vicinity of the TATA element located from positions Ϫ23 to Ϫ29. These ligand-induced changes in the chromatin architecture of the TSH␣ pro-moter as revealed by MPE accessibility are in keeping with the T 3 -induced change in supercoiling shown in Fig. 3c (lane 4), which reflects a decrease in nucleosome density. It should be noted that since these analyses were performed in the presence of ␣-amanitin, which inhibits RNA polymerase II action, the structural changes are occurring independent of transcription.
Comparison of the TSH␣ promoter with the TR␤A promoter revealed a notable difference in the effects conferred on chromatin structure by liganded TR (Fig. 3b). In contrast to the TSH␣ promoter, TR␤A exhibited the clearly defined periodicity indicative of organized nucleosomal packaging seen previously (19), both in the absence and presence of unliganded receptor (lanes 1-4). Both promoters exhibited induced hypersensitivity to MPE upon the addition of T 3 (see arrow in Fig. 3b). For TR␤A, this induced hypersensitive region disrupted an existing nucleosome between two TREs (compare lanes 3 and 4 with  lanes 5 and 6). This is in agreement with the ligand-induced change in supercoiling observed in Fig. 3c (lane 8). However, unlike with the TSH␣ promoter, the presence of T 3 did not appear to stabilize the position of other nucleosomes, suggesting a differing requirement for structural reorganization of chromatin upon activation of these two promoters in the oocyte.
Natural TR Mutants Are Differentially Impaired in Their Capacity to Regulate the TSH␣ and TR␤A Promoters-The wild-type receptor and the mutants used in this study are all the human TR␤1 isoform. Another isoform, TR␤2, differs from TR␤1 at the amino terminus and is expressed primarily in the pituitary, where it is believed to be a major regulator of TSH␣ (35,36). Given the unexpected up-regulation of TSH␣ by TR␤1 in the oocyte, we sought to ascertain whether this result was isoform-dependent. We expressed TR␤2 in the oocyte (Fig. 4a) and compared its transactivation capacity with that of TR␤1 (Fig. 4b). We found no significant difference between the activities of these two isoforms in this system.
We have previously characterized the naturally occurring TR␤ mutations L454V and L454W identified in individuals with resistance to T 3 (21), a dominantly inherited clinical dis-order characterized by elevated levels of circulating thyroid hormone, but inappropriately normal TSH levels, as well as variations in goiter, attention-deficit hyperactivity disorder, reduced IQ, and growth retardation. These mutant receptors are impaired both in their T 3 -dependent transactivation function and their ability to recruit coactivators such as steroid receptor coactivator-1, yet bind T 3 with near wild-type affinity (21). 2 Since recruitment of coactivators is believed to play an integral role in the regulation of chromatin structure, we utilized the Xenopus oocyte system to examine the influence of these mutations on transcriptional activation and chromatin remodeling of both the TSH␣ and TR␤A promoters.
To confirm the expression of the TR mutants in the oocyte, we performed Western blot analysis of oocyte extracts using an antibody to Xenopus TR␤. Fig. 4a shows that all three mutants were expressed to a level similar to that of the wild-type receptor. We next examined the ability of each mutant to activate the TR␤A and TSH␣ promoters. Fig. 4c shows that on the TR␤A promoter, each mutant TR was able to repress basal transcription in the absence of ligand at least as well as wildtype TR (10 -20% of the basal level). In the presence of saturating levels of T 3 , the more mildly affected natural mutant, L454V, was able to achieve 75% of the wild-type activity, whereas the severely affected mutant, L454W, did not activate above basal levels (ϳ38% of the activated wild-type level). Furthermore, the artificial mutant, L454A, which has previously been shown to be inactive in mammalian cells and devoid of coactivator binding (21,37), was even incapable of fully releasing basal repression, reaching only ϳ14% of the wild-type maximum. However, when tested on the TSH␣ promoter, the phenotype of these mutants was notably more severe, with L454V achieving only 12% of the wild-type activity, whereas both L454W and L454A showed Ͻ4% of the wild-type level. This marked difference between mutant TR responses on the two promoters suggests key promoter-dependent differences in the nature of TR-mediated regulation.
