Specific Histone Lysine 4 Methylation Patterns Define TR-Binding Capacity and Differentiate Direct T₃ Responses

The histone code, notably H3 methylation, contributes to the precise control of gene expression that underlies complex physiological T₃ responses.

Abstract

The diversity of thyroid hormone T₃ effects in vivo makes their molecular analysis particularly challenging. Indeed, the current model of the action of T₃ and its receptors on transcription does not reflect this diversity. Here, T₃-dependent amphibian metamorphosis was exploited to investigate, in an in vivo developmental context, how T₃ directly regulates gene expression. Two, direct positively regulated T₃-response genes encoding transcription factors were analyzed: thyroid hormone receptor β (TRβ) and TH/bZIP. Reverse transcription-real-time quantitative PCR analysis on Xenopus tropicalis tadpole brain and tail fin showed differences in expression levels in premetamorphic tadpoles (lower for TH/bZIP than for TRβ) and differences in induction after T₃ treatment (lower for TRβ than for TH/bZIP). To dissect the mechanisms underlying these differences, chromatin immunoprecipitation was used. T₃ differentially induced RNA polymerase II and histone tail acetylation as a function of transcriptional level. Gene-specific patterns of TR binding were found on the different T₃ -responsive elements (higher for TRβ than for TH/bZIP), correlated with gene-specific modifications of H3K4 methylation (higher for TRβ than for TH/bZIP). Moreover, tissue-specific modifications of H3K27 were found (lower in brain than in tail fin). This first in vivo analysis of the association of histone modifications and TR binding/gene activation during vertebrate development for any nuclear receptor indicate that chromatin context of thyroid-responsive elements loci controls the capacity to bind TR through variations in histone H3K4 methylation, and that the histone code, notably H3, contributes to the fine tuning of gene expression that underlies complex physiological T₃ responses.

Thyroid hormones (THs) regulate multiple developmental and physiological functions in vertebrates. At the cellular level, T₃, the active form of TH, controls cell metabolism, proliferation, and commitment to differentiation or apoptosis. A large part of these regulations is achieved by T₃ binding to the thyroid hormone receptors (TRs). TRs are transcription factors that belong to the subfamily of nuclear receptors (1). TH is a versatile player, not only up-regulating but also down-regulating gene expression.

Current knowledge suggests that both the positive and negative effects of T₃ on gene transcription implicate TRs. To date, most of the studies on the mechanisms of action of TRs have been carried out on positively regulated T₃-response genes. Such studies have shown that TRs bind to specific sequences, thyroid-responsive elements (T₃REs) present in the promoter regions of their target genes. TR-induced transcriptional regulation requires chromatin modification and/or chromatin remodeling. Chromatin modification corresponds to posttranslational modification of N-terminal tail of histones, including a process not limited to acetylation and methylation (2). Such modifications allow control of transcriptional output. Chromatin remodeling will also affect DNA accessibility by localized alteration of nucleosomic structure.

More than a decade ago, a working model was proposed to explain the mechanism of repression by unliganded TR and activation of transcription by liganded TR on positively regulated genes (3). It is still largely valid today. Succinctly, in the absence of T₃, TRs bind T₃REs and recruit a nuclear receptor corepressor complex with histone deacetylase activity, creating a closed chromatin conformation inaccessible to transcriptional machinery and leading to gene repression. T₃ binding induces a conformational change of TR that relieves its inhibitory effect with first, the release of the corepressor complex and second, the recruitment of the steroid receptor coactivator/p300 coactivator complex that contains histone acetyl transferase activity, the SWItch/Sucrose NonFermentable complex involved in chromatin remodeling and the Mediator complex directly involved in transcription activation (4). The resulting chromatin reorganization leads to chromatin opening and gene activation. However, this model cannot explain the physiological diversity of T₃ effects, the understanding of which requires more detailed study of the molecular mechanisms underlying individual gene regulation in vivo.

To better understand transcriptional regulation by T₃ in an integrated physiological system, we use the well-studied Xenopus metamorphosis model. Anuran amphibian metamorphosis is one of the most striking developmental processes where tadpole transformation is marked by dramatic T₃-induced changes including de novo morphogenesis (limbs), tissue remodeling (nervous system), and organ resorption (tail) (5). These changes involve cascades of gene regulation initiated by T₃ and TRs. The diversity of T₃ effects requires tissue- and time-specific control of gene expression leading to the coordination of different transformations at different developmental stages in various organs. The levels of the TR and the localized activities of deiodinases that activate and inactivate TH and determine endogenous T₃ concentrations play important roles in the heterochronic responses of metamorphosis (6). At the single-cell level, the amplitude of direct T₃ responses and the starting-point level of expression need to be controlled to ensure coherent unfolding of the developmental program. The existence of T₃REs with different affinities, the implication of other transcription factors and their binding sites, and varied chromatin landscapes will all contribute to generating differential gene-specific responses to T₃. However, the fact that gene-specific responses are difficult to assess and analyze means that it is, in turn, difficult to derive generalizations about the contribution of various elements to the expression before induction and magnitude of TH responses.

We exploited in vivo chromatin immunoprecipitation (ChIP) to address the question of mechanisms of induction of individual, direct-response genes. Expression profiles, binding of TR, recruitment of RNA polymerase II (RNA PolII), and occupancy and function of several histone modifications were investigated on two direct T₃-target genes in the developing brain and tail fin of Xenopus tropicalis (X. tropicalis). The genes chosen were TRβ, a T₃/TR autoregulated gene and TH/bZIP, another rapid T₃-response transcription factor (7, 8). It has already been reported that, in Xenopus leavis, the T₃RE in TRβ shows a greater affinity for TR than TH/bZIP response elements (9) and that individual T₃-response genes have distinct coregulator requirements, the T₃-dependent corepressor to coactivator switch being gene specific (10). Here, we show that these two direct response genes display more differential features including specific magnitude of T₃ responses, gene-specific degrees of repression in absence of ligand, gene-specific dynamics of TR binding levels, and gene-specific chromatin modifications, notably distinct variations in histone lysine methylation. These findings have major implications for our understanding of the fine-tuning and gene-specific actions of T₃-mediated transcriptional control in a vertebrate developmental context and other nuclear receptor mechanisms of action.

