Different Configurations of Specific Thyroid Hormone Response Elements Mediate Opposite Effects of Thyroid Hormone and GC-1 on Gene Expressi
http://www.100md.com
《内分泌学杂志》
Division of Endocrinology and Metabolism (B.G., E.A.S., W.H.D.), University of California, San Diego, La Jolla, California 92093
Department of Physiology and Biophysics (G.G.) and Department of Cell and Developmental Biology (A.S.M.), ICB University of Sao Paulo, 05508-900 Sao Paulo, Brazil
Departments of Pharmaceutical Chemistry and Cellular and Molecular Pharmacology (G.C., T.S., J.D.B.), University of California, San Francisco, San Francisco, California 94143
Abstract
T3 regulates transcription of the rat sarcoendoplasmic reticulum calcium ATPase in the heart. The T3 effect is mediated by three differently configured T3 response elements (TREs). Here we report the mutation of each individual TRE in the promoter and the contribution of each TRE on gene expression. Mutation of TRE1, a direct repeat element, exerted the strongest T3 response, compared with TRE2 and TRE3, which are inverted palindromes. The isolated TRE2 and TRE3, which showed no response (TRE2) or were weakly positive with T3 (TRE3), became strong negative regulatory elements with the T3 analog GC-1. We found that TRE1 recruits corepressor complexes containing nuclear receptor corepressor and histone deacetylase 3 in the absence of ligand, and steroid receptor coactivator-1-containing coactivator complexes with both T3 and GC-1. TRE3 bound the same corepressor complexes without ligand but showed only a weak association with steroid receptor coactivator-1 with T3 and a strong association with corepressor complexes with GC-1. Thus, GC-1 appears to control cofactor association differentially on these two sarcoendoplasmic reticulum calcium ATPase TREs, which could be the mechanism of ligand-dependent transcriptional activation and repression observed with the isolated TRE1 and TRE3 elements. Because the x-ray crystal structures of GC-1 and T3 complexed with the TR ligand binding domain are superimposable, the results imply that GC-1 and T3 induce differential effects on the receptor that are not evident in the static structures but must occur in the dynamic setting of receptor function. These results have implications for selective modulation of receptor function by agonist ligands.
Introduction
THYROID HORMONE RECEPTORS (TRs) are members of the nuclear receptor superfamily that includes receptors for steroid hormones, peroxisomal proliferator-activated receptors, retinoids, and many other ligands. Two TR isoforms, TR and TR mediate T3 signaling in a variety of organs including the heart. The contractile function of the heart is markedly influenced by alterations of the thyroid status (1, 2). In particular, a delay in diastolic relaxation occurs with hypothyroidism, and this change can be linked to a delay in the calcium transient during the contractile cycle. One of the important mediators of calcium flux is the activity of the calcium ATPase of the sarcoplasmic reticulum (SERCa2), and the expression of this gene in cardiac myocytes is markedly influenced by thyroid hormone (T3) levels (3). We previously analyzed the enhancer and promoter region of the SERCa2 gene and identified three separate thyroid hormone response elements (TREs) in the region extending 500 nucleotides upstream from the transcription start (4). TRE1 is a direct repeat spaced by four nucleotides (DR4) and is localized between –481 and –458. TRE2 is an inverted palindrome, also termed everted repeat element, spaced by four nucleotides between –310 and –289. TRE3 is also an inverted palindrome but is spaced by six nucleotides and is localized between –218 and –195. It is currently unclear what functional contribution the three differently configured TREs make as individual TREs to the T3-mediated alteration in SERCa2 transcription and in which way they may interact with each other. It is of interest to determine how much each individual TRE contributes both to the transcriptional activation on ligand binding to the TR and transcriptional suppression or silencing by unliganded TR. The contribution that the individual TREs in the SERCa2 promoter make to these functions is further explored in this report.
Two separate genes encode TRs, and in addition, there are several splice variants (5, 6, 7). One of the genes codes for TR1, which binds T3 and a separate splice variant for the non-T3 binding TR2 isoform (8). The second gene encodes the T3 binding variants TR1, TR2, and TR3 (9, 10, 11). TRs have a modular design, which has been well characterized. The carboxy-terminal ligand binding domain undergoes T3-induced conformational changes and interacts with coactivators and corepressors (12).
Structural analysis of the ligand binding pocket of the TR1 and TR1 has demonstrated that the T3 ligand is an intrinsic part of the receptor structure largely buried inside the protein fold (13). It appears therefore likely that the binding of a modified T3 analog may have an influence on the configuration of the T3 receptor surface, especially in the regions in which coactivators and corepressors bind. The most noted examples of this are the selective estrogen receptor modulators such as tamoxifen and raloxifene (14). Recently novel thyroid hormone analogs like the compound GC-1 have been synthesized, and it has been demonstrated that administration of GC-1 to hypothyroid and euthyroid animals has effects that are distinct from those of T3 (15, 16).
Although it is difficult to predict a higher-order structure of chromatin in the region of these differently configured TREs in the SERCa2 promoter, individual mutation has led to some insight into which role each TRE may play. For example, binding on a direct repeat spaced by four nucleotides favors TR/retinoid X receptor (RXR) heterodimer binding with the RXR binding to the 5' half-site (17). In contrast, binding to inverted palindromes prefers homodimeric T3 receptor configurations, and on some of these elements, the T3 ligand leads to transcriptional repression (18). This half-site-dependent characteristic of each TRE may be enhanced by binding of a modified ligand. Therefore, we compared the transactivation behavior of the novel T3 analog GC-1 and T3 with TRs bound to the DR4 vs. inverted palindromic TREs of the SERCa2 promoter. We noticed significant differences in T3 vs. GC-1 transactivation activity with TR bound to the differently configured TREs, which was independent of the TR isoform that was cotransfected. Using chromatin immunoprecipitation (ChIP) assays, we studied cofactor association on the isolated TRE3 and TRE1 within the SERCa2 promoter. Our results may provide a mechanism for the differential transactivation activity observed with T3 and GC-1 on these elements.
Materials and Methods
Experimental animals
Experimental animal procedures were reviewed and approved by the Animal Subjects Program at the University of California, San Diego.
Cloning of the rat SERCa2 promoter up to position –3260 bp
The 3.2rSERCa2-luciferase construct was engineered by first digesting an insert (–3260 to +74 SERCa2 promoter) from 3.2pBLCAT3 (19) with HindIII. This fragment was subsequently blunt ended, digested with XhoI, and cloned into a luciferase-containing vector (pGL2, Promega, Madison, WI) previously digested with MluI, blunted with Klenow-polymerase, and digested with XhoI. The insert and the vector were ligated by using standard cloning methods. The 0.6rSERCa2-luciferase was constructed by digesting an insert (–560 to +74 SERCa2 promoter) from 0.6pBLCAT3 (19) also with HindIII. This insert was treated the same way as above and ligated to the vector pGL2 also digested with MluI, blunt ended with Klenow-polymerase, and digested with XhoI.
Mutation of the individual TREs in the rat SERCa2 promoter
The two guanines present in all TRE half-sites have been shown to be essential for receptor/DNA binding (20), independently of the ligand, and their substitution by adenosines completely abrogates receptor/DNA binding. Consequently we performed the same substitutions in all SERCa2 TREs to determine the individual contributions of those elements in SERCa2 promoter responsiveness to T3/GC-1. This was achieved by introducing point mutations by a Kunkel-based method. In brief, the SERCa2 promoter fragment encompassing +74 to –561 bp was inserted into the multiple cloning site of a M13 vector, which was used to transform a bacterial host strain deficient for deoxyuridine triphosphatase and uracil-N-glycosylase, which results in occasional substitutions of uracil for thymine in newly synthesized DNA. Subsequently the substitution of uracil for thymine in newly synthesized DNA was annealed to a set of primers containing the desired mutations, targeting simultaneously both TRE half-sites (TRE1 = primer 1; TRE2 = primer 2; TRE3 = primer 3; see Table 1 and Fig. 1 for sequence information). After annealing, the second strand was synthesized by T7 DNA polymerase and T4 DNA ligase. Finally, the double-stranded DNA was used to transform an uracil-N-glycosylase+ bacterial strain, which eliminates the wild-type strand. By using this method it was possible to generate SERCa2 promoters presenting TREs 1, 2, and 3 individually mutated. The different SERCa2 promoter TRE mutant combinations in the same promoter were accomplished by standard cloning procedures, using unique endonuclease restriction sites, flanking the three TREs (HindIII at –559 bp, ApaI at –395 bp and SmaI at –278 bp).