Changes in Chromatin Supercoiling Are Not Sufficient for TR-mediated Activation-To examine the capacity for the TR mutants to remodel chromatin on the TSH␣ and TR␤A promoters, we used a DNA supercoiling assay to assess ligand-induced changes in DNA supercoiling. In this assay, a loss of supercoiling upon the addition of T 3 is represented by a general downshift in the distribution of topoisomer bands and, in the strong cases, the appearance of a cluster of diffuse bands toward the bottom of the gel. Note that the minor variation in distribution of topoisomers between each mutant in the absence of T 3 is not considered significant. The data in Fig. 5a show that, in the presence of wild-type TR, T 3 induced a loss of supercoiling in the TR␤A promoter, confirming an earlier report from this laboratory (19). The L454V mutant also exhibited a wild-type level of change in supercoiling, in keeping with its capacity for transactivation on this promoter. Furthermore, the supercoiling changes observed with L454W and L454A were moderate and poor, respectively, again correlating with the extent of T 3 -induced activation of transcription (Fig. 4c). However, for the TSH␣ promoter, the correlation between supercoiling and transcription did not hold. Although the transcriptionally inactive L454W and L454A mutants were completely incapable of inducing significant changes in DNA supercoiling (Fig. 5b), the L454V mutant, which retains a low level of activity on this promoter, elicited a change in supercoiling that was similar to that of the wild-type reporter. This suggests that although a change in chromatin supercoiling may be a prerequisite for transactivation, it is not, by itself, sufficient and is in accord FIG. 4. Transcriptional regulation of the Xenopus TR␤A and human TSH␣ promoters by TR␤ mutants. a, comparative expression levels of TR␤ mutants in Xenopus oocytes. Oocytes were left uninjected (Nil) or were injected with 0.5 ng of mRNA for wild-type TR␤1 (WT); the TR␤1 point mutant L454V, L454W, or L454A; or the TR␤2 isoform. After a 12-h incubation, oocyte extracts were subjected to Western analysis using an antibody to TR. In each case, the upper band represents the full-length receptor. b, comparison of transactivation of the TSH␣ promoter by the TR␤1 and TR␤2 isoforms. Oocytes were left uninjected or were injected with 0.5 ng each of RXR␣ and TR␤1 or TR␤2 mRNAs, followed after 4 h by 1 ng of TSH␣-Luc DNA, and incubated for 12 h in the presence or absence of 100 nM T 3 . TSH␣ promoter activity was assayed by primer extension. c, regulation of TR␤A and TSH␣ promoter activities by TR␤1 mutants. Oocytes were left uninjected or were injected with 0.5 ng each of RXR␣ and wild-type or mutant TR␤1 mRNAs, followed after 4 h by 1 ng of pTR␤A or TSH␣-Luc DNA, and incubated for 12 h in the presence or absence of 1000 nM T 3 . The higher than usual concentration of T 3 was to ensure that the small reduction in T 3 binding affinity for the mutants did not account for observed functional differences. Promoter activity was assayed by primer extension. Activity is reported relative (Rel.) to that of each promoter in the absence of receptor and T 3 . Txn, reporter message; xTR␤A, Xenopus TR␤A; hTSH␣, human TSH␣; H4, internal control message.
with earlier observations on the TR␤A promoter (19). Furthermore, for both activation of transcription (Fig. 4c) and topological change, the extent to which each mutation affects receptor function is promoter type-dependent. DISCUSSION In this study, we have shown that (i) the TSH␣ promoter is transcriptionally activated in Xenopus oocytes; (ii) T 3 induces major changes in the translational positioning of nucleosomes in the TSH␣ promoter; (iii) changes in chromatin architecture are not sufficient for T 3 -mediated gene activation; and (iv) unlike the TR␤A promoter, activation of the TSH␣ promoter cannot be fully accounted for by relief of deacetylase-mediated repression.