Results

TRβ and TH/bZIP regulation by T₃ revisited in X. tropicalis

Thyroid hormone receptor β (TRβ) and basic leucine-zipper thyroid hormone-response gene (TH/bZIP) were selected for study because they are two well-known, T₃ direct-response genes in X. laevis, with sequenced and characterized promoters, respectively (11, 12). To identify their cDNA sequences and regulatory regions in X. tropicalis, a BLAST search was carried out using X. laevis sequences (respectively GenBank accession nos. M35361 and U41859 for cDNA and nos. U04675 and AF192491 for promoter). Homologs were found on the X. tropicalis genomic scaffold 26 for TRβ and 448 for TH/bZIP (http://genome.jgi-psf.org/Xentr4/Xentr4.home.html).

The isolated cDNA sequences were used to design primers (Supplemental Fig. 1 published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org) to measure the initial levels of expression and the effects of T₃ treatment on TRβ and TH/bZIP expression in X. tropicalis premetamorphic (stage NF 53) tadpole brain and tail fin by reverse transcription-quantitative PCR (RT-qPCR). rpl8 was used as a stable internal control gene for normalization, following Normfinder (13) analysis (data not shown). rpl8 is on scaffold 35 at position 982,964–1,030,015. T₃ treatment significantly increased the mRNA levels of TRβ and TH/bZIP in brain (Fig. 1A) and tail fin (Fig. 1B), but significant differences between genes were observed. First, and as previously described (12), the initial mRNA levels of TRβ were higher than those of TH/bZIP in tadpole brain (3.42 ± 0.36-fold; P < 0.01) and tail (6.3 ± 2.23-fold; P < 0.01) before T₃ treatment. After 48 h of T₃ treatment, mRNA levels of both genes increased significantly (P < 0.001; Fig. 1). However, large differences in fold response between the two genes were observed 48 h after T₃ treatment: for TRβ, 5.84 ± 0.46-fold in the brain and 43.87 ± 12.95-fold in the tail fin (P < 0.01), and for TH/bZIP 44.23 ± 2.91-fold in the brain and 144.4 ± 31.35-fold in the tail fin (P < 0.05). This result highlights a much stronger effect of T₃ on tail fin than on brain. Finally, at 48 h, the mRNA levels of the two genes differed significantly in the brain (Fig. 1A; higher for TH/bZIP than for TRβ), but not in tail fin (Fig. 1B). Because rpl8 expression levels were similar in both tissues in absence and presence of T₃ (data not shown), the expression levels of both genes could be compared between brain and tail fin. The initial mRNA levels of TRβ and TH/bZIP were higher in tadpole brain than in tail, respectively: 4.26 ± 1.55-fold (P < 0.01) and 5.93 ± 2.62-fold (P < 0.05), before T₃ treatment.

Fig. 1.

T₃ induces TRβ and TH/bZIP transcription. Tadpoles were treated for 48 h with 10 nm T₃. Total RNA was extracted from brain and tail fin and used for RT and real-time qPCR analysis of TRβ and TH/bZIP (a basic leucine-zipper TH-response gene) expression. Gene expression was normalized against rpl8 RNA. The data plotted are the log of 2^−ΔCT value where ΔCT is the difference between the control gene and the gene of interest. The results represents the mean and sem of three or six independent experiments. Statistical significance is indicated as not significant (ns); **, P < 0.01; or ***, P < 0.001.

Next, the T₃REs involved in this regulation were identified using BLAST to compare X. tropicalis sequences against X. laevis TRβ and TH/bZIP regulatory regions, respectively (GenBank accession nos. U04675 and AF192491). In scaffold 26 (TRβ), 77 bp with 97% identity were found to contain the T₃RE sequence described in X. laevis (Supplemental Fig. 2A). In scaffold 448 (TH/bZIP), 190 bp with 95% identity were found to contain the two previously described T₃RE (Supplemental Fig. 2B). One would never be absolutely sure that there are no other bindings sites that may contribute to the regulation. At the level of the T₃REs, which are classical direct repeats with four-nucleotide spacing (DR4), the identity is 100% for both genes (Supplemental Fig. 2, T₃RE highlighted in bold). The T₃REs of both genes differ from the consensus DR4 sequence (AGGTCANNNNAGGTCA) by a few nucleotides. TRβ T₃RE has a single nucleotide difference whereas the two TH/bZIP T₃REs each have three- and four-nucleotide differences. It seems that, as in X. laevis, TRβ and TH/bZIP are direct T₃-response genes in X. tropicalis but both genes behave differently regarding basal levels of expression and mRNA induction after T₃ treatment.

TR recruitment on T₃ response genes

To obtain more direct information on whether these differences in gene expression correlate with TR binding to chromatin, ChIP assays were used. A T₃ treatment period of 48 h was chosen because at this time point both TRβ and TH/bZIP gene expression is significantly up-regulated by T₃ whatever the tissue, presumably with a maximum of cells responding to the stimulus. This is vital when using an in vivo model, because maximal homogeneity of the T₃-response in the whole tissue is required. However, the treatments used could be sufficiently long to induce secondary changes that could have indirect and tissue-specific consequences on TR function. Chromatin isolated from control or T₃-treated premetamorphic (stage NF53) X. tropicalis tadpole brains and tail fins was immunoprecipitated with an antibody recognizing both TRα and TRβ. The TR-bound DNA fragments were quantified by real-time quantitative PCR (qPCR) (Fig. 2). Two kinds of DNA regions were compared (Fig. 2A for TRβ and Fig. 2B for TH/bZIP): the T₃-response promoters containing the transcription start site (TSS) and the TR binding site (T₃RE) and, as a control, a region distant by a few thousand bp from the promoter studied and that was not expected to bind TR [control zone (CZ)]. ChIP without antibody led to the unspecific precipitation of weak amount of DNA in brain (Supplemental Fig. 3A) and in tail fin (Supplemental Fig. 3B). Similar results were obtained when irrelevant antibodies were used (data not shown).

Fig. 2.

