INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Thyroid hormone receptors (TRs)¹ are transcription factors that belong to the steroid/thyroid nuclear receptor superfamily, which are encoded by two distinct genes, c-erbA and c-erbA. Alternative splicing of the  gene primary transcripts at their 3' extremity generates mRNAs encoding TR1, which binds thyroid hormone (triiodothyronine; T3), as well as the c-erbA2 and c-erbA3 variants, which do not. The  gene expresses two T3-binding isoforms, TR1 and the minor form TR2 (reviewed in Ref. 1). In the adult animal, the 1 and 1 receptor mRNAs are found in all T3-sensitive tissues, albeit with differences in their relative abundance (2, 3). In contrast, expression of the two receptor isotypes is strikingly different during brain development. Whereas TR1 expression predominates at early stages and remains steady, TR1 expression is initially very low but exhibits a dramatic transient increase during the critical period of the central nervous system maturation (4-6). At this moment, 1 transcripts prevail in the proliferation layer, while those of 1 are localized in the differentiation layer of the developing rat cerebral cortex. These patterns of expression are consistent with specific functions of the two receptors in brain development (7-9).

TR transactivation activities are mediated through binding to specific DNA elements called thyroid hormone response elements (TREs) present in the promoter of target genes. TREs are composed of two hexameric half-sites closely related to the consensus AGGTCA, arranged either as a direct repeat (DR), or as palindromic or inverted palindromic (IP) arrays (10). An intriguing feature of TRs is their ability to interact with these TREs as monomer, homodimer, or heterodimer with the 9-cis-retinoic acid receptor (RXR). The relatively high binding affinity of the TR/RXR heterodimer for the TREs (11-17), the broad expression of RXR (18-20), and the requirement of RXR for in vitro ligand-dependent transcription by TRs (21) emphasize the predominant role of this heterodimer in transcription activation. In contrast, the role of TR homodimers that are disrupted upon T3 addition in in vitro binding assays (22-24) and that of TR monomers whose binding requires an extended TRE half-site are not yet elucidated. The recent identification of nuclear receptor co-repressors and co-activators suggests that TR-mediated transcriptional regulation is a stepwise process. It is thought that, in the absence of T3, TR/RXR binds to the TRE and represses basal promoter activity by interacting with silencing factors such as N-CoR and SMRT (25, 26). The addition of T3 would release co-repressors, thereby allowing the interaction of the receptors with co-activators such as SRC-1 and p300/CBP (27, 28). Thus, transactivation of target genes would consist of both a derepression and then a further stimulation of the target promoter.

While there is evidence for specific roles of TR1 and TR1 in mediating T3 effects as mentioned above, the mechanism by which the two receptors discriminate between specific target genes remains largely unexplained. To address this question, we analyzed two T3-regulated genes, the malic enzyme (ME) and the myelin basic protein (MBP) genes. The ME gene is involved in lipid metabolism, more particularly in lipid synthesis. Its expression is ubiquitous but is up-regulated by T3 in rat liver, kidney, and heart. So far, no specific developmental expression pattern of this gene has been described. The product of the second gene studied herein, MBP, is a major protein constituent of the myelin sheet surrounding the axons. Expression of this gene is nervous system-specific and developmentally regulated. Up-regulation by T3 is mainly observed during the high myelinating activity at the critical period of brain maturation (29). Interestingly, the onset of myelination is marked by transiently elevated levels of TR1 mRNA that parallel those of MBP mRNA, suggesting the involvement of TR1 in the regulated expression of the MBP gene (30). Moreover, when analyzed in a reporter gene transfection assay, TR1 is a more potent T3-dependent inducer of the MBP promoter activity than TR1. In contrast, the ME promoter is more efficiently stimulated by TR1 as compared with TR1 in response to T3 (31). This differential regulation of these two genes by TR1 and TR1 is thus physiologically relevant and provides a model to gain insight into the mechanisms of TR isotype-specific action.