Construction of the luciferase reporter plasmids
The constructs in which the rat SERCa2 promoter was fused to the HSV-thymidine kinase (tk) promoter have been cloned using the pT81luc vector described by Nordeen (21). A HindIII/NaeI fragment from position –562 to position –163 (NaeI site) was cloned into the pT81luc vector cut with HindIII and SmaI. The SERCa2 promoter fragments in which individual or pairs of the TREs were mutated were cloned in the exact same way into pT81luc. Plasmids were prepared as Maxipreps (QIAGEN, Valencia, CA) following the standard protocol supplied by the manufacturer and were subsequently verified by sequencing using primer 8 (Table 1) within the luciferase gene. The plasmid T4T was constructed by cloning the annealed primers 9 and 10 (Table 1) into a plasmid that contained a minimal promoter of the human -retinoic acid receptor gene including the TATA-box upstream of the luciferase gene (21). Construction of the individual TREs upstream of the SERCa2 minimal promoter was done using the plasmid pGL2 basic as a vector backbone. The rat SERCa2 promoter from position –38 to +32 (19) was cloned into pGL2 basic, and the three SERCa2 TREs were cloned upstream of this promoter as annealed oligonucleotides with the following primers from Table 1: TRE1, primer 11/primer 12; TRE2, primer 13/primer 14; and TRE3, primer 15/primer 16. To generate the TRE swap mutants, we used a bidirectional PCR strategy in which an oligo was annealed to the sequences immediately upstream of the 5' TRE half-site and that had a mismatching extension reaching to the middle of the spacer of the TRE to be changed. Another oligo was annealed immediately downstream of the 3' half-site and extended also to the middle of the spacer of the TRE to be changed. As a template, the –0.6rSERCa2 tk-Luc circular plasmid was used with Pfu polymerase (Stratagene, La Jolla, CA). After PCR, DpnI treatment was used to select against parental DNA molecules, and the new strands, containing the swapped TREs, were ligated with T4-DNA ligase. The recircularized vector DNA incorporating the desired mutations was transformed into Escherichia coli. To change TRE1 to TRE3, we used a sense oligo primer 4 and antisense oligo primer 5 (Table 1). To change TRE3 to TRE1, we used as sense oligo primer 6 and as antisense primer 7. The mismatched bases are shown in Table 1 (lower case).
Other plasmids
Expression plasmids for transient transfections were all prepared by standard techniques using a QIAGEN Maxiprep kit. The rat cDNAs for TR1, TR1, and human RXR were cloned into the expression plasmid pCMX.
Transient transfection of neonatal myocytes
Neonatal myocytes from 1- to 2-d-old rats were prepared as described earlier (19). For transient transfections these primary cultures were maintained in DMEM containing 5% fetal bovine serum stripped of hormones (OMEGA Scientific, Tarzana, CA). Cells (100,000 per well of a 6-well plate) were transfected with a total of 4 μg DNA using the calcium phosphate coprecipitation technique. Per well 100 ng of reporter plasmid, 250 ng pCMX-TR ( or ) or empty expression vector for controls, 10 ng pCMX RXR, 3 μg pBKS II (for efficient transfection as a carrier plasmid), and 650 ng -galactosidase (-gal) reporter plasmid to control for transfection efficiency were coprecipitated and added to the cell culture for 16 h. After that, the culture medium was changed and replaced with fresh medium containing no ligand, T3, or GC-1 at 10–7 M final concentration. Cell culture was continued for another 30 h after which cells were washed on the plate with PBS and harvested for luciferase assays and -gal assays or treated with 1% formaldehyde for ChIP assays. All experiments were repeated at least three times, and data points were collected in duplicates for each experiment and normalized to the -gal values. Evaluation and statistical analyses were done using the Microsoft Excel program with the Student t test function (Microsoft Corp., Redmond, WA).
ChIP assays
The ChIP assays were performed according to the protocols of Upstate Biotechnology (Lake Placid, NY) with some modifications. Neonatal myocytes were transfected with 100 ng luciferase reporter per well of a 6-well plate, 250 ng of expression vector for TR, and 10 ng of expression vector for RXR using Polyfect reagent (QIAGEN) and cultured as described above. After no treatment or treatment with T3 or GC-1 at 10–7 M final concentration, cells were cross-linked with 1% formaldehyde for 15 min at room temperature. Cells were collected by washing twice with ice-cold PBS and centrifuged for 4 min at 1000 x g. Cell pellets were resuspended in 250 μl of cell lysis buffer [50 mM Tris-HCl (pH 8.0), 85 mM KCl, and 0.5% Nonidet P-40] with protease inhibitor (Roche Molecular Biochemicals, Indianapolis, IN), incubated on ice for 10 min, and centrifuged for 4 min at 2000 x g. The pellets were resuspended in 250 μl of sodium dodecyl sulfate (SDS) lysis buffer [1% SDS, 10 mM EDTA, and 50 mM Tris-HCl (pH 8.1)] with protease inhibitor and sonicated four times for 15 sec each time followed by centrifugation at 14,000 x g for 10 min. Supernatants were collected and diluted 10-fold with dilution buffer [0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl (pH 8.1), and 167 mM NaCl] with protease inhibitor followed by preclearing with 2 μg of sheared salmon sperm DNA and protein A/G agarose [50 μl of a 50% slurry in 10 mM Tris-HCl (pH 8.1)-1 mM EDTA] for 2 h at 4 C.
Immunoprecipitation with the following antibodies was performed at 4 C overnight: anti-nuclear receptor corepressor (NCoR) (sc-8994), anti-histone deacetylase (HDAC)3 (sc-11417), anti-steroid receptor coactivator (SRC)-1 (sc-8995), and normal rabbit IgG (sc-2027) (Santa Cruz Biotechnology, Santa Cruz, CA). Immunoprecipitated complexes were collected with protein A/G agarose beads followed by sequential washes in low-salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), 150 mM NaCl], high-salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1),and 500 mM NaCl], LiCl wash buffer [0.25 M LiCl, 1% Nonidet P-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl (pH 8.1)], and Tris-EDTA buffer. Precipitates were eluted with elution buffer (1% SDS, 0.1 M NaHCO3), and 5 M NaCl to a final concentration of 200 mM were added to reverse cross-links at 65 C for 6 h. DNA fragments were purified with a PCR purification kit (QIAGEN). A total of 0.5 to 3 μl of purified sample was used in 23 to 28 cycles of PCR. The (forward) primer for the pT81luc reporter plasmid was primer 17 (Table 1) and (reverse primer) for the SERCa2 promoter primer 18 (Table 1).
Results
T3 responsiveness of SERCa2 5'-flanking DNA is preserved when linked to a minimal tk promoter
We previously demonstrated that 3.2 and 0.6 kb of rat SERCa2 5'-flanking DNA mediated strong responses to T3 (19, 22). Figure 1 shows the position of the TREs within 500 bp upstream of the transcription start as well as the orientation, sequence, and mutation of the half-sites in each TRE. Figure 2 demonstrates that the overall stimulation observed with T3 in neonatal myocytes was similar for the 3.2 and 0.6 kb of rat SERCa2 5'-flanking DNA linked to luciferase. In these experiments we observed a relatively high basal activity of the 0.6 kb SERCa2 regulatory region that may obscure the subtle differences in activity upon mutation of individual TREs. We therefore linked the SERCa2 regulatory region containing all three TREs to the heterologous tk promoter present in the vector pT81luc. This reduced the basal level about 4-fold and enhanced the relative repression and activation by the three SERCa2 TREs 3- to 4-fold, as shown in Fig. 2. We also noted that TR1 and TR1 had the same stimulating or silencing activity in our experimental setup using neonatal myocytes (data not shown).
Individual TREs in the SERCa2 promoter differ in their responses to T3 and unliganded TRs
To evaluate the function of individual TREs, we determined how elimination of one of the TREs in the 0.6 kb SERCa2 promoter would influence transcriptional effects (Fig. 3). Elimination of TRE1 results in a SERCa2 promoter termed M1. This promoter retains TR-induced silencing but loses responsiveness to T3 by TR-induced transcriptional stimulation, indicating that TRE1 mediates to a large extent and is necessary for the positive T3 action on the wild-type SERCa2 regulatory region. Elimination of TRE2 (M2) or TRE3 (M3) still shows significant T3 receptor mediated silencing and also T3-induced transcriptional activity, however, to a significantly diminished degree. Mutation of all three TREs (M1, M2, and M3) eliminated TR silencing and T3-induced transcriptional activity, which was expected. Thus, all three elements contribute to both silencing and induction, although only one element (TRE1) is absolutely required for induction.
Next we compared effects of individual TREs when the other two TREs were mutated to non-T3 receptor binding nucleotide sequences. Figure 4 shows that when TRE1 is present as the only functional TRE in the SERCa2 promoter construct termed M2,3, the silencing effect of the unoccupied TR and the TR plus T3 transcriptional stimulatory effect is preserved. Thus, TRE2 and TRE3 contribute to the induction because they suppress the actions at TRE1, an effect that is lost when neither TRE2 nor TRE3 is present. In contrast, when TRE2 is the only functional TRE as it occurs in the M1,3 construct, the silencing and T3 plus TR stimulatory functions are abolished, indicating that TRE2 can function along with only one or two other elements. When TRE3 is the only functional TRE, a modest but significant silencing and stimulatory effect is preserved.
In these and the following experiments in which only one TRE was present, either in the context of the SERCa2 5' flanking region or as an isolated element upstream of the natural SERCa2 (S2) promoter, we noted an unexpected silencing by the thyroid hormone analog GC-1 on TRE2 and TRE3 but not on TRE1. This was not observed when more than one TRE was present in the SERCa2 regulatory region. Figure 4 shows the effect of GC-1 in transient transfections with the wild-type and mutant SERCa2 regulatory region. Whereas the wild-type and M2,3 mutant were stimulated by T3 and GC-1 to a similar extent, M1,3 and M1,2 showed significant silencing in the presence of GC-1; by contrast, T3 showed no effect on TRE2 and stimulated TRE3. The silencing ranged from one half on TRE2 to one fifth on TRE3 in comparison with the basal activity of the same TREs. To determine whether the difference between T3 and GC-1 effects were dependent on the position of TRE3 in the SERCa2 regulatory region and relative to the other TREs was addressed by swap mutants. In S3-2-1 the order of the TREs was reversed, and this construct showed a higher basal level, compared with the wild-type or the other mutants, but could be repressed to the same level with unliganded TR. Both T3 and GC-1 could stimulate this construct about 4-fold above the repressed level. When TRE3 was placed at the position of TRE1 and the other two TREs were mutated (S3M2,3, Fig. 4), silencing by unliganded TR was abolished, and no stimulation with T3 could be observed. However, silencing in the presence of GC-1 was evident, although somewhat weaker than when TRE3 resides in its natural position.