Effects of T₃ on DNA binding by TR at T₃-response genes. Chromatin isolated from brain or tail fin of T₃-treated X. tropicalis tadpoles (10 nm T₃ for 48 h) was immunoprecipitated, and the products were analyzed by qPCR for the presence of T₃RE or an upstream control region (CZ) as schematically represented in panel A for the TRβ gene and in panel B for the TH/bZIP gene. The distance between the T₃RE area and the CZ area, the position of the primers used for qPCR, and the position of the TSS are also indicated. Sequences are not drawn to scale. C, T₃ increases TR binding to TRβ and TH/bZIP T₃RE in tadpole brain. Chromatin was immunoprecipitated with antibodies against TR (ChIPαTR). D, T₃ induces TR binding to TRβ and TH/bZIP T₃RE in tadpole tail fin. Chromatin was immunoprecipitated with antibodies against TR (ChIPαTR). E, TRβ overexpression and TR recruitment on TRβ and TH/bZIP T₃RE in brain. Chromatin isolated from brain of T₃-treated X. laevis transgenic tadpoles overexpressing GFP fused to TRβ in neurons (10 nm T₃ for 48 h) was immunoprecipitated with antibodies against GFP (ChIPαGFP). The product was analyzed by qPCR for the presence of the conserved between X. tropicalis and X. laevis T₃RE containing region (T₃RE) of TRβ and TH/bZIP gene. TH/bZIP coding locus was used as negative control (Exon1 TH/bZIP). The mean values and sem of four (panels C and D) or three (panel E) independent experiments are expressed as percent of input. Statistical significance as compared with untreated animals is indicated as not significant (ns); *, P < 0.05; **, P < 0.01; or ***, P < 0.001.

In the absence of T₃, TR was present on the TRβ-regulatory region in brain (Fig. 2C) and in tail fin (Fig. 2D). As expected, no TR was found on TRβ CZ whether from brain (Fig. 2C) or tail fin (Fig. 2D). In the same samples, a weak but significant signal of TR on TH/bZIP T₃RE was seen in the brain compared with the CZ (Fig. 2C). In the tail fin, a clear signal of TR on TH/bZIP T₃RE was seen (Fig. 2D). After T₃ treatment, TR binding increased significantly on T₃RE region of each gene in both tissues (Fig. 2C for brain and Fig. 2D for tail fin). All these changes observed for TR binding after T₃ treatment were independent of any change in recruitment on CZ for both genes and in both tissues (Fig. 2C for brain and Fig. 2D for tail fin). TR presence was between 8- and 10-fold higher on the TRβ T₃RE than on the TH/bZIP T₃RE in both tissues analyzed (Fig. 2C for brain, P < 0.01; and Fig. 2D for tail fin, P < 0.001). Also for each gene, TR presence was also between 3- and 4-fold lower in the brain than in the tail fin [compare Fig. 2, panels C and D: TRβ minus T₃ (P < 0.05); TRβ plus T₃ (P < 0.01); TH/bZIP minus T₃ (P < 0.001); and TH/bZIP plus T₃ (P < 0.01)].

Our data suggest that in both brain and tail fin only the TRβ T₃RE is occupied by the TR expressed during premetamorphosis and that TR recruitment to TH/bZIP T₃RE only occurs when either TR expression and or T₃ levels increase during metamorphosis. The premetamorphic level of TR could therefore be a limiting factor for TR binding, implying that T₃-induced TRβ expression is required for the recruitment of TR to certain T₃REs such as TH/bZIP. To address this point, TRβ fused to green fluorescent protein (GFP) was overexpressed under the control of the Neuronal tubuline β promoter (NβT) by germinal transgenesis in X. laevis tadpole central nervous system. X. laevis was used instead of X. tropicalis because data regarding TR binding are comparable (our data and Refs. 7 and 9) and transgenesis in X. laevis gives higher success rates than in X. tropicalis. We used the NβT promoter to drive TRβ-GFP expression because expression from the ubiquitous Cytomegalovirus promoter decreases significantly during metamorphosis (14). Expression of the transgene in the F₁ generation can be controlled by observation of the GFP signal in the central nervous system (15) that allowed establishment of groups with similar levels of transgene expression (16). ChIP assay with antibodies raised against GFP were used to specifically immunoprecipitate exogenous TRβ-bound chromatin from nuclei isolated from control or T₃-treated premetamorphic (stage NF 53) tadpoles brains. The GFP fused TRβ-bound DNA fragments were analyzed by real-time qPCR (Fig. 2E). The presence in the ChIP product of three DNA regions was analyzed: X. laevis TRβ and TH/bZIP T₃REs and, as a control, a region corresponding to the first exon of TH/bZIP (exon1 TH/bZIP) distant from the T₃REs studied. As expected, TRβ-GFP was absent from the non-TR-binding zone (exon1 TH/bZIP) irrespective of the presence or absence of T₃ (Fig. 2E). In the absence of T₃, TRβ-GFP showed a clear signal on the TRβ-regulatory region, but only a very low signal on the TH/bZIP T₃RE (Fig. 2E) as observed for endogenous TR (Fig. 2C). After T₃ treatment, TRβ-GFP binding increased significantly on both T₃RE regions (Fig. 2E). Thus, TRβ overexpression does not result in higher TR binding to the TH/bZIP T₃RE region in absence of T₃, but does induce higher liganded TR binding to TH/bZIP T₃RE (Fig. 2E).

RNA PolII recruitment and H3K36 methylation differ on individual T₃ response genes

To link TRβ and TH/bZIP mRNA levels with transcriptional status, the presence of RNA PolII on both genes was studied by ChIP. RNA PolII recruitment was analyzed in the genomic regions previously described for TRβ (Fig. 3A) and TH/bZIP (Fig. 3B), with the T₃REs for both genes being near their TSS (in the middle of the T₃RE for TRβ and for TH/bZIP the T₃RE is 63 bp upstream the TSS), the CZ not being transcribed and the exons being transcribed [transcribed zone (TZ)]. In tail fin, RNA PolII was significantly recruited in the absence of T₃, on the TRβ TSS (Fig. 3C) and on the TH/bZIP TSS (Fig. 3D), compared with their respective CZ. After T₃ treatment, RNA PolII recruitment increased on both TRβ (Fig. 3C) and TH/bZIP T₃REs (Fig. 3D). Similar results were obtained when chromatin was isolated from brain (Supplemental Fig. 4), except that RNA PolII was not significantly recruited on TH/bZIP TSS in the absence of T₃ (Supplemental Fig. 4B). Comparing brain and tail fin, the levels and variations of RNA PolII recruitment were correlated with the increase of mRNA levels observed by RT-qPCR. As expected, RNA PolII was absent from the CZ irrespective of the presence of T₃ (for TRβ and TH/bZIP, Fig. 3, panels C and D, respectively). In the TZ of both genes, the sensitivity of the method did not allow us to detect significant levels of RNA PolII (Fig. 3, C and D).

Fig. 3.