In this study, we demonstrate that in NIH3T3 cells the TREs of the ME and MBP genes (ME_TRE and MBP_TRE) are not functionally equivalent. The ME_TRE, which is a direct repeat motif with a 4-base pair gap between two hexamers (DR4, Fig. 1A) (32), mediates high T3-dependent response by both TR1 and TR1. In contrast, the MBP_TRE, which consists of an inverted palindrome formed by two hexamers spaced by 6 base pairs (IP6, Fig. 1A) (31), confers a poor TR1-mediated transactivation as compared with TR1. This poor TR1 activity correlates with its ability to efficiently bind the MBP_TRE as monomer. We thus further explored the characteristics of the TR1 monomer in terms of DNA sequence requirements, functional properties, and co-factor interactions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Plasmids-- The construction of chloramphenicol acetyltransferase (CAT) reporter genes, MBP(256/+3)-CAT and ME(315/+3)-CAT, has been previously described (29, 33), and these genes are here referred to as MBP256 and ME315, respectively. The MBP/ME_TRE and the MBP256/5'm reporter gene were obtained by PCR mutagenesis of the MBP256 reporter according to the protocol described for PCR technology (34). The following primers were used for the MBP256/5'm construct: MBP187(+), 5'-AATGTTGAACGAGCTGAGGACACGGC-3'; MBP176(), 5'-AGCTCGTTCAACATTGTTCTGGATCGC-3'; MBP260(+), 5'-GCTCTAGACTCTAGGCCTCGTACA-3'; MBP+9(), 5'-CCGCTCGAGCGGATGGTCTGAAGCTCGT-3'. The following primers were used for the MBP/ME_TRE construct: p1, 5'-GGACGTTCGGGTTAGGGGAGGACACGGCGGTGACA-3'; p2, 5'-CCCTAACCCCAACGTCCTCTGGATCGCATCTGCCT-3'; p3, 5'-AACTGCAGGTCGACTCTAGAAGGCCTCGTACAGGC-3'; p4, 5'-AATGGCGAGGATCCCTCGAGTGAAGCTCGTCGGAC-3'.

Transfections-- NIH3T3 cells were maintained in culture and transiently transfected by electroporation as described previously (32). Briefly, each cuvette contained 4 × 10⁶ cells at a density of 12.5 × 10⁶ cells/ml and a total of 70 µg of plasmid DNA (30 µg of CAT reporter plasmid, 12 µg of each expression vector (or as indicated in the figure legends), 1 µg of pCMVgal as internal control for transfection efficiency, and pUC19 to complete to 70 µg of transfected DNA). After electroporation, cells were distributed in four 60-mm dishes containing the transfection medium supplemented with 100 nM T3 or with vehicle alone. After 48 h, cell extracts were prepared and -galactosidase and CAT activities were determined as described previously (32).

Electrophoretic Mobility Shift Assays-- Proteins for electrophoretic mobility shift assays (EMSA) were obtained either by in vitro transcription and translation using the rabbit reticulocyte lysate system (Promega) or from Xenopus oocytes whole cell extracts prepared as described below. RXR was prepared from nuclear extract of Sf9 cells infected with a recombinant baculovirus overexpressing mouse RXR as described (38). The fusion proteins GST-N-CoR and GST-SRC were expressed in Escherichia coli and purified as described by the manufacturer (pGEX vectors; Amersham Pharmacia Biotech). The probes used correspond to the double-stranded oligonucleotides indicated in the figures flanked on their 5' side and 3' side by BamHI and HindIII sequence, respectively, and labeled with [-³²P]dATP.

Microinjection of Xenopus Oocytes-- The preparation of Xenopus stage VI oocytes and the microinjection procedure were essentially as described in Wong et al. (39) using the Nanoliter Injector (A203XVZ, World Precision Instruments, Inc.). The rTR1 (3 ng/oocyte) and human RXR (3 ng/oocyte) mRNAs were injected into the cytoplasms of oocytes, which were incubated at 18 °C overnight in MBSH buffer (10 mM Hepes, pH 7.6, 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.82 mM MgSO₄, 0.41 mM CaCl₂, 0.33 mM Ca(NO₃)₂) supplemented with antibiotics (50 units/ml of ampicillin and streptomycin). Then the ME315, MBP256, or MBP256/5'm single-stranded reporter genes were microinjected into oocyte nuclei (1 ng/oocyte) with the internal control mycI(6-9)CAT (0.2 ng/oocyte) as indicated in the legend of Fig. 6. The oocytes were incubated in MBSH buffer supplemented with antibiotics during 6 h and then collected for transcription analysis (see below). For EMSA, 10 oocytes were injected, and after an overnight incubation, whole cell extracts were prepared as described previously (39).

Primer Extension Analysis-- A group of 15-20 oocytes was collected per sample and homogenized in 10 µl/oocyte of 0.25 M Tris-HCl at pH 8.0. The equivalent of 10 oocytes was used for RNA purification and transcription analysis. Total RNA was prepared from the homogenate using the RNazol^TM method (Tel-Test Inc.) as described by Wong et al. (39). Half of the RNA sample (5 oocytes) was analyzed by primer extension.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