In a surprising observation, we found that in the Xenopus oocyte, the TSH␣ promoter was repressed by unliganded TR, but activated upon the addition of T 3 (Fig. 1b), contrary to its perceived mode of negative regulation in vivo in the thyrotroph cells of the anterior pituitary (15). Furthermore, this effect was not specific to TSH␣ or genes active in the pituitary since the promoters for the hypothalamic TRH (Fig. 1b) and keratinocyte-specific keratin 17 (data not shown) (30) genes that are ordinarily down-regulated by T 3 also demonstrated this reversal of T 3 response. The most likely explanation for this would be the coexistence of specific regulatory factors that are not conserved between mammalian cells and Xenopus oocytes. The promoter region of the TSH␣ gene contains many regulatory elements central to its expression both in the pituitary and in the placenta that have demonstrated cell-type specificity for their usage (38 -47). The concerted action of these regulatory factors in the correct combination may be the key determinant of the nature of the transcriptional response on the TSH␣ promoter, with TR itself functioning largely as the switch. This suggests that the mechanism of nuclear receptor-mediated control of mammalian gene expression involves receptors functioning as integrators of a regulatory pathway that is predetermined by the concerted action of specific promoter-bound transcription factors. To that end, the receptor-mediated changes in chromatin architecture seen here on the TSH␣ promoter may be instrumental in facilitating the coordinate DNA binding and activity of such factors.
That TR regulates the TSH␣ promoter directly seems clear given our observation that T 3 induced activity in the presence of the protein synthesis inhibitor cycloheximide. Historically, the precise nature of the TRE has been difficult to define, with one report suggesting a region near the transcription start site that resembles a degenerate palindromic TRE (23). Although the element identified in the present study appears to confer T 3 responsiveness in the oocyte, it does not account for the entire T 3 response. Two possibilities could explain this. The first is the existence of other TREs. In addition to that described above (23), the observed proximity of the TREs to the MPE-hypersensitive sites in both the TSH␣ and TR␤A promoters (Fig. 3, a  and b) suggests that the other strongly T 3 -induced MPE-hypersensitive site in TSH␣, between nucleosomes D and E, points to another region of TR binding. The second possibility is that TR may regulate the TSH␣ promoter through mechanisms other than direct DNA binding, as recently suggested (48). The differing importance of activation versus repression on these promoters shown in Fig. 1 (b and c) is likely due to differences in promoter structure and utilization of regulatory factors and mechanisms. The data in Fig. 1e suggest that acetylation is the major effector of activation of the TR␤A promoter and that, in the absence of deacetylase activity, nontargeted acetyltransferases may acetylate this promoter sufficiently to facilitate maximal activation. In contrast, on the TSH␣ and TRH promoters, full activation required the concerted action of both TSA and T 3 -activated TR. This suggests either that additional acetylation of these promoters, beyond that achieved by TSA alone, requires targeted acetylase recruitment or that other mechanisms in addition to acetylation are important. The latter scenario has precedent in other nuclear receptor studies. Glucocorticoid receptor-mediated remodeling of a reconstituted mouse mammary tumor virus nucleosomal array has been shown to require ATP and remodeling factors, as well as interaction with acetyltransferase coactivators, supporting the idea of distinct requirements for each of these two remodeling mechanisms (10,11). The estrogen and retinoic acid receptors also have been shown to require ATP-dependent remodeling complexes in the chromatin context (49,50).
Both the TSH␣ and TR␤A promoters exhibited T 3 -induced MPE hypersensitivity around the TR-binding sites, but TSH␣ exhibited greater T 3 -dependent changes in nucleosome translational positioning (Fig. 3). Chromatin remodeling is likely to facilitate transcription factor access and formation of a transcriptionally permissive state. These additional structural changes in the TSH␣ promoter that are not seen with TR␤A may account for the greater potential for activation of transcription. Such highly organized chromatin structures have been previously identified in the regulatory regions of many inducible genes (51)(52)(53)(54)(55)(56), with both rotational positioning of DNA on the nucleosomes as well as translational positioning of the nucleosomes along the DNA being important (12,(57)(58)(59). Changes in chromatin structure appear to be generally confined to regulatory regions (60), and in vivo footprinting studies on the mouse mammary tumor virus promoter have shown that transcription factors such as nuclear factor-1 and the transcription factor IID complex do not bind unless hormone-induced chromatin remodeling mediated by the glucocorticoid receptor has occurred (61,62). It is likely that the observed remodeling of the TSH␣ promoter also facilitates its binding to transcription factors.