To further elucidate the functions and separate them from the influence of surrounding elements, the individual TREs were placed in front of a minimal SERCa2 promoter as described in the experimental procedures. Figure 5 illustrates results of transient transfections of these constructs into neonatal myocytes. TRE1 showed repression by the unliganded receptor and stimulation by T3 and GC-1, similar to that observed with TRE1 present as the only functional TRE in the SERCa2 5' flanking region, although repression and activation from this minimal promoter element were decreased relative to the element present in the context of the SERCa2 regulatory region. TRE2 had no silencing activity with the unliganded TR and no T3 plus TR stimulating activity but mediated GC-1-dependent repression in agreement with the results in which TRE2 was the only functional TRE in the SERCa2 5' flanking region. TRE3 upstream of the minimal SERCa2 promoter showed modest silencing with unliganded TR but was not statistically significant. Significant silencing was mediated by GC-1, whereas T3 stimulated this construct, in line with the results of the M1,2 mutant in the SERCa2 5' flanking region. The T4T construct was included as an unrelated control to show that on a synthetic DR+4 linked to a heterologous promoter both T3 and GC-1 have agonist function. Similar synthetic constructs with inverted palindromic half-sites spaced by four or six nucleotides showed T3 and GC-1 stimulation and thus no GC-1-dependent repression (data not shown).
Differential complex formation on TRE3 with GC-1 and T3
To further investigate the differential behavior of T3 and GC-1 on TRE1 vs. TRE3, we performed ChIP assays to determine whether GC-1 leads to a different complex assembly on TRE1, compared with TRE3 (Fig. 6A). Anti-HDAC3 and anti-NCoR antibodies precipitated both elements in a similar pattern. Lysates from nontransfected cells, cells in which TR was omitted, or immunoprecipitation with a control antiserum did not lead to an amplified PCR product. In the absence of T3 and GC-1, both TRE1 and TRE3 associated with NCoR- and HDAC3-containing complexes. The association with NCoR and HDAC3 was diminished with T3 and GC-1 on TRE1 but only with T3 on the TRE3. On TRE3, NCoR and HDAC3 associations were slightly enhanced in the presence of GC1, which is consistent with the luciferase results in Figs. 4 and 5 in which GC-1 led to repression of TRE3. In a reciprocal experiment with anti-SRC-1 antibodies, we found that in the absence of ligand, very little SRC-1 was associated with both TREs. In the presence of T3, both elements contained complexes with SRC-1; however, only TRE1 bound an SRC-1 containing complex with GC-1, whereas TRE3 did not bind SRC-1 in the presence of GC-1. These results are also consistent with the transactivation behavior of both ligands in the luciferase assays. Immunoprecipitation with anti-silencing mediator of retinoic acid and thyroid hormone receptor, anti-transducin -like 1 protein (TBL-1), anti-HDAC1, and anti-HDAC2 antibodies did not bring down any elements above the background level (data not shown). Removal of a small aliquot (20 μl of 1 ml lysate) of the cross-linked lysate before immunoprecipitation served as an input control for the presence of the reporters in the lysate and consistently yielded a PCR product. A schematic is presented to illustrate the T3- or GC-1 mediated association of coactivators or corepressors on TRE1 and TRE3 (Fig. 6B).
Discussion
Thyroid hormone-regulated genes with promoters that contain multiple TREs, i.e. the rat GH gene, S14 gene, CPT-1 gene, Xenopus TRA gene, and the HIV-1 LTR genes, have been described previously (23, 24, 25, 26, 27). A systematic analysis of the individual TREs and their functional interaction in these promoters has not yet been fully explored. Here we attempted to define the role of the individual SERCa2 TREs within the promoter context or isolated upstream of a minimal promoter. In addition, we present somewhat surprising results on the ability of T3 vs. GC1 to activate or suppress transcription from TRE3 of the SERCa2 promoter, which has the configuration of an inverted palindrome.
Our results, that both T3 and GC-1 stimulate the wild-type SERCa2 regulatory region to the same extent, indicate that at a concentration of 10–7 M of ligand, both ligands can enter the neonatal myocytes and elicit similar transactivation function through TR or TR. This is not unexpected because it was shown previously (28) that at 10–7 M, the binding of both ligands to TR and TR were similar. We did notice a stronger stimulation of the wild-type SERCa2 promoter by GC-1 through the TR receptor, compared with the TR receptor, at lower ligand concentrations (data not shown). However, GC-1-specific repression via TRE2 and TRE3 was only marginal at lower ligand concentrations and was maximal at 10–7 M.
Mutation of single TREs in the SERCa2 regulatory region suggests that a higher-order structure exists, involving complexes on all three TREs. This structure seems to be dependent on TRE1 for recruitment of coactivator complexes because TRE2 and TRE3 together mediate only repression by unliganded and liganded TRs as shown in Fig. 3. This is in contrast to the results with individual TREs in which unliganded TRs repress and liganded TRs activateTRE3. A combination of TRE1 with TRE2 or TRE3 demonstrates the dominance of TRE1 for transactivation. The lower magnitude of transactivation shown with the M2 mutant could be explained by the decreased formation of the potent transactivation complex that forms on the wild-type SERCa2 regulatory region in the absence of TR binding to TRE2 or the possibility that other factors may bind to the mutated TRE2 region that could interfere with T3 signaling. Predictably mutation of all three TREs leads to loss of TR-mediated repression and ligand-activated stimulation of the SERCa2 regulatory region.
Mutation of two TREs, leaving only one functional TRE in the promoter, revealed the strong activation function potential of TRE1 and a differential effect of GC-1 compared with T3 on TRE2 or TRE3. This suggests that all three TREs together assemble a higher-order structure of TRs and their cofactors in which TRE2 and TRE3 function in the context of the entire promoter to modulate TRE1 function in an inhibitory role but that allows optimal function for the overall promoter. The isolated TRE2 (M1,3) lacks TR-mediated repression and T3-induced transactivation but interestingly mediates GC-1-dependent silencing. This indicates that TRE2 can bind TRs, which we had shown previously (4), although the specific structure of two inverted palindromic half-sites spaced by four nucleotides has not been reported in another promoter before. The mutant M1,2 in which TRE3 is the only functional TRE, shows unliganded repression and small transactivation with T3. GC-1, in contrast, shows a strong repression, demonstrating an opposite effect to T3.
The striking difference that we observed on TRE3 with T3 and GC-1 prompted us to investigate whether the order of the TREs within the SERCa2 regulatory region or the particular position of TRE3 was important for the GC-1-specific effect. In Fig. 4 we show the results with swap mutants in which the order of the TREs was changed from 1-2-3 in the wild-type to 3-2-1 in the swap mutant S3-2-1. The repositioning of TRE3 did not change the overall profile of unliganded repression and ligand-induced transactivation with both T3 and GC-1. This shows that as long as TRE1 is present either in a distal position or a proximal position relative to the transcription start site, GC-1 functions as an agonist. The other swap mutant that we tested (S3,M2,3) placed TRE3 at the TRE1 position, whereas the other two remaining TREs were mutated. This swap eliminated the unliganded repression and T3-dependent transactivation of TRE3, possibly due to the different surrounding DNA binding factors in the SERCa2 regulatory region or the relative distance of TRE3 to the transcription start site. Interestingly, the ligand-dependent silencing by GC-1 was preserved, however, to a much smaller extent than in the natural position of TRE3.
Placing the individual TREs outside of their promoter context upstream of a minimal SERCa2 (S2) promoter, as shown in Fig. 5, mimicked their qualitative function within the promoter. These results showed that the TREs did not need the neighboring DNA binding factors in the SERCa2 regulatory region for unliganded repression or ligand-dependent activation and silencing, which was observed in the intact promoter.
To provide a mechanistic explanation for the differential GC-1 vs. T3-dependent silencing mediated by TRE3, we compared the composition of coactivator and corepressor complexes of TRE1 and TRE3 alone within the promoter context by ChIP of transient transfected plasmid reporters (29) in primary cultures of neonatal cardiac myocytes. Figure 6A shows that the complex assembled on TRE1 contained very little or no SRC-1 in the absence of ligand but associated with SRC-1 in the presence of T3 and GC-1. This is consistent with the agonist function of both ligands shown in Fig. 4. In contrast, TRE3 did not associate with SRC-1 in the presence of GC-1 but did associate in the presence of T3. This is also consistent with the differential GC-1- vs. T3-dependent silencing observed with M1,2 shown in Fig. 4. ChIP assays with the corepressor antibody for NCoR and the histone deacetylase antibody for HDAC3 paralleled each other. Here we observed a good association of both proteins, NCoR and HDAC3, in the absence of ligand on TRE1 and TRE3, consistent with the finding that both mutants M2,3 and M1,2 show unliganded TR-mediated silencing. The association with NCoR and HDAC3 is weakened in the presence of T3 and GC-1 on TRE1. On TRE3, however, only T3 weakened NCoR and HDAC3 association, whereas GC-1 appeared to increase NCoR and HDAC3 binding to the TRE3 complex. The functional consequence of these observations would be a GC-1-dependent silencing on TRE3 but not on TRE1, which is what we observed with M2,3 and M1,2, shown in Fig. 4. It is of interest to note that GC-1 largely behaves as a T3 agonist but recruits corepressors to the complex formed on isolated TREs configured as inverted palindromes.