The ME and MBP Genes Are Differentially Transactivated by TR1 and TR1-- To detect functional differences between the two main TR subtypes in the regulation of the ME and MBP promoters, we used transfection experiments in NIH3T3 cells. These cells express RXR as most cell lines do but have very low TR mRNA levels (data not shown). The two CAT reporter constructs used contain either the promoter region of the ME gene up to 315 (ME315), which encompasses the TRE between 287 and 260 (32) or the MBP promoter up to position 256 (MBP256) including the TRE located between positions 195 and 165 (31) (Fig. 1A). The T3-mediated response of the ME315 reporter gene increased with the amounts of cotransfected expression plasmids for TR1 or TR1 (Fig. 1B). However, repression by the unliganded TR1 (mean value of 58% of repression when the cells were transfected with 3 µg of expression plasmid, p < 0.05) was more important than in the presence of TR1 (mean value of 73% for the corresponding point, statistically not significant). The overall T3-induced response was also more potent with TR1 than with TR1, a difference that cannot be attributed to TR1 and TR1 differences in expression levels (see below). As seen in Fig. 1C, the MBP256 reporter gene poorly responded to T3 and TR1, and this response was not improved by transfection of higher amounts of TR expression vector. In contrast, TR1-mediated activity was dose-dependent, leading to a significant difference in T3 response between the two TR subtypes. Because of the opposite results obtained when using the ME and the MBP reporter genes, we can rule out an apparent functional difference that would be due to a difference in the level of receptor expression. This was further assessed by a RNase protection assay that showed that the level of TR1 and TR1 mRNA after transfection were similar (data not shown). Moreover, the same pattern of activation and repression was observed using different vectors for expressing TRs (pRSV instead of pSG5PL; data not shown). Thus, in transfected NIH3T3 cells, TR1 induces a strong T3 response of ME315 but a weak activation of MBP256. In contrast, TR1 mediates an efficient T3 stimulation of both ME and MBP reporter genes.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Differential transactivation properties of TRs. A, schematic representation of the CAT reporter genes. The respective positions and sequences of the METRE, organized as a direct repeat (DR4), and MBPTRE, organized as an inverted palindrome (IP6), are given within their wild type promoter context (ME315, MBP256). B, comparison of the T3 response mediated by TR1 (filled bars) and TR1 (striped bars) using the ME315 reporter construct. NIH3T3 cells were cotransfected with the reporter construct (7.5 µg/dish) and increasing amounts of pSG5TR1 or pSG5TR1 in the presence (+) or absence () of 100 nM T3 as indicated. The total amount of expression plasmid was adjusted to 3 µg with the empty vector pSG5. A pCMV (0.25 µg/dish) was included as an internal control. CAT activity levels were standardized to -galactosidase expression. The basal activity of the reporter plasmid in the absence of receptor and T3 was set to 1. Results are the mean ± S.E. of at least three independent experiments. Statistical analyses show that the repression by the unliganded TR1 is significant (Student's t test value, p = 0.02) in contrast to that mediated by TR1. The overall difference of the T3 response mediated by TR1 and TR1 is also significant (p < 0.05). C, comparison of the T3 response mediated by TR1 (filled bars) and TR1 (stripped bars) using MBP256. The experimental conditions used are as described in B. The difference in the T3 response mediated by TR1 and TR1 is significant at 1.5 and 3 µg of expression plasmid (Student's t test value, p < 0.01). D, effect of the TRE structure in the T3 response mediated by TRs. Transfection in NIH3T3 cells of MBP256 or MBP/METRE CAT reporter genes with pSG5, pSG5TR1, and pSG5TR1 (3 µg/dish), as described for B is shown. Both MBPTRE/TATA and METRE/TATA CAT reporter genes were cotransfected in NIH3T3 cells with pSG5 or pSG5TR1 (3 µg/dish).

Role of the TRE Structure in the Transactivation Potency of TR1 and TR1-- One distinct feature of the ME and MBP promoter that may participate in their different patterns of regulation by T3 concerns the TRE structure, which corresponds to a DR4 in the ME promoter and to an inverted palindrome with a 6-base pair spacing (IP6) in the MBP promoter (Fig. 1A). To evaluate whether the nature of the TRE was responsible of the poor TR1-mediated T3 responsiveness of the MBP promoter, we changed the MBP_TRE for ME_TRE within the context of the MBP promoter (MBP/ME_TRE; Fig. 1A). As seen in Fig. 1D, TR1 mediated a similar T3 responsiveness of the wild type and mutated promoter, in contrast to TR1, which was significantly more efficient when the ME_TRE replaces the MBP_TRE. A distinct role of these two TREs in the potency of TR1-mediated transcriptional activation was further analyzed in the context of a common minimal promoter that only contains the TATA region from 33 to +3 of the MBP promoter (MBP_TRE/TATA and ME_TRE/TATA). In this context again, the MBP_TRE solely conveyed a low activation by TR1 as compared with that mediated by ME_TRE (5-fold versus 15-fold, respectively). We thus can conclude that in these different promoter contexts, the MBP_TRE element is by itself responsible of the poor TR1-mediated response of the MBP promoter, while it mediates efficient regulation by TR1.