Effects of T₃ on RNA PolII recruitment and Me3H3K36 occupancy at T₃-response genes. Chromatin isolated from tail fins of T₃-treated X. tropicalis tadpoles (10 nm T₃ for 48 h) was immunoprecipitated, and the product was analyzed by qPCR for the presence of the TSS-containing region, which is also the T₃RE-containing region, an upstream control region (CZ), or a coding region (TZ) as schematically represented in panel A for the TRβ gene and in panel B for the TH/bZIP gene. The distance between each areas and the position of the primers used for qPCR are also indicated. Exonic sequences are boxed and intronic or upstream sequences are indicated with solid line. Sequences are not drawn to scale. C, T₃ increases RNA PolII recruitment to TRβ TSS. Chromatin was immunoprecipitated with antibodies against RNA PolII (ChIPαRNA PolII). D, T₃ induces RNA PolII recruitment to TH/bZIP promoter. ChIP was done as described in panel C, but the product was analyzed with primers for TH/bZIP genomic regions. E, T₃ increases Me3H3K36 deposition at the TRβ transcribed genomic locus. ChIP was done using antibodies against Me3H3K36 (ChIPαMe3H3K36) and primers on TRβ genomic locus. F, T₃ induces H3K36 trimethylation at TH/bZIP. ChIP was done using antibodies against Me3H3K36 (ChIPαMe3H3K36) and primers for TH/bZIP genomic locus. The average values and sem of four independent experiments are expressed as percent of input. Statistical significance as compared with untreated animals is indicated as not significant (ns); *, P < 0.05; or **, P < 0.01.

To analyze further the transcriptional status of TRβ and TH/bZIP genes, the level of histone H3 trimethylation on lysine 36 (Me3H3K36) was measured. Me3H3K36 generally accumulates on the coding regions of transcribed genes (17). Before T₃ treatment, low levels of Me3H3K36 were seen on the TRβ transcribed region, but not on either the CZ or the TSS (Fig. 3E). In the same physiological conditions, Me3H3K36 was not found on any part of the TH/bZIP gene (Fig. 3F). After T₃ treatment, Me3H3K36 strongly accumulated, uniquely in the TZ of both genes (for TRβ and TH/bZIP, Fig. 3, panels E and F, respectively). Thus, there is a strong correlation between RNA PolII recruitment, Me3H3K36 deposition, and transcription levels.

H3 and H4 histone acetylation on T₃-responsive gene promoters

To better understand the molecular mechanisms responsible for the difference in T₃ regulation of direct T₃-response genes, the chromatin states of both the TRβ and TH/bZIP gene were analyzed. Histone acetylation of the promoters was investigated first, because TR regulation of transcription has been associated with recruitment of either histone deacetylase activity or histone acetyl transferase activity (10). Using ChIP assay with an antibody specific to pan-acetylated histone H4 (AcH4), ChiP products were analyzed for the presence of the different genomic area already presented in Fig. 2, A and B (for TRβ and TH/bZIP, respectively). T₃ treatment increased histone H4 acetylation on the TRβ and TH/bZIP T₃REs in brain (Fig. 4A) and tail fin (Fig. 4B). These increases were specific because they contrast with lack of effect T₃ on AcH4 for dimethyladenosine transferase 1 like (DIMT1L; Supplemental Fig. 5). DIMT1L is a distant housekeeping gene localized on scaffold 87 and is not a T₃-response gene (as revealed with Normfinder analysis; data not shown). For brain and tail fin, AcH4 levels before or after T₃ treatment were similar at TRβ T₃RE. The AcH4 levels were higher at THb/ZIP T₃RE in tail fin compared with brain but the induction rate after T₃ treatment was identical. On the CZ, in both tissues, the level of AcH4 was lower than around the T₃RE for TRβ (Fig. 4, A and B) and in the same order of magnitude for TH/bZIP (Fig. 4, A and B). Moreover, T₃ treatment increased AcH4 levels on the TRβ and TH/bZIP CZ in tail fin (Fig. 4B), but not in brain (Fig. 4A).

Fig. 4.

T₃ treatment effects on histone H3 and H4 acetylation at T₃-response genes. Chromatin isolated from brains or tail fins of T₃-treated tadpoles (10 nm T₃ for 48 h) was immunoprecipitated with antibodies against pan-acetylated histone H4 (AcH4), acetylated histone H3 (AcH3) at lysine 9 (K9) or lysine 18 (K18). ChIP products were analyzed by qPCR for the presence of T₃RE-containing region and CZ region as presented in Fig. 2A for TRβ gene and Fig. 2B for TH/bZIP gene. A, T₃ increases pan-AcH4 to T₃RE in brain. Chromatin was immunoprecipitated with antibodies against pan-AcH4 (ChIPαAcH4). B, T₃ increases pan-AcH4 to T₃RE in tail fin. ChIP was done with antibodies against pan-AcH4 (ChIPαAcH4). C, T₃ does not affect H3K9 acetylation at T₃RE in brain. ChIP was done using antibodies against AcH3K9 (ChIPαAcH3K9). D, T₃ increases H3K9 acetylation on TH/bZIP T₃RE but has no effect on TRβ T₃RE in tail fin. Antibodies against AcH3K9 (ChIPαAcH3K9) were used. E, T₃ increases H3K18 acetylation at T₃RE in brain. ChIP was done using antibodies against AcH3K18 (ChIPαAcH3K18). F, T₃ increases T₃RE H3K18 acetylation in tail fin. Chromatin was immunoprecipitated with antibodies against AcH3K18 (ChIPαAcH3K18). The average values and sem of four independent experiments are expressed as percent of input. Statistical significance as compared with untreated animals is indicated as not significant (ns); *, P < 0.05; **, P < 0.01; or ***, P < 0.001. CZ at least 2000 bp upstream T₃RE area.