The most likely explanation for the different effects of T3 and GC-1 rests with the structural divergence between the two molecules and possible changes in the surface of the TR that are induced when T3 or GC-1 occupies the ligand binding pocket. The T3 ligand forms an intimate part of the occupied TR structure, and a ligand with a different structure from T3, such as GC-1, will make different contacts with amino acid residues that line the ligand binding pocket. Indeed, on the basis of crystal structures, different contacts between the GC-1/TR complex and the T3/TR complex have been described (13). However, the surfaces of the T3 and GC-1 liganded TR ligand binding domains are identical in the crystal structures. Thus, it seems likely that differences in the interactions between the two ligands with the TR ligand binding domain result in differences in the surfaces of the receptor responsible for coactivator or corepressor recruitment that are not apparent in the static x-ray crystal structures but occur in the dynamic setting of receptor function. To our knowledge, this phenomenon has not been reported for other nuclear receptors but has major implications for selective modulation of their functions. Selective modulation in other contexts, such as with the selective estrogen receptor modulators, has involved rather major differential effects of the ligand on the placement of helix 12 that is readily apparent in the x-ray crystal structure result in major differential effects. By contrast, the more subtle effects observed here are likely to involve more subtle differences in the actions of ligands. Whether this explains the differential actions of GC-1 vs. T3 in animals, i.e. on triglyceride levels, is not known; however, the differential actions of two ligands that do not impart structural changes that can be detected by static x-ray crystal structures deserve further study. There is also a possible contribution of the different ligands to structural changes of homodimers or heterodimers bound to different half-site configurations, which has been suggested previously (30). The differences in half-site sequence and the structure of the TRE itself could alter receptor dimerization or cofactor recruitment with T3 vs. GC-1, and we look forward to studies involving crystal structure data with the full-length receptors bound to various TREs in the presence and absence of ligands. Here we show a combination of a ligand-induced and individual TRE-dependent effect of a thyroid hormone analog, GC-1, that acts as an agonist on a classical DR4 element but as an antagonist on TRE3 configured as an everted repeat.
Acknowledgments
We thank Dr. Brian West for advice on the bidirectional PCR.
Footnotes
This work was supported by National Institutes of Health Grants HL 25022-19 (to W.H.D.), DK52798 (to T.S.), and DK41842 and DK09516 (to J.D.B.) and the Foundation of Support to the Research of the State of So Paulo (FAPESP) 99/11981-3 (to G.G.). J.D.B. has proprietary interests in, and serves as a consultant to, Karo Bio AB (Huddinge, Sweden), which has commercial interests in this area of research.
Current address for B.G.: Duke University Medical Center, Department of Neurobiology, Box 3209, Durham, North Carolina 27710.
Abbreviations: ChIP, Chromatin immunoprecipitation; -gal, -galactosidase; HDAC, histone deacetylase; NCoR, nuclear receptor corepressor; RXR, retinoid X receptor; SDS, sodium dodecyl sulfate; SERCa2, sarcoendoplasmic reticulum calcium ATPase; SRC, steroid receptor coactivator; tk, thymidine kinase; TR, T3 receptor; TRE, T3 response element.
References
Klein I, Ojamaa K 2001 Thyroid hormone and the cardiovascular system. N Engl J Med 344:501–509
Bluhm WF, Meyer M, Sayen MR, Swanson EA, Dillmann WH 1999 Overexpression of sarcoplasmic reticulum Ca(2+)-ATPase improves cardiac contractile function in hypothyroid mice. Cardiovasc Res 43:382–388
Rohrer D, Dillmann WH 1988 Thyroid hormone markedly increases the mRNA coding for sarcoplasmic reticulum Ca2+-ATPase in the rat heart. J Biol Chem 263:6941–6944
Hartong R, Wang N, Kurokawa R, Lazar MA, Glass CK, Apriletti JW, Dillmann WH 1994 Delineation of three different thyroid hormone-response elements in promoter of rat sarcoplasmic reticulum Ca2+ATPase gene. Demonstration that retinoid X receptor binds 5' to thyroid hormone receptor in response element 1. J Biol Chem 269:13021–13029
Evans RM 1988 The steroid and thyroid hormone receptor superfamily. Science 240:889–895
Wood WM, Ocran KW, Gordon DF, Ridgway EC 1991 Isolation and characterization of mouse complementary DNAs encoding and thyroid hormone receptors from thyrotrope cells: the mouse pituitary-specific 2 isoform differs at the amino terminus from the corresponding species from rat pituitary tumor cells. Mol Endocrinol 5:1049–1061
Mitsuhashi T, Nikodem VM 1989 Regulation of expression of the alternative mRNAs of the rat -thyroid hormone receptor gene. J Biol Chem 264:8900–8904
Lazar MA, Hodin RA, Darling DS, Chin WW 1988 Identification of a rat c-erbA -related protein which binds deoxyribonucleic acid but does not bind thyroid hormone. Mol Endocrinol 2:893–901
Sakurai A, Nakai A, DeGroot LJ 1990 Structural analysis of human thyroid hormone receptor gene. Mol Cell Endocrinol 71:83–91
Hodin RA, Lazar MA, Wintman BI, Darling DS, Koenig RJ, Larsen PR, Moore DD, Chin WW 1989 Identification of a thyroid hormone receptor that is pituitary-specific. Science 244:76–79
Williams GR 2000 Cloning and characterization of two novel thyroid hormone receptor isoforms. Mol Cell Biol 20:8329–8342
Feng W, Ribeiro RC, Wagner RL, Nguyen H, Apriletti JW, Fletterick RJ, Baxter JD, Kushner PJ, West BL 1998 Hormone-dependent coactivator binding to a hydrophobic cleft on nuclear receptors. Science 280:1747–1749
Wagner RL, Huber BR, Shiau AK, Kelly A, Cunha Lima ST, Scanlan TS, Apriletti JW, Baxter JD, West BL, Fletterick RJ 2001 Hormone selectivity in thyroid hormone receptors. Mol Endocrinol 15:398–410
Bryant HU, Dere WH 1998 Selective estrogen receptor modulators: an alternative to hormone replacement therapy. Proc Soc Exp Biol Med 217:45–52
Trost SU, Swanson EA, Gloss B, Wang-Iverson DB, Zhang H, Volodarsky T, Grover GJ, Baxter JD, Chiellini G, Scanlan TS, Dillmann WH 2000 The thyroid hormone receptor--selective agonist GC-1 differentially affects plasma lipids and cardiac activity. Endocrinology 141:3057–3064
Ribeiro MO, Carvalho SD, Schultz JJ, Chiellini G, Scanlan TS, Bianco AC, Brent GA 2001 Thyroid hormone-sympathetic interaction and adaptive thermogenesis are thyroid hormone receptor isoform-specific. J Clin Invest 108:97–105
Kurokawa R, DiRenzo J, Boehm M, Sugarman J, Gloss B, Rosenfeld MG, Heyman RA, Glass CK 1994 Regulation of retinoid signalling by receptor polarity and allosteric control of ligand binding. Nature 371:528–531
Saatcioglu F, Deng T, Karin M 1993 A novel cis element mediating ligand-independent activation by c-ErbA: implications for hormonal regulation. Cell 75:1095–1105
Rohrer DK, Hartong R, Dillmann WH 1991 Influence of thyroid hormone and retinoic acid on slow sarcoplasmic reticulum Ca2+ ATPase and myosin heavy chain gene expression in cardiac myocytes. Delineation of cis-active DNA elements that confer responsiveness to thyroid hormone but not to retinoic acid. J Biol Chem 266:8638–8646
Liu HC, Towle HC 1994 Functional synergism between multiple thyroid hormone response elements regulates hepatic expression of the rat S14 gene. Mol Endocrinol 8:1021–1037
Nordeen SK 1988 Luciferase reporter gene vectors for analysis of promoters and enhancers. Biotechniques 6:454–458
Moriscot AS, Sayen MR, Hartong R, Wu P, Dillmann WH 1997 Transcription of the rat sarcoplasmic reticulum Ca2+ adenosine triphosphatase gene is increased by 3,5,3'-triiodothyronine receptor isoform-specific interactions with the myocyte-specific enhancer factor-2a. Endocrinology 138:26–32
Norman MF, Lavin TN, Baxter JD, West BL 1989 The rat growth hormone gene contains multiple thyroid response elements. J Biol Chem 264:12063–12073
Zilz ND, Murray MB, Towle HC 1990 Identification of multiple thyroid hormone response elements located far upstream from the rat S14 promoter. J Biol Chem 265:8136–8143
Jansen MS, Cook GA, Song S, Park EA 2000 Thyroid hormone regulates carnitine palmitoyltransferase I gene expression through elements in the promoter and first intron. J Biol Chem 275:34989–34997
Urnov FD, Wolffe AP 2001 An array of positioned nucleosomes potentiates thyroid hormone receptor action in vivo. J Biol Chem 276:19753–19761
Hsia SC, Shi YB 2002 Chromatin disruption and histone acetylation in regulation of the human immunodeficiency virus type 1 long terminal repeat by thyroid hormone receptor. Mol Cell Biol 22:4043–4052
Chiellini G, Apriletti JW, Yoshihara HA, Baxter JD, Ribeiro RC, Scanlan TS 1998 A high-affinity subtype-selective agonist ligand for the thyroid hormone receptor. Chem Biol 5:299–306
Ishizuka T, Lazar MA 2003 The N-CoR/histone deacetylase 3 complex is required for repression by thyroid hormone receptor. Mol Cell Biol 15:5122–5131
Sjoberg M, Vennstrom B 1995 Ligand-dependent and -independent transactivation by thyroid hormone receptor 2 is determined by the structure of the hormone response element. Mol Cell Biol 15:4718–4726(Bernd Gloss, Gisele Giann)
Department of Physiology and Biophysics (G.G.) and Department of Cell and Developmental Biology (A.S.M.), ICB University of Sao Paulo, 05508-900 Sao Paulo, Brazil
Departments of Pharmaceutical Chemistry and Cellular and Molecular Pharmacology (G.C., T.S., J.D.B.), University of California, San Francisco, San Francisco, California 94143
Abstract
T3 regulates transcription of the rat sarcoendoplasmic reticulum calcium ATPase in the heart. The T3 effect is mediated by three differently configured T3 response elements (TREs). Here we report the mutation of each individual TRE in the promoter and the contribution of each TRE on gene expression. Mutation of TRE1, a direct repeat element, exerted the strongest T3 response, compared with TRE2 and TRE3, which are inverted palindromes. The isolated TRE2 and TRE3, which showed no response (TRE2) or were weakly positive with T3 (TRE3), became strong negative regulatory elements with the T3 analog GC-1. We found that TRE1 recruits corepressor complexes containing nuclear receptor corepressor and histone deacetylase 3 in the absence of ligand, and steroid receptor coactivator-1-containing coactivator complexes with both T3 and GC-1. TRE3 bound the same corepressor complexes without ligand but showed only a weak association with steroid receptor coactivator-1 with T3 and a strong association with corepressor complexes with GC-1. Thus, GC-1 appears to control cofactor association differentially on these two sarcoendoplasmic reticulum calcium ATPase TREs, which could be the mechanism of ligand-dependent transcriptional activation and repression observed with the isolated TRE1 and TRE3 elements. Because the x-ray crystal structures of GC-1 and T3 complexed with the TR ligand binding domain are superimposable, the results imply that GC-1 and T3 induce differential effects on the receptor that are not evident in the static structures but must occur in the dynamic setting of receptor function. These results have implications for selective modulation of receptor function by agonist ligands.