Work from our laboratory has previously shown that the TR␤A promoter in Xenopus oocytes can be fully activated by acetylation alone and that TR-dependent chromatin disruption is not required for transcriptional activation by the deacetylase inhibitor TSA (28). It was suggested that this observation would be anticipated if histone acetylation was the only alter- FIG. 5. Topological changes induced in TSH␣ and TR␤A chromatin structure by wild-type and mutant TRs. Oocytes were left uninjected or were injected with 5 ng each of RXR␣ and wild-type (WT) or mutant TR mRNAs, followed after 4 h by 1 ng of pTR␤A (a) or TSH␣-Luc (b) DNA, and incubated for 12 h in the presence or absence of 1000 nM T 3 . The higher than usual concentration of T 3 was to ensure that the small reduction in T 3 binding affinity for the mutants did not account for observed functional differences. Oocytes were harvested, and the promoter DNA was analyzed by the supercoiling assay as described under "Experimental Procedures." Input lanes contain uninjected supercoiled plasmid DNA. White dots denote the positions of the most prevalent topoisomers for each condition as determined by visual assessment of band intensity. xTR␤A, Xenopus TR␤A; hTSH␣, human TSH␣. ation to chromatin structure necessary for transcriptional activation of this promoter. In line with this idea, we found that the more severe TR mutants, L454W and L454A, not only were incapable of activating the TR␤A promoter to the level attained by wild-type TR, but also appeared unable to fully release repression of basal promoter activity. One explanation for this is that these mutant receptors are impaired in their ability to release corepressors in response to T 3 . In support of this notion is the observation that a similar natural mutant of TR␤ involving the same residue, L454S, exhibits both a stronger interaction with the corepressor nuclear receptor corepressor than the wild-type receptor and is markedly impaired in T 3 -induced corepressor release, despite preservation of high affinity T 3 binding (63). These results further support a major role for the acetylation state in TR-mediated regulation of the TR␤A promoter.
The greater effect of receptor mutations on activation of the TSH␣ promoter compared with TR␤A (Fig. 4c) is intriguing. On TR␤A, the TRE configurations are typically those of a direct repeat with a 4-base pair spacing (DRϩ4), to which a TR/RXR heterodimer is known to bind well (64). However, the nature of the TRE in the TSH␣ promoter and of TR binding is unclear, but does not appear to involve a normal DRϩ4 element. It is plausible that the configuration of the receptor when bound to the TSH␣ promoter (cf. TR␤A) may exacerbate the effect of these mutations. Such response element configuration dependence for mutational effects on DNA binding by TR and homo-or heterodimerization with RXR has previously been reported (29).
Finally, we found that the degree of T 3 -induced changes in supercoiling of the TR␤A promoter in the presence of the TR mutants reflects the effects of the mutants seen upon activation of transcription. Specifically, the greater the change in supercoiling, the greater is the level of activation (Fig. 5a). That this generality did not hold for the TSH␣ promoter is intriguing and suggests that although a change in supercoiling may be necessary for activation of the TSH␣ promoter, it is not, by itself, sufficient.
In summary, this study indicates fundamental differences in the mechanisms by which TR regulates expression of the TR␤A and TSH␣ promoters in the Xenopus oocyte. We suggest that for TR␤A, gross chromatin remodeling is not required and that changes in the acetylation state alone can confer full regulatory response. In contrast, activation of the TSH␣ promoter requires other mechanisms in addition to acetylation, and these mechanisms induce the necessary changes in chromatin architecture for transcriptional activation. Further studies will be required to determine the absolute need for these structural changes as well as to determine the factors required for this change.