Histone H3 is another target for NH₂-terminal tail lysine acetylation, and its state reflects gene activation. ChIP with pan-acetylated histone H3 (AcH3) antibodies, revealed high AcH3 occupancy on the TRβ T₃RE and TH/bZIP T₃RE. T₃ increased AcH3 occupancy significantly for TRβ but not for TH/bZIP (data not shown). Five H3 lysines can be acetylated, all recognized by the antibody used. Therefore, antibodies recognizing two lysine-specific acetylations were used next: lysines 9 (H3K9) and 18 (H3K18). Looking at the genomic area presented in Fig. 2, A and B, ChIP assays showed no T₃ effect on H3K9 acetylation for either target gene in brain (Fig. 4C). In tail fin, T₃ increased H3K9 acetylation for TH/bZIP T₃RE but not for TRβ T₃RE (Fig. 4D). T₃ increased H3K18 acetylation on both TRβ and TH/bZIP T₃REs in brain (Fig. 4E) and tail fin (Fig. 4F). As for AcH3K9, no effects were seen on DIMT1L (Supplemental Fig. 5). Interestingly, AcH3K9 and AcH3K18 occupancies at the TH/bZIP CZ in tail fin were also increased by T₃ treatment. Thus, global levels of histone acetylation were positively correlated with levels of gene expression, TR binding, and RNA PolII recruitment.

Histone methylation status of T₃-response gene promoters

Six major lysine residues (H3K4, H3K9, H3K27, H3K36, H3K79, and H4K20) can be mono-, di-, or trimethylated. Unlike histone lysine acetylation, which is generally coupled to activation, both the position of the lysine residue and the degree of methylation can have different transcriptional consequences (18–20). T₃ effects on the methylation status of lysines 9, 4, and 27 were examined. H3 lysine 9 methylation is associated with closed chromatin and gene repression (18). As shown in Supplemental Fig. 6, trimethylated and dimethylated H3K9 (Me3H3K9 and Me2H3K9, respectively) gave only weak insignificant signals on TRβ and TH/bZIP T₃REs and their CZ, independent of T₃ status.

The pattern of H3 methylation at lysine 4 (MeH3K4) is more ambiguous, correlating with either activation or repression (19). ChIP with antibody against mono/di/trimethyl H3K4 on brain, showed T₃ to significantly increase methylation at the TH/bZIP T₃RE, but not the TRβ T₃RE where the methylation level was already high and T₃ independent (data not shown). To address the specificity of H3K4 methylation, antibodies against dimethyl H3K4 (Me2H3K4) and trimethyl H3K4 (Me3H3K4) were used (Fig. 5, A–D). In both brain and tail fin, T₃ decreased levels of dimethylation on TRβ T₃RE (Fig. 5A) and increased it on TH/bZIP T₃RE (Fig. 5B). Interestingly, specific differences were seen with Me3H3K4, which was present at high levels on the TRβ T₃RE in tail fin (Fig. 5D) and to a lesser extent in brain (Fig. 5C), whereas Me3H3K4 was virtually absent from TH/bZIP T₃RE in brain (Fig. 5C) and tail fin (Fig. 5D). T₃ did not modify Me3H3K4 levels. Only the CZ of TH/bZIP in tail fin showed any significant levels of H3K4 methylation with no T₃ effects (Fig. 5, A–D).

Fig. 5.

T₃ treatment effects on histone H3 methylation at T₃-response genes. Chromatin isolated from brains or tail fins of T₃-treated tadpoles (10 nm T₃ for 48 h) was immunoprecipitated with antibodies against dimethyl histone H3 at lysine 4 (Me2H3K4), trimethyl histone H3 at lysine 4 (Me3H3K4), or trimethyl histone H3 at lysine 27 (Me3H3K27). ChIP products were analyzed by qPCR for the presence of T₃RE-containing region and CZ region as presented in Fig. 2A for TRβ gene and Fig. 2B for TH/bZIP gene. A, In brain, T₃ decreases Me2H3K4 at TRβ T₃RE and increases Me2H3K4 at TH/bZIP T₃RE. ChIP was done using antibodies against Me2H3K4 (ChIPαMe2H3K4). B, As observed in brain, T₃ decreases Me2H3K4 at TRβ T₃RE and increases Me2H3K4 at TH/bZIP T₃RE in tail fin. ChIP was done using antibodies against Me2H3K4. C, In brain, Me3H3K4 is present on TRβ T₃RE and not on TH/bZIP T₃RE with no T₃ effect. ChIP was done using antibodies against Me3H3K4 (ChIPαMe3H3K4). D, In tail fin, Me3H3K4 is present on TRβ T₃RE and slightly detected on TH/bZIP T₃RE with no T₃ effect. ChIP used antibodies against Me3H3K4. E, In brain, T₃ decreases Me3H3K27 at TRβ T₃RE but not on TH/bZIP T₃RE. Chromatin was immunoprecipitated with antibodies against Me3H3K27 (ChIPαMe3H3K27). F, T₃ strongly decreases Me3H3K27 occupancy at T₃RE. ChIP was done using antibodies against Me3H3K27. The average values and sem of four independent experiments are expressed as percent of input. Statistical significance as compared with untreated animals is indicated as not significant (ns); *, P < 0.05; or **, P < 0.01; or ***, P < 0.001. CZ at least 2000 bp upstream from T₃RE area.

Finally, the trimethylation of histone H3 at lysine 27 (Me3H3K27), a mark of transcriptional repression (20), was examined. In brain, T₃ treatment decreased slightly the level of Me3H3H27 for TRβ whereas it had no effect on TH/bZIP T₃RE (Fig. 5E). In tail fin, the levels of Me3H3K27 before T₃ treatment were higher than in the brain for both genes (Fig. 5F). Moreover, T₃ decreased Me3H3K27 occupancy on TRβ and TH/bZIP T₃REs to reach levels measured in brain (Fig. 5F). Me3H3K27 was observed on DIMTL1 but was not affected by T₃ (Supplemental Fig. 5). At the CZs, the levels of Me3H3K27 without T₃ were in the same order of magnitude as levels at TRβ T₃RE and were lower compared with TH/bZIP T₃RE (Fig. 5, panel E for brain and panel F for tail fin). Only in tail fin at the TH/bZIP CZ, did Me3H3K27 levels decrease significantly after T₃ treatment (Fig. 5F). Thus, as for H3K4 methylation, Me3H3K27 level on TRβ and TH/bZIP loci showed gene-specific variations and to a lesser extent, tissue specificity.

T₃-response gene-specific histone demethylase (HDM) and methyltransferase requirements

The differences in H3K4 and H3K27 methylation status for TRβ and TH/bZIP in response to T₃ suggest a functional interplay between histone methyltransferases and HDMs. Histone methyltransferase and HDM can be targeted to promoters to influence transcriptional status (21). To better understand the requirements of histone lysine modifying enzymes on T₃-response gene regulation, the effect of an HDM inhibitor, pargylin, was investigated. Pargylin is a selective monoamine oxidase inhibitor that also blocks LSD1 (22), the first lysine HDM described (23). However, adding pargylin to aquarium water led to tadpole death within minutes. Therefore, pargylin was used on cultured tail tips that show similar T₃-induced regression and T₃-target gene responses to those in the whole organism (24).