Introduction
THYROID HORMONE RECEPTORS (TRs) are members of the nuclear receptor superfamily that includes receptors for steroid hormones, peroxisomal proliferator-activated receptors, retinoids, and many other ligands. Two TR isoforms, TR and TR mediate T3 signaling in a variety of organs including the heart. The contractile function of the heart is markedly influenced by alterations of the thyroid status (1, 2). In particular, a delay in diastolic relaxation occurs with hypothyroidism, and this change can be linked to a delay in the calcium transient during the contractile cycle. One of the important mediators of calcium flux is the activity of the calcium ATPase of the sarcoplasmic reticulum (SERCa2), and the expression of this gene in cardiac myocytes is markedly influenced by thyroid hormone (T3) levels (3). We previously analyzed the enhancer and promoter region of the SERCa2 gene and identified three separate thyroid hormone response elements (TREs) in the region extending 500 nucleotides upstream from the transcription start (4). TRE1 is a direct repeat spaced by four nucleotides (DR4) and is localized between –481 and –458. TRE2 is an inverted palindrome, also termed everted repeat element, spaced by four nucleotides between –310 and –289. TRE3 is also an inverted palindrome but is spaced by six nucleotides and is localized between –218 and –195. It is currently unclear what functional contribution the three differently configured TREs make as individual TREs to the T3-mediated alteration in SERCa2 transcription and in which way they may interact with each other. It is of interest to determine how much each individual TRE contributes both to the transcriptional activation on ligand binding to the TR and transcriptional suppression or silencing by unliganded TR. The contribution that the individual TREs in the SERCa2 promoter make to these functions is further explored in this report.
Two separate genes encode TRs, and in addition, there are several splice variants (5, 6, 7). One of the genes codes for TR1, which binds T3 and a separate splice variant for the non-T3 binding TR2 isoform (8). The second gene encodes the T3 binding variants TR1, TR2, and TR3 (9, 10, 11). TRs have a modular design, which has been well characterized. The carboxy-terminal ligand binding domain undergoes T3-induced conformational changes and interacts with coactivators and corepressors (12).
Structural analysis of the ligand binding pocket of the TR1 and TR1 has demonstrated that the T3 ligand is an intrinsic part of the receptor structure largely buried inside the protein fold (13). It appears therefore likely that the binding of a modified T3 analog may have an influence on the configuration of the T3 receptor surface, especially in the regions in which coactivators and corepressors bind. The most noted examples of this are the selective estrogen receptor modulators such as tamoxifen and raloxifene (14). Recently novel thyroid hormone analogs like the compound GC-1 have been synthesized, and it has been demonstrated that administration of GC-1 to hypothyroid and euthyroid animals has effects that are distinct from those of T3 (15, 16).
Although it is difficult to predict a higher-order structure of chromatin in the region of these differently configured TREs in the SERCa2 promoter, individual mutation has led to some insight into which role each TRE may play. For example, binding on a direct repeat spaced by four nucleotides favors TR/retinoid X receptor (RXR) heterodimer binding with the RXR binding to the 5' half-site (17). In contrast, binding to inverted palindromes prefers homodimeric T3 receptor configurations, and on some of these elements, the T3 ligand leads to transcriptional repression (18). This half-site-dependent characteristic of each TRE may be enhanced by binding of a modified ligand. Therefore, we compared the transactivation behavior of the novel T3 analog GC-1 and T3 with TRs bound to the DR4 vs. inverted palindromic TREs of the SERCa2 promoter. We noticed significant differences in T3 vs. GC-1 transactivation activity with TR bound to the differently configured TREs, which was independent of the TR isoform that was cotransfected. Using chromatin immunoprecipitation (ChIP) assays, we studied cofactor association on the isolated TRE3 and TRE1 within the SERCa2 promoter. Our results may provide a mechanism for the differential transactivation activity observed with T3 and GC-1 on these elements.
Materials and Methods
Experimental animals
Experimental animal procedures were reviewed and approved by the Animal Subjects Program at the University of California, San Diego.
Cloning of the rat SERCa2 promoter up to position –3260 bp
The 3.2rSERCa2-luciferase construct was engineered by first digesting an insert (–3260 to +74 SERCa2 promoter) from 3.2pBLCAT3 (19) with HindIII. This fragment was subsequently blunt ended, digested with XhoI, and cloned into a luciferase-containing vector (pGL2, Promega, Madison, WI) previously digested with MluI, blunted with Klenow-polymerase, and digested with XhoI. The insert and the vector were ligated by using standard cloning methods. The 0.6rSERCa2-luciferase was constructed by digesting an insert (–560 to +74 SERCa2 promoter) from 0.6pBLCAT3 (19) also with HindIII. This insert was treated the same way as above and ligated to the vector pGL2 also digested with MluI, blunt ended with Klenow-polymerase, and digested with XhoI.
Mutation of the individual TREs in the rat SERCa2 promoter
The two guanines present in all TRE half-sites have been shown to be essential for receptor/DNA binding (20), independently of the ligand, and their substitution by adenosines completely abrogates receptor/DNA binding. Consequently we performed the same substitutions in all SERCa2 TREs to determine the individual contributions of those elements in SERCa2 promoter responsiveness to T3/GC-1. This was achieved by introducing point mutations by a Kunkel-based method. In brief, the SERCa2 promoter fragment encompassing +74 to –561 bp was inserted into the multiple cloning site of a M13 vector, which was used to transform a bacterial host strain deficient for deoxyuridine triphosphatase and uracil-N-glycosylase, which results in occasional substitutions of uracil for thymine in newly synthesized DNA. Subsequently the substitution of uracil for thymine in newly synthesized DNA was annealed to a set of primers containing the desired mutations, targeting simultaneously both TRE half-sites (TRE1 = primer 1; TRE2 = primer 2; TRE3 = primer 3; see Table 1 and Fig. 1 for sequence information). After annealing, the second strand was synthesized by T7 DNA polymerase and T4 DNA ligase. Finally, the double-stranded DNA was used to transform an uracil-N-glycosylase+ bacterial strain, which eliminates the wild-type strand. By using this method it was possible to generate SERCa2 promoters presenting TREs 1, 2, and 3 individually mutated. The different SERCa2 promoter TRE mutant combinations in the same promoter were accomplished by standard cloning procedures, using unique endonuclease restriction sites, flanking the three TREs (HindIII at –559 bp, ApaI at –395 bp and SmaI at –278 bp).
Construction of the luciferase reporter plasmids
The constructs in which the rat SERCa2 promoter was fused to the HSV-thymidine kinase (tk) promoter have been cloned using the pT81luc vector described by Nordeen (21). A HindIII/NaeI fragment from position –562 to position –163 (NaeI site) was cloned into the pT81luc vector cut with HindIII and SmaI. The SERCa2 promoter fragments in which individual or pairs of the TREs were mutated were cloned in the exact same way into pT81luc. Plasmids were prepared as Maxipreps (QIAGEN, Valencia, CA) following the standard protocol supplied by the manufacturer and were subsequently verified by sequencing using primer 8 (Table 1) within the luciferase gene. The plasmid T4T was constructed by cloning the annealed primers 9 and 10 (Table 1) into a plasmid that contained a minimal promoter of the human -retinoic acid receptor gene including the TATA-box upstream of the luciferase gene (21). Construction of the individual TREs upstream of the SERCa2 minimal promoter was done using the plasmid pGL2 basic as a vector backbone. The rat SERCa2 promoter from position –38 to +32 (19) was cloned into pGL2 basic, and the three SERCa2 TREs were cloned upstream of this promoter as annealed oligonucleotides with the following primers from Table 1: TRE1, primer 11/primer 12; TRE2, primer 13/primer 14; and TRE3, primer 15/primer 16. To generate the TRE swap mutants, we used a bidirectional PCR strategy in which an oligo was annealed to the sequences immediately upstream of the 5' TRE half-site and that had a mismatching extension reaching to the middle of the spacer of the TRE to be changed. Another oligo was annealed immediately downstream of the 3' half-site and extended also to the middle of the spacer of the TRE to be changed. As a template, the –0.6rSERCa2 tk-Luc circular plasmid was used with Pfu polymerase (Stratagene, La Jolla, CA). After PCR, DpnI treatment was used to select against parental DNA molecules, and the new strands, containing the swapped TREs, were ligated with T4-DNA ligase. The recircularized vector DNA incorporating the desired mutations was transformed into Escherichia coli. To change TRE1 to TRE3, we used a sense oligo primer 4 and antisense oligo primer 5 (Table 1). To change TRE3 to TRE1, we used as sense oligo primer 6 and as antisense primer 7. The mismatched bases are shown in Table 1 (lower case).