After 48 h, tail fins were used for RNA extraction and RT-qPCR analysis. rpl8 was used again as a stable internal control gene for normalization, after Normfinder (13) analysis (data not shown). Its expression was not affected by T₃ and/or pargylin treatment (Supplemental Fig. 7A). As expected, T₃ treatment increased T₃-response gene expression (Fig. 6A). Pargylin treatment alone also increased T₃-response gene expression but to a lesser extent than T₃. To highlight the specific effect of pargylin on T₃-response genes, expression of non T₃-regulated gene was also analyzed. On the four genes used for Normfinder analysis, none of their expressions were affected by the presence of pargylin (Supplemental Fig. 7A). Pargylin+T₃ cotreatment led to increased TRβ and TH/bZIP mRNA levels. Interestingly, RNA levels after pargylin+T₃ are not different to T₃ alone, suggesting lack of synergy. The results could also indicate that HDM requirement is part of the T₃ regulation processes. ChIP assay on similarly treated tails was used to determine Me2H3K4 and AcH4 levels on TRβ and TH/bZIP T₃REs. The Me2H3K4 mark was chosen because LSD1 specifically demethylates mono- and dimethylated H3K4 (23). As observed in whole tadpole, in tail culture, T₃ decreased Me2H3K4 on TRβ (Fig. 6B), but increased it on TH/bZIP (Fig. 6C). Pargylin alone significantly increased Me2H3K4 on TH/bZIP (Fig. 6C), but not on TRβ, where methylation was unaffected (Fig. 6B). Pargylin and T₃ cotreatment increased Me2H3K4 on TH/bZIP (Fig. 6C), but not on TRβ (Fig. 6B). Looking at the CZ, pargylin treatment did not change significantly the level of Me2H3K4 marks for either T₃-response gene (Fig. 6, B and C, respectively). No pargylin effects were seen on DIMT1L (Supplemental Fig. 7B).

Fig. 6.

Pargylin, an inhibitor of HDM, increases T₃-response gene expression and increases TR binding to T₃RE. The tail tips of stage NF 52–53 tadpoles were isolated and placed in cultures dishes. After 48 h with or without 10 nm T₃ and/or 100 nm pargylin (P) for 48 h, tail fins were isolated for RNA extraction or ChIP studies. A, Pargylin treatment increases T₃-response genes expression. The RNA were analyzed by RT-qPCR for the presence of TRβ and TH/bZIP expression. Gene expressions were normalized against the value for rpl8 RNA. The average values and sem of three independent experiments are expressed as multiples of induction, where 1 is equal to expression in the absence of T₃ or pargylin treatment (Ct). B, Effect of T₃ and pargylin treatment on Me2H3K4 occupancy at TRβ locus. After 2 d of culture, tail fins were isolated from tail explants for chromatin extraction. The presence of Me2H3K4 on T₃RE located near the TSS and 2000 bp upstream the TSS of TRβ were analyzed by ChIP (ChIPαMe2H3K4). The average values and sem of at least four independent experiments are expressed as percent of input. C, T₃ and pargylin effect on TH/bZIP T₃RE H3K4 dimethylation. ChIP was done using antibodies against Me2H3K4 and primers around TH/bZIP T₃RE and 3000 bp upstream from its TSS. D, T₃ and pargylin increases AcH4 at TRβ T₃RE. ChIP was done using antibodies against pan-AcH4 (ChIPαACH4) and primers on TRβ T₃RE region and 2000 bp upstream from its TSS. E, T₃ and pargylin effect on AcH4 occupancy at TH/bZIP T₃RE. ChIP was done using antibodies against pan-AcH4 and primers around TH/bZIP T₃RE and 3000 bp upstream from the TSS. F, Pargylin increases TR recruitment at TRβ T₃RE. ChIP was done as using antibodies against TR (ChIPαTR) and primers on TRβ T₃RE region. G, Pargylin treatment increases TR binding at TH/bZIP T₃RE. ChIP was done using antibodies against TR and primers around TH/bZIP T₃RE. Statistical significance as compared with untreated animals is indicated as *, P < 0.05; **, P < 0.01; or ***, P < 0.001. P, Pargylin; Ct, Control tadpoles corresponding to untreated animal.

Pargylin effects on AcH4 were also investigated using ChIP with a pan-AcH4 antibody. In tail fin, T₃, pargylin and T₃+pargylin treatment proportionally increased AcH4 both on the TRβ T₃RE (Fig. 6D) and TH/bZIP T₃RE (Fig. 6E). These results are in accordance with the increase of TRβ and TH/bZIP mRNA levels measured in each condition. Moreover, the increase of AcH4 after inhibition of HDM by pargylin, strongly underlines a functional interplay between demethylation and deacetylation.

Finally, as the data showed first, that H3K4 methylation levels correlated with TR binding to T₃RE and second, that pargylin increased H3K4 methylation (at least at TH/bZIP T₃RE; Fig. 6C), ChIP assays were used to study the recruitment of TR in tail fin after pargylin treatment of isolated tail in culture. Pargylin treatment alone significantly increased TR binding on TRβ (Fig. 6F) and TH/bZIP T₃RE (Fig. 6G). The present data thus show a strong link between H3K4 methylation and TR binding.

Discussion

The current model to explain the mechanism of repression and activation of transcription by TR can be summarized as follows: In the absence of ligand, TR, constitutively present on the T₃RE, can repress gene expression by recruiting a corepressor complex, whereas liganded TR recruits coactivator complexes for gene activation. Corepressor and coactivator complexes induce chromatin remodeling to mediate TR regulation of transcription with special emphasis on histone acetylation. The data reported here allow us to modify this conceptual model. Our results highlight first, that initial premetamorphic expression levels of direct T₃-responses gene are gene specific and dictate the magnitude of regulation by T₃. Second, although both genes have similar TR binding sites, the TR occupancy on T₃RE differs, indicating that TR binding to DNA in the absence of hormone is not a obligatory phenomenon. Third, the variations of AcH4 and AcH3K18 levels are in agreement with the basic model for TR action on positively regulated genes. However, the pattern of AcH3K9 modification is gene and tissue specific with no clear correlation with gene expression. Future studies will be necessary to test the role of this histone mark. Fourth, Me3H3K27 levels negatively correlate with the level of gene expression. Finally, direct T₃-response gene specificity corresponds to a complex pattern of H3K4 methylation strongly linked to TR binding. This methylation mark thus represents a fundamental controller of TR recruitment, permitting fine tuning of transcription. We propose that these diverse molecular codes could be the key to differential modulation of the sensitivity of T₃-response genes in a physiological context. Promoter-specific regulation is particularly interesting in the context of metamorphosis. Potentially, the different gene-regulatory mechanisms could be correlated with the multiple T₃-induced cellular responses underlying tissue remodeling during this complex developmental phenomenon.

Me3H3K27, a mark to impose stronger repression

Our results point to a novel, and tissue-specific, role of H3K27 methylation status for regulating T₃-response genes. The Polycomb repressive complex 2 is involved in Me3H3K27 acquisition (25) and the trithorax complex is involved in removing this mark (26). Little is known about their targeting mechanisms and their role in nuclear receptor function. To date, only Polycomb has been shown to be involved in regulation by the retinoic acid receptor (27). The present data show that H3K27 methylation has a strong impact on TR function. First, the high level of Me3H3K27 occupancy correlates with the lower expression of both genes in the tail fin compared with the brain without T₃. Second, the tail fin-specific, 5-fold decreases of Me3H3K27 after T₃ treatment correlate with the large induction mRNA levels induced by T₃. Me3H3K27 and Me3H3K4 are thought to be opposing modifications with regard to their consequences for gene activity (28). However, certain genes have been found to contain both marks simultaneously (28), with the repression function of Me3H3K27 dominating the activating Me3H3K4 modification (29). Most of these genes are poised for transcription, with preinitiated RNA PolII at their promoter and no Me3H3K36 on the transcribed DNA locus (30). This mechanism does not apply to TRβ. Interestingly, and consistently with our data for TRβ, these loci are transcribed at low levels (31). Furthermore, cobinding of HDAC1 and p300 at the promoter of a silent locus has been observed (31) as well as for TRβ T₃RE (10). A final possibility is that the two modifications do not co-occur within the same cell. Such a situation has been described for Xenopus (32).

H3K4 methylation status, a code to govern specificity of T₃-response gene regulation

Of all the histone modifications analyzed, the H3K4 methylation marks showed the strongest gene-specific pattern. The level of Me3H3K4 on the TRβ locus was constantly about 10 time higher than that of TH/bZip, and this high level was maintained despite T₃-induced changes in transcription. The presence of Me3H3K4 in premetamorphic tadpoles could either be independent of transcription initiation or be a historical mark of previous transcriptional initiation (30). T₃ decreases Me2H3K4 marks on TRβ gene, as observed for estrogen receptor with its ligand (33). The levels resulting from T₃ treatment are consistent with activation of transcription, because Me2H3K4 is totally absent on repressed targets (23). This selective variation might be dictated by the surrounding histone marks around the promoter. One such mark could be AcH3K9, known to block complete H3K4 demethylation (34). Thus, decreased Me2H3K4 occupancy without leading to gene repression could limit gene activation. Attenuation of promoter activation by Me2/Me3H3K4 was also recently described (35). Many histone methylations serve as recruitment marks for specific proteins, again limiting activation by implementation of enzymatic activities at the site of recruitment. TH/bZIP shows an unusual pattern, with absence of Me3H3K4 with or without T₃ and an increase of Me2H3K4 after T₃ treatment. Whereas Me2H3K4 and Me3H3K4 are usually concordant at most genes, a subset of differentially methylated, Me2H3K4+/Me3H3K4− genes exist (36). The common feature of this gene category could be the presence of an enhancer region that imposes stricter control of initiation than that typically observed at other genes where transcription is controlled by initiation and elongation. Such a mechanism would fit with the tight repression of TH/bZIP in absence of T₃ and its strong activation with T₃.

TR binding and specificity of T₃-response gene regulation

In vitro studies in a variety of cell types from different vertebrate species show that TR binds to T₃RE independent of hormone presence. Given the significant levels of TRα expression in premetamorphic tadpoles (37), it is surprising to find first, no binding or very low binding of TR to the TH/bZIP T₃RE and second, increased binding of TR to both TH/bZIP and TRβ T₃RE after addition of T₃. The abundance of TRα vs. TRβ and their specificity of DNA binding could account for such difference. Further studies will be necessary to address this point. TRβ level of expression alone does not appear to be responsible for the difference in TR occupancy before hormone treatment. In presence of T₃, the overexpression of TRβ leads to similar levels of TRβ binding on both TRβ and TH/bZIP T₃RE, confirming the need of sufficient TRβ expression for binding to the TH/bZIP T₃RE (9) and suggesting that T₃ increases TRβ binding capacity. Thus, the type of T₃RE, TRβ level of expression, and T₃ could all contribute to controlling TR binding.

These results are also pertinent to interpretation of the dual function model for the role of TR during amphibian development (37). In this model, T₃-response genes would be repressed by unliganded-TR during premetamorphosis, with T₃ playing a critical, activating role during metamorphosis. As TR binding was not detected on the TH/bZIP promoter during premetamorphosis, the very low levels of expression of TH/bZIP before metamorphosis would appear not to involve TR-dependent repression. However, TR binding to TH/bZIP T₃RE might be less stable or more dynamic, precluding detection by ChIP. Moreover, repression could also implicate other transcription factors. Moreover, some data support a working model in which initial unstable recruitment of silencing mediator of retinoid and TR/ nuclear receptor corepressor complexes via unliganded TR generates a histone code that serves to stabilize their own recruitment (38, 39) and thus to propagate repression (40).

Chromatin configuration also contributes to controlling TR binding. Levels of methylated H3K4 correlate with TR occupancy. Pharmacological modification of the methylation status of H3K4 confirmed these observations because increasing H3K4 methylation leads to an increase of TR binding at the T₃RE. Thus, H3K4 methylation could provide landmarks for high TR binding. A similar process has been described for estrogen receptor where histone H3K9 methylation modulates estrogen receptor binding to DNA in absence of 17β-estradiol (33). Because methylated H3K4 colocalizes with enhancer regions (41, 42), this mark could have a functional role for binding of transcription factors. This observation has considerable broad significance because a large number of putative sequences with regulatory elements characteristics can be found using bioinformatic tools, but not all of them bind transcription factors. Thus, combining sequence searches and chromatin landscape data will provide new opportunities to define true, functional regulatory elements. Now, the main question is what factors determine the presence of the methylation on H3K4 before metamorphosis. It is possible that DNA methylation is a downstream event to safeguard the silent status of a promoter (43). Future studies will be necessary to test this hypothesis and will complete this first in vivo analysis during vertebrate development for any nuclear receptors of the link between histone marks and T₃-response gene regulation or TR binding to DNA.