Other plasmids
Expression plasmids for transient transfections were all prepared by standard techniques using a QIAGEN Maxiprep kit. The rat cDNAs for TR1, TR1, and human RXR were cloned into the expression plasmid pCMX.
Transient transfection of neonatal myocytes
Neonatal myocytes from 1- to 2-d-old rats were prepared as described earlier (19). For transient transfections these primary cultures were maintained in DMEM containing 5% fetal bovine serum stripped of hormones (OMEGA Scientific, Tarzana, CA). Cells (100,000 per well of a 6-well plate) were transfected with a total of 4 μg DNA using the calcium phosphate coprecipitation technique. Per well 100 ng of reporter plasmid, 250 ng pCMX-TR ( or ) or empty expression vector for controls, 10 ng pCMX RXR, 3 μg pBKS II (for efficient transfection as a carrier plasmid), and 650 ng -galactosidase (-gal) reporter plasmid to control for transfection efficiency were coprecipitated and added to the cell culture for 16 h. After that, the culture medium was changed and replaced with fresh medium containing no ligand, T3, or GC-1 at 10–7 M final concentration. Cell culture was continued for another 30 h after which cells were washed on the plate with PBS and harvested for luciferase assays and -gal assays or treated with 1% formaldehyde for ChIP assays. All experiments were repeated at least three times, and data points were collected in duplicates for each experiment and normalized to the -gal values. Evaluation and statistical analyses were done using the Microsoft Excel program with the Student t test function (Microsoft Corp., Redmond, WA).
ChIP assays
The ChIP assays were performed according to the protocols of Upstate Biotechnology (Lake Placid, NY) with some modifications. Neonatal myocytes were transfected with 100 ng luciferase reporter per well of a 6-well plate, 250 ng of expression vector for TR, and 10 ng of expression vector for RXR using Polyfect reagent (QIAGEN) and cultured as described above. After no treatment or treatment with T3 or GC-1 at 10–7 M final concentration, cells were cross-linked with 1% formaldehyde for 15 min at room temperature. Cells were collected by washing twice with ice-cold PBS and centrifuged for 4 min at 1000 x g. Cell pellets were resuspended in 250 μl of cell lysis buffer [50 mM Tris-HCl (pH 8.0), 85 mM KCl, and 0.5% Nonidet P-40] with protease inhibitor (Roche Molecular Biochemicals, Indianapolis, IN), incubated on ice for 10 min, and centrifuged for 4 min at 2000 x g. The pellets were resuspended in 250 μl of sodium dodecyl sulfate (SDS) lysis buffer [1% SDS, 10 mM EDTA, and 50 mM Tris-HCl (pH 8.1)] with protease inhibitor and sonicated four times for 15 sec each time followed by centrifugation at 14,000 x g for 10 min. Supernatants were collected and diluted 10-fold with dilution buffer [0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl (pH 8.1), and 167 mM NaCl] with protease inhibitor followed by preclearing with 2 μg of sheared salmon sperm DNA and protein A/G agarose [50 μl of a 50% slurry in 10 mM Tris-HCl (pH 8.1)-1 mM EDTA] for 2 h at 4 C.
Immunoprecipitation with the following antibodies was performed at 4 C overnight: anti-nuclear receptor corepressor (NCoR) (sc-8994), anti-histone deacetylase (HDAC)3 (sc-11417), anti-steroid receptor coactivator (SRC)-1 (sc-8995), and normal rabbit IgG (sc-2027) (Santa Cruz Biotechnology, Santa Cruz, CA). Immunoprecipitated complexes were collected with protein A/G agarose beads followed by sequential washes in low-salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), 150 mM NaCl], high-salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1),and 500 mM NaCl], LiCl wash buffer [0.25 M LiCl, 1% Nonidet P-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl (pH 8.1)], and Tris-EDTA buffer. Precipitates were eluted with elution buffer (1% SDS, 0.1 M NaHCO3), and 5 M NaCl to a final concentration of 200 mM were added to reverse cross-links at 65 C for 6 h. DNA fragments were purified with a PCR purification kit (QIAGEN). A total of 0.5 to 3 μl of purified sample was used in 23 to 28 cycles of PCR. The (forward) primer for the pT81luc reporter plasmid was primer 17 (Table 1) and (reverse primer) for the SERCa2 promoter primer 18 (Table 1).
Results
T3 responsiveness of SERCa2 5'-flanking DNA is preserved when linked to a minimal tk promoter
We previously demonstrated that 3.2 and 0.6 kb of rat SERCa2 5'-flanking DNA mediated strong responses to T3 (19, 22). Figure 1 shows the position of the TREs within 500 bp upstream of the transcription start as well as the orientation, sequence, and mutation of the half-sites in each TRE. Figure 2 demonstrates that the overall stimulation observed with T3 in neonatal myocytes was similar for the 3.2 and 0.6 kb of rat SERCa2 5'-flanking DNA linked to luciferase. In these experiments we observed a relatively high basal activity of the 0.6 kb SERCa2 regulatory region that may obscure the subtle differences in activity upon mutation of individual TREs. We therefore linked the SERCa2 regulatory region containing all three TREs to the heterologous tk promoter present in the vector pT81luc. This reduced the basal level about 4-fold and enhanced the relative repression and activation by the three SERCa2 TREs 3- to 4-fold, as shown in Fig. 2. We also noted that TR1 and TR1 had the same stimulating or silencing activity in our experimental setup using neonatal myocytes (data not shown).
Individual TREs in the SERCa2 promoter differ in their responses to T3 and unliganded TRs
To evaluate the function of individual TREs, we determined how elimination of one of the TREs in the 0.6 kb SERCa2 promoter would influence transcriptional effects (Fig. 3). Elimination of TRE1 results in a SERCa2 promoter termed M1. This promoter retains TR-induced silencing but loses responsiveness to T3 by TR-induced transcriptional stimulation, indicating that TRE1 mediates to a large extent and is necessary for the positive T3 action on the wild-type SERCa2 regulatory region. Elimination of TRE2 (M2) or TRE3 (M3) still shows significant T3 receptor mediated silencing and also T3-induced transcriptional activity, however, to a significantly diminished degree. Mutation of all three TREs (M1, M2, and M3) eliminated TR silencing and T3-induced transcriptional activity, which was expected. Thus, all three elements contribute to both silencing and induction, although only one element (TRE1) is absolutely required for induction.
Next we compared effects of individual TREs when the other two TREs were mutated to non-T3 receptor binding nucleotide sequences. Figure 4 shows that when TRE1 is present as the only functional TRE in the SERCa2 promoter construct termed M2,3, the silencing effect of the unoccupied TR and the TR plus T3 transcriptional stimulatory effect is preserved. Thus, TRE2 and TRE3 contribute to the induction because they suppress the actions at TRE1, an effect that is lost when neither TRE2 nor TRE3 is present. In contrast, when TRE2 is the only functional TRE as it occurs in the M1,3 construct, the silencing and T3 plus TR stimulatory functions are abolished, indicating that TRE2 can function along with only one or two other elements. When TRE3 is the only functional TRE, a modest but significant silencing and stimulatory effect is preserved.
In these and the following experiments in which only one TRE was present, either in the context of the SERCa2 5' flanking region or as an isolated element upstream of the natural SERCa2 (S2) promoter, we noted an unexpected silencing by the thyroid hormone analog GC-1 on TRE2 and TRE3 but not on TRE1. This was not observed when more than one TRE was present in the SERCa2 regulatory region. Figure 4 shows the effect of GC-1 in transient transfections with the wild-type and mutant SERCa2 regulatory region. Whereas the wild-type and M2,3 mutant were stimulated by T3 and GC-1 to a similar extent, M1,3 and M1,2 showed significant silencing in the presence of GC-1; by contrast, T3 showed no effect on TRE2 and stimulated TRE3. The silencing ranged from one half on TRE2 to one fifth on TRE3 in comparison with the basal activity of the same TREs. To determine whether the difference between T3 and GC-1 effects were dependent on the position of TRE3 in the SERCa2 regulatory region and relative to the other TREs was addressed by swap mutants. In S3-2-1 the order of the TREs was reversed, and this construct showed a higher basal level, compared with the wild-type or the other mutants, but could be repressed to the same level with unliganded TR. Both T3 and GC-1 could stimulate this construct about 4-fold above the repressed level. When TRE3 was placed at the position of TRE1 and the other two TREs were mutated (S3M2,3, Fig. 4), silencing by unliganded TR was abolished, and no stimulation with T3 could be observed. However, silencing in the presence of GC-1 was evident, although somewhat weaker than when TRE3 resides in its natural position.