Materials and Methods

Animals and treatments

X. tropicalis adult frogs were obtained from NASCO (Fort Atkinson, WI) and maintained at 24 C in aquatic housing system (MPAquarien, Rockenhausen, Germany). Mating was induced by injection of 200 U of human chorionic gonadotropine for females and 100 U for males (Chorulon; Intervet, Beaucouze, France). Tadpoles were raised at 26 C. X. laevis frogs and tadpoles were raised as previously described (44). For T₃ treatment, tadpoles were maintained 48 h, in 5 liters with 10 nm T₃ (Sigma, St. Quentin Fallavier, France). The water was changed daily. Tadpoles were killed by decapitation after anesthesia (0.01% MS222, Sigma) before brain isolation and tail fin dissection. Tadpoles were staged according to the normal table of X. laevis (Daudin) of Nieuwkoop and Faber (45). Animal care was in accordance with institutional and national guidelines.

RNA isolation and RT-qPCR analysis

For each physiological conditions, brain or tail fin isolated from groups of 10 tadpoles were collected, flash frozen, and stored at −80 C. The tissue lysis was performed in 500 μl of RNAble (Gex-ex-T00–0U; Eurobio, Les Ulis, France) with one bead (INOX AISI 304 grade 100 AFBMA) using Tissue Lyser II apparatus (QIAGEN, Courtaboeuf, France) for 1 min at 30 Hz. The lysed tissues were mixed with chloroform and incubated on ice for 5 min before centrifugation (12,000 × g, 15 min, 4 C). The supernatant was subjected to RNA purification with RNeasy MinElute Cleanup kit according to manufacturer (ref : 74204, QIAGEN). RNA concentration was measured by optic density. RNA quality was estimated by microcapillary electrophoresis using Qiaxcel (QIAGEN). Potential contamination by genomic DNA was removed using DNAse treatment as described by the provider (Turbo DNA free; Ambion, Applied Biosystems, Courtaboeuf, France). Reverse transcription (RT) was done as previously described (37) using Superscript III reverse transcriptase (Invitrogen, Fisher Scientific, Illkirch, France). RT products were analyzed by real-time qPCR performed on an ABI 7300 (Applied Biosystems). Primers were designed using Primer express (Applied Biosystems). The list of used primers is given in Supplemental Table 1. Prism 7300 system software (Applied Biosystems) was used to analyze the results. In Fig. 1, the data are presented in means of Log(2^−ΔCT) and sem (CT, cycle time). The raw data of three to six biological replicates are first normalized on the endogenous control gene rpl8 (ΔCT: mean CT rpl8 minus mean CT gene of interest) followed by Log transformation, so the variances between groups succeed the F test. In Fig. 6, the results are presented in fold increase (means and sem). The raw data are normalized on the endogenous control gene rpl8 and on the nontreated sample control by the 2^−ΔΔCT method. For statistical analysis, the Log of the normalized data were (after Kolmogorov and Smirnov normality test) subjected to a one-sample two-tailed t test (α = 5%).

ChIP and qPCR analysis

Dissected brains or tail fins, up to 30 to 50 mg, from seven euthanized tadpoles were used to isolate chromatin. ChIP was done as previously described (46) with slight modifications. ChIP products were analyzed by real-time qPCR as previously described for RT products analysis. Primers are given in Supplemental Fig. 1. The results were expressed as percent of input and presented as means sem of at least four independent experiments. Statistical analyses were performed with a paired one-tailed t test (α = 5%). A detailed procedure is presented in Supplemental Fig. 8.

Antibodies

Antibody against TR was previously described and used in ChIP assay (37). Antibody against RNA PolII (CTD4H8) was from Epigentec (A-2032; Euromedex, Souffelweyersheim, France). All the antibodies against chromatin modifications pan-AcH4 (06-866), AcH3K9 (07-352), AcH3K18 (07-354), Me2H3K4 (07-030), Me3H3K4 (07-473), Me2H3K9 (07-441), Me3H3K9 (17-625), and Me3H3K27 (17-622) were from Upstate Biotechnology (Millipore, Saint-Quentin en Yvelines, France). Antibody against Me3H3K36 (ab1785) was from Abcam (Paris, France). Antibody against GFP was from Torrey Pines Biolabs (CliniSciences, Montrouge, France).

Transgenesis

F₀ transgenic animals were generated by restriction enzyme-mediated integration nuclear transplantation (47) with slight modifications (16). The introduced DNA is a plasmid expressing the X. laevis TRβ protein fused to the C-terminal part of GFP from the cytomegalovirus promoter (16) or from neurotubulin β (NβT) promoter. The NβT-GFP-TRβ construct was generated by inserting the GFP-TRβ AgeI-DraIII insert in front of the NβT promoter of the NβT-GFP construct previously described (15). where the cytomegalovirus promoter was replaced by the NβT promoter. To select founder transgenic animals, fluorescent proteins were observed on anesthetized (0.01% MS222) intact embryos and latter tadpoles, under a dissecting scope equipped with fluorescent light. F₀ generation was used to investigate and confirm the functionality of the GFP-fused TRβ protein compared with TRβ wild-type protein as well as the comparable levels of expression between groups of transgenic animals (16). F₁ generation was used for ChIP assay analysis. Mating with two transgenic animals was induced by injection of 500 U of human chorionic gonadotropine for females and 200 U for males. F₁ transgenic tadpoles with similar level of GFP expression were carefully selected as F₀ tadpoles.

Tail culture

Tail cultures was done as previously described (23). Samples were placed in culture dishes (12-well plates) containing the culture medium and maintained at 24 C. The tails were treated for 48 h with or without 10 nm T₃ (Sigma) and/or 0.1 mm Pargylin (P8013, Sigma) immediately at the start of the culture. Prior chromatin extraction and RNA isolation tail fins were isolated.

NURSA Molecule Pages:

• Nuclear Receptors: TR-α | TR-β;

• Ligands: Thyroid hormone.