To further elucidate the functions and separate them from the influence of surrounding elements, the individual TREs were placed in front of a minimal SERCa2 promoter as described in the experimental procedures. Figure 5 illustrates results of transient transfections of these constructs into neonatal myocytes. TRE1 showed repression by the unliganded receptor and stimulation by T3 and GC-1, similar to that observed with TRE1 present as the only functional TRE in the SERCa2 5' flanking region, although repression and activation from this minimal promoter element were decreased relative to the element present in the context of the SERCa2 regulatory region. TRE2 had no silencing activity with the unliganded TR and no T3 plus TR stimulating activity but mediated GC-1-dependent repression in agreement with the results in which TRE2 was the only functional TRE in the SERCa2 5' flanking region. TRE3 upstream of the minimal SERCa2 promoter showed modest silencing with unliganded TR but was not statistically significant. Significant silencing was mediated by GC-1, whereas T3 stimulated this construct, in line with the results of the M1,2 mutant in the SERCa2 5' flanking region. The T4T construct was included as an unrelated control to show that on a synthetic DR+4 linked to a heterologous promoter both T3 and GC-1 have agonist function. Similar synthetic constructs with inverted palindromic half-sites spaced by four or six nucleotides showed T3 and GC-1 stimulation and thus no GC-1-dependent repression (data not shown).
Differential complex formation on TRE3 with GC-1 and T3
To further investigate the differential behavior of T3 and GC-1 on TRE1 vs. TRE3, we performed ChIP assays to determine whether GC-1 leads to a different complex assembly on TRE1, compared with TRE3 (Fig. 6A). Anti-HDAC3 and anti-NCoR antibodies precipitated both elements in a similar pattern. Lysates from nontransfected cells, cells in which TR was omitted, or immunoprecipitation with a control antiserum did not lead to an amplified PCR product. In the absence of T3 and GC-1, both TRE1 and TRE3 associated with NCoR- and HDAC3-containing complexes. The association with NCoR and HDAC3 was diminished with T3 and GC-1 on TRE1 but only with T3 on the TRE3. On TRE3, NCoR and HDAC3 associations were slightly enhanced in the presence of GC1, which is consistent with the luciferase results in Figs. 4 and 5 in which GC-1 led to repression of TRE3. In a reciprocal experiment with anti-SRC-1 antibodies, we found that in the absence of ligand, very little SRC-1 was associated with both TREs. In the presence of T3, both elements contained complexes with SRC-1; however, only TRE1 bound an SRC-1 containing complex with GC-1, whereas TRE3 did not bind SRC-1 in the presence of GC-1. These results are also consistent with the transactivation behavior of both ligands in the luciferase assays. Immunoprecipitation with anti-silencing mediator of retinoic acid and thyroid hormone receptor, anti-transducin -like 1 protein (TBL-1), anti-HDAC1, and anti-HDAC2 antibodies did not bring down any elements above the background level (data not shown). Removal of a small aliquot (20 μl of 1 ml lysate) of the cross-linked lysate before immunoprecipitation served as an input control for the presence of the reporters in the lysate and consistently yielded a PCR product. A schematic is presented to illustrate the T3- or GC-1 mediated association of coactivators or corepressors on TRE1 and TRE3 (Fig. 6B).
Discussion
Thyroid hormone-regulated genes with promoters that contain multiple TREs, i.e. the rat GH gene, S14 gene, CPT-1 gene, Xenopus TRA gene, and the HIV-1 LTR genes, have been described previously (23, 24, 25, 26, 27). A systematic analysis of the individual TREs and their functional interaction in these promoters has not yet been fully explored. Here we attempted to define the role of the individual SERCa2 TREs within the promoter context or isolated upstream of a minimal promoter. In addition, we present somewhat surprising results on the ability of T3 vs. GC1 to activate or suppress transcription from TRE3 of the SERCa2 promoter, which has the configuration of an inverted palindrome.
Our results, that both T3 and GC-1 stimulate the wild-type SERCa2 regulatory region to the same extent, indicate that at a concentration of 10–7 M of ligand, both ligands can enter the neonatal myocytes and elicit similar transactivation function through TR or TR. This is not unexpected because it was shown previously (28) that at 10–7 M, the binding of both ligands to TR and TR were similar. We did notice a stronger stimulation of the wild-type SERCa2 promoter by GC-1 through the TR receptor, compared with the TR receptor, at lower ligand concentrations (data not shown). However, GC-1-specific repression via TRE2 and TRE3 was only marginal at lower ligand concentrations and was maximal at 10–7 M.
Mutation of single TREs in the SERCa2 regulatory region suggests that a higher-order structure exists, involving complexes on all three TREs. This structure seems to be dependent on TRE1 for recruitment of coactivator complexes because TRE2 and TRE3 together mediate only repression by unliganded and liganded TRs as shown in Fig. 3. This is in contrast to the results with individual TREs in which unliganded TRs repress and liganded TRs activateTRE3. A combination of TRE1 with TRE2 or TRE3 demonstrates the dominance of TRE1 for transactivation. The lower magnitude of transactivation shown with the M2 mutant could be explained by the decreased formation of the potent transactivation complex that forms on the wild-type SERCa2 regulatory region in the absence of TR binding to TRE2 or the possibility that other factors may bind to the mutated TRE2 region that could interfere with T3 signaling. Predictably mutation of all three TREs leads to loss of TR-mediated repression and ligand-activated stimulation of the SERCa2 regulatory region.
Mutation of two TREs, leaving only one functional TRE in the promoter, revealed the strong activation function potential of TRE1 and a differential effect of GC-1 compared with T3 on TRE2 or TRE3. This suggests that all three TREs together assemble a higher-order structure of TRs and their cofactors in which TRE2 and TRE3 function in the context of the entire promoter to modulate TRE1 function in an inhibitory role but that allows optimal function for the overall promoter. The isolated TRE2 (M1,3) lacks TR-mediated repression and T3-induced transactivation but interestingly mediates GC-1-dependent silencing. This indicates that TRE2 can bind TRs, which we had shown previously (4), although the specific structure of two inverted palindromic half-sites spaced by four nucleotides has not been reported in another promoter before. The mutant M1,2 in which TRE3 is the only functional TRE, shows unliganded repression and small transactivation with T3. GC-1, in contrast, shows a strong repression, demonstrating an opposite effect to T3.
The striking difference that we observed on TRE3 with T3 and GC-1 prompted us to investigate whether the order of the TREs within the SERCa2 regulatory region or the particular position of TRE3 was important for the GC-1-specific effect. In Fig. 4 we show the results with swap mutants in which the order of the TREs was changed from 1-2-3 in the wild-type to 3-2-1 in the swap mutant S3-2-1. The repositioning of TRE3 did not change the overall profile of unliganded repression and ligand-induced transactivation with both T3 and GC-1. This shows that as long as TRE1 is present either in a distal position or a proximal position relative to the transcription start site, GC-1 functions as an agonist. The other swap mutant that we tested (S3,M2,3) placed TRE3 at the TRE1 position, whereas the other two remaining TREs were mutated. This swap eliminated the unliganded repression and T3-dependent transactivation of TRE3, possibly due to the different surrounding DNA binding factors in the SERCa2 regulatory region or the relative distance of TRE3 to the transcription start site. Interestingly, the ligand-dependent silencing by GC-1 was preserved, however, to a much smaller extent than in the natural position of TRE3.
Placing the individual TREs outside of their promoter context upstream of a minimal SERCa2 (S2) promoter, as shown in Fig. 5, mimicked their qualitative function within the promoter. These results showed that the TREs did not need the neighboring DNA binding factors in the SERCa2 regulatory region for unliganded repression or ligand-dependent activation and silencing, which was observed in the intact promoter.
To provide a mechanistic explanation for the differential GC-1 vs. T3-dependent silencing mediated by TRE3, we compared the composition of coactivator and corepressor complexes of TRE1 and TRE3 alone within the promoter context by ChIP of transient transfected plasmid reporters (29) in primary cultures of neonatal cardiac myocytes. Figure 6A shows that the complex assembled on TRE1 contained very little or no SRC-1 in the absence of ligand but associated with SRC-1 in the presence of T3 and GC-1. This is consistent with the agonist function of both ligands shown in Fig. 4. In contrast, TRE3 did not associate with SRC-1 in the presence of GC-1 but did associate in the presence of T3. This is also consistent with the differential GC-1- vs. T3-dependent silencing observed with M1,2 shown in Fig. 4. ChIP assays with the corepressor antibody for NCoR and the histone deacetylase antibody for HDAC3 paralleled each other. Here we observed a good association of both proteins, NCoR and HDAC3, in the absence of ligand on TRE1 and TRE3, consistent with the finding that both mutants M2,3 and M1,2 show unliganded TR-mediated silencing. The association with NCoR and HDAC3 is weakened in the presence of T3 and GC-1 on TRE1. On TRE3, however, only T3 weakened NCoR and HDAC3 association, whereas GC-1 appeared to increase NCoR and HDAC3 binding to the TRE3 complex. The functional consequence of these observations would be a GC-1-dependent silencing on TRE3 but not on TRE1, which is what we observed with M2,3 and M1,2, shown in Fig. 4. It is of interest to note that GC-1 largely behaves as a T3 agonist but recruits corepressors to the complex formed on isolated TREs configured as inverted palindromes.
The most likely explanation for the different effects of T3 and GC-1 rests with the structural divergence between the two molecules and possible changes in the surface of the TR that are induced when T3 or GC-1 occupies the ligand binding pocket. The T3 ligand forms an intimate part of the occupied TR structure, and a ligand with a different structure from T3, such as GC-1, will make different contacts with amino acid residues that line the ligand binding pocket. Indeed, on the basis of crystal structures, different contacts between the GC-1/TR complex and the T3/TR complex have been described (13). However, the surfaces of the T3 and GC-1 liganded TR ligand binding domains are identical in the crystal structures. Thus, it seems likely that differences in the interactions between the two ligands with the TR ligand binding domain result in differences in the surfaces of the receptor responsible for coactivator or corepressor recruitment that are not apparent in the static x-ray crystal structures but occur in the dynamic setting of receptor function. To our knowledge, this phenomenon has not been reported for other nuclear receptors but has major implications for selective modulation of their functions. Selective modulation in other contexts, such as with the selective estrogen receptor modulators, has involved rather major differential effects of the ligand on the placement of helix 12 that is readily apparent in the x-ray crystal structure result in major differential effects. By contrast, the more subtle effects observed here are likely to involve more subtle differences in the actions of ligands. Whether this explains the differential actions of GC-1 vs. T3 in animals, i.e. on triglyceride levels, is not known; however, the differential actions of two ligands that do not impart structural changes that can be detected by static x-ray crystal structures deserve further study. There is also a possible contribution of the different ligands to structural changes of homodimers or heterodimers bound to different half-site configurations, which has been suggested previously (30). The differences in half-site sequence and the structure of the TRE itself could alter receptor dimerization or cofactor recruitment with T3 vs. GC-1, and we look forward to studies involving crystal structure data with the full-length receptors bound to various TREs in the presence and absence of ligands. Here we show a combination of a ligand-induced and individual TRE-dependent effect of a thyroid hormone analog, GC-1, that acts as an agonist on a classical DR4 element but as an antagonist on TRE3 configured as an everted repeat.
Acknowledgments
We thank Dr. Brian West for advice on the bidirectional PCR.
Footnotes
This work was supported by National Institutes of Health Grants HL 25022-19 (to W.H.D.), DK52798 (to T.S.), and DK41842 and DK09516 (to J.D.B.) and the Foundation of Support to the Research of the State of So Paulo (FAPESP) 99/11981-3 (to G.G.). J.D.B. has proprietary interests in, and serves as a consultant to, Karo Bio AB (Huddinge, Sweden), which has commercial interests in this area of research.
Current address for B.G.: Duke University Medical Center, Department of Neurobiology, Box 3209, Durham, North Carolina 27710.
Abbreviations: ChIP, Chromatin immunoprecipitation; -gal, -galactosidase; HDAC, histone deacetylase; NCoR, nuclear receptor corepressor; RXR, retinoid X receptor; SDS, sodium dodecyl sulfate; SERCa2, sarcoendoplasmic reticulum calcium ATPase; SRC, steroid receptor coactivator; tk, thymidine kinase; TR, T3 receptor; TRE, T3 response element.
References
Klein I, Ojamaa K 2001 Thyroid hormone and the cardiovascular system. N Engl J Med 344:501–509
Bluhm WF, Meyer M, Sayen MR, Swanson EA, Dillmann WH 1999 Overexpression of sarcoplasmic reticulum Ca(2+)-ATPase improves cardiac contractile function in hypothyroid mice. Cardiovasc Res 43:382–388
Rohrer D, Dillmann WH 1988 Thyroid hormone markedly increases the mRNA coding for sarcoplasmic reticulum Ca2+-ATPase in the rat heart. J Biol Chem 263:6941–6944
Hartong R, Wang N, Kurokawa R, Lazar MA, Glass CK, Apriletti JW, Dillmann WH 1994 Delineation of three different thyroid hormone-response elements in promoter of rat sarcoplasmic reticulum Ca2+ATPase gene. Demonstration that retinoid X receptor binds 5' to thyroid hormone receptor in response element 1. J Biol Chem 269:13021–13029
Evans RM 1988 The steroid and thyroid hormone receptor superfamily. Science 240:889–895
Wood WM, Ocran KW, Gordon DF, Ridgway EC 1991 Isolation and characterization of mouse complementary DNAs encoding and thyroid hormone receptors from thyrotrope cells: the mouse pituitary-specific 2 isoform differs at the amino terminus from the corresponding species from rat pituitary tumor cells. Mol Endocrinol 5:1049–1061
Mitsuhashi T, Nikodem VM 1989 Regulation of expression of the alternative mRNAs of the rat -thyroid hormone receptor gene. J Biol Chem 264:8900–8904
Lazar MA, Hodin RA, Darling DS, Chin WW 1988 Identification of a rat c-erbA -related protein which binds deoxyribonucleic acid but does not bind thyroid hormone. Mol Endocrinol 2:893–901
Sakurai A, Nakai A, DeGroot LJ 1990 Structural analysis of human thyroid hormone receptor gene. Mol Cell Endocrinol 71:83–91
Hodin RA, Lazar MA, Wintman BI, Darling DS, Koenig RJ, Larsen PR, Moore DD, Chin WW 1989 Identification of a thyroid hormone receptor that is pituitary-specific. Science 244:76–79
Williams GR 2000 Cloning and characterization of two novel thyroid hormone receptor isoforms. Mol Cell Biol 20:8329–8342
Feng W, Ribeiro RC, Wagner RL, Nguyen H, Apriletti JW, Fletterick RJ, Baxter JD, Kushner PJ, West BL 1998 Hormone-dependent coactivator binding to a hydrophobic cleft on nuclear receptors. Science 280:1747–1749
Wagner RL, Huber BR, Shiau AK, Kelly A, Cunha Lima ST, Scanlan TS, Apriletti JW, Baxter JD, West BL, Fletterick RJ 2001 Hormone selectivity in thyroid hormone receptors. Mol Endocrinol 15:398–410
Bryant HU, Dere WH 1998 Selective estrogen receptor modulators: an alternative to hormone replacement therapy. Proc Soc Exp Biol Med 217:45–52
Trost SU, Swanson EA, Gloss B, Wang-Iverson DB, Zhang H, Volodarsky T, Grover GJ, Baxter JD, Chiellini G, Scanlan TS, Dillmann WH 2000 The thyroid hormone receptor--selective agonist GC-1 differentially affects plasma lipids and cardiac activity. Endocrinology 141:3057–3064
Ribeiro MO, Carvalho SD, Schultz JJ, Chiellini G, Scanlan TS, Bianco AC, Brent GA 2001 Thyroid hormone-sympathetic interaction and adaptive thermogenesis are thyroid hormone receptor isoform-specific. J Clin Invest 108:97–105
Kurokawa R, DiRenzo J, Boehm M, Sugarman J, Gloss B, Rosenfeld MG, Heyman RA, Glass CK 1994 Regulation of retinoid signalling by receptor polarity and allosteric control of ligand binding. Nature 371:528–531
Saatcioglu F, Deng T, Karin M 1993 A novel cis element mediating ligand-independent activation by c-ErbA: implications for hormonal regulation. Cell 75:1095–1105
Rohrer DK, Hartong R, Dillmann WH 1991 Influence of thyroid hormone and retinoic acid on slow sarcoplasmic reticulum Ca2+ ATPase and myosin heavy chain gene expression in cardiac myocytes. Delineation of cis-active DNA elements that confer responsiveness to thyroid hormone but not to retinoic acid. J Biol Chem 266:8638–8646
Liu HC, Towle HC 1994 Functional synergism between multiple thyroid hormone response elements regulates hepatic expression of the rat S14 gene. Mol Endocrinol 8:1021–1037
Nordeen SK 1988 Luciferase reporter gene vectors for analysis of promoters and enhancers. Biotechniques 6:454–458
Moriscot AS, Sayen MR, Hartong R, Wu P, Dillmann WH 1997 Transcription of the rat sarcoplasmic reticulum Ca2+ adenosine triphosphatase gene is increased by 3,5,3'-triiodothyronine receptor isoform-specific interactions with the myocyte-specific enhancer factor-2a. Endocrinology 138:26–32
Norman MF, Lavin TN, Baxter JD, West BL 1989 The rat growth hormone gene contains multiple thyroid response elements. J Biol Chem 264:12063–12073
Zilz ND, Murray MB, Towle HC 1990 Identification of multiple thyroid hormone response elements located far upstream from the rat S14 promoter. J Biol Chem 265:8136–8143
Jansen MS, Cook GA, Song S, Park EA 2000 Thyroid hormone regulates carnitine palmitoyltransferase I gene expression through elements in the promoter and first intron. J Biol Chem 275:34989–34997
Urnov FD, Wolffe AP 2001 An array of positioned nucleosomes potentiates thyroid hormone receptor action in vivo. J Biol Chem 276:19753–19761
Hsia SC, Shi YB 2002 Chromatin disruption and histone acetylation in regulation of the human immunodeficiency virus type 1 long terminal repeat by thyroid hormone receptor. Mol Cell Biol 22:4043–4052
Chiellini G, Apriletti JW, Yoshihara HA, Baxter JD, Ribeiro RC, Scanlan TS 1998 A high-affinity subtype-selective agonist ligand for the thyroid hormone receptor. Chem Biol 5:299–306
Ishizuka T, Lazar MA 2003 The N-CoR/histone deacetylase 3 complex is required for repression by thyroid hormone receptor. Mol Cell Biol 15:5122–5131
Sjoberg M, Vennstrom B 1995 Ligand-dependent and -independent transactivation by thyroid hormone receptor 2 is determined by the structure of the hormone response element. Mol Cell Biol 15:4718–4726(Bernd Gloss, Gisele Giann)