Binding Pattern of TR1 and TR1 on ME_TRE and MBP_TRE-- To evaluate whether the specific pattern of T3 response through particular response elements by TR1 may reflect differences in DNA binding properties of the TR subtypes, EMSAs were performed with labeled double-stranded oligonucleotides corresponding to the two TREs. Neither TR1 nor TR1 was able to bind to the ME probe in absence of RXR. The addition of RXR to the TRs resulted in the formation of a major complex that corresponds to the TR/RXR heterodimer (Fig. 2A). The absence or presence of T3 in the reaction did not change the strength of this signal (data not shown). Binding of TRs to the MBP_TRE was quite different from this pattern. In absence of RXR, TR1 formed two complexes: a fast migrating complex that corresponds to a TR1 monomer and a slower complex corresponding to a TR1 homodimer (Fig. 2B). The addition of T3 triggered the disappearance of the upper complex (Fig. 2B, compare lanes 3 and 4) while provoking a small but reproducible downshift of the faster migrating complex, consistent with their identification as TR homodimer and monomer, respectively (22-24). Increasing amounts of TR1 resulted in a parallel increase of the intensity of both the monomer and the homodimer bands (Fig. 2B, lanes 7-9), indicating the absence of cooperativity for the formation of the homodimer. In the presence of both TR1 and RXR, a third complex was detected. This complex corresponds to a TR1/RXR heterodimer and was not disrupted by T3 (Fig. 2B, lanes 5 and 6). As seen in Fig. 2C, TR1 was also able to form a retarded complex on the MBP_TRE in absence of RXR. Its slow migration and T3-induced disruption identified it as a TR1 homodimer (Fig. 2C, lanes 1 and 2). No complex corresponding to a TR1 monomer was observed (Fig. 2C, lanes 5-7), although trace amounts were occasionally detected in presence of high concentrations of TR1. In the presence of RXR, TR1/RXR heterodimers were formed and showed a slightly increased mobility in presence of T3 (Fig. 2C, lanes 3 and 4). The addition of more RXR to TR1 or TR1-containing binding reactions with the MBP_TRE probe resulted in the exclusive detection of TR1/RXR or TR1/RXR complexes (data not shown). In summary, TR1/RXR and TR1/RXR complexes form with a similar efficiency on both TREs. However, a striking difference resides in the binding of TR1 monomers and TR1 homodimers to the MBP_TRE, whereas such complexes are not seen on the ME_TRE.

View larger version (37K): [in this window] [in a new window]   Fig. 2.   Distinct in vitro binding pattern of TRs to the METRE and MBPTRE. EMSAs using in vitro translated TR1 or TR1 in the presence or absence of Sf9-expressed recombinant RXR and of 100 nM T3 as indicated were incubated with a radiolabeled oligonucleotide encompassing the METRE sequence (A) or the MBPTRE sequence (B and C). In B, lanes 7-9, increasing amounts of in vitro translated TR1 (6, 15, and 30 fmol, respectively) show the noncooperative binding of TR1 monomers in the absence of RXR. In C, lanes 5-7, increasing amounts of in vitro translated TR1 (6, 15, and 30 fmol, respectively) do not allow the detection of TR1 monomer. In all panels, TR indicates the monomer complex, TR/TR indicates the homodimer complex, and TR/RXR indicates the heterodimer complex.

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Competition and kinetic DNA binding studies of TR1 monomers, homodimers, and heterodimers. EMSAs using in vitro translated TR1 incubated with the radiolabeled MBPTRE probe were performed in the presence of Sf9-expressed recombinant RXR so that the three complexes (monomer, homodimer, and heterodimer) could be observed. A, after 15 min of incubation, competitions were performed during 10 min in the absence of or with a 1.25-, 2.5-, 5-, and 12.5-fold molar excess of unlabeled MBPTRE (lanes 1-5, respectively). B, after 15 min of incubation, a 50-fold molar excess of unlabeled MBPTRE was added to the reaction, which was stopped after 0, 1, 3, 5, and 10 min (lanes 1-5, respectively). The graphs on the left correspond to the densitometric analysis of the autoradiographs shown on the right.

Binding of TR1 as Monomer Requires a Specific Extended Half-site-- The occurrence of a high monomer binding efficiency that was TRE and TR subtype-specific prompted us to identify the corresponding sequence requirement present in MBP_TRE. Mutation of the downstream half-site with respect to the orientation of the TRE in the wild-type promoter (MBP 3'm; Fig. 4A) completely abolished TR1 as well as TR1 binding in the absence and in the presence of RXR (Fig. 4B, lanes 5-8). In contrast, mutation of the 5' half-site (MBP 5'm; Fig. 4A) did not affect TR1 monomer binding, but neither TR1 nor TR1 homodimers could form (Fig. 4B, lanes 9 and 11). These results further excluded the possibility that the complex migrating as homodimer corresponds to the independent and adjacent binding of two TR molecules. However, most unexpectedly, TR1/RXR and TR1/RXR did bind to MBP 5'm although to a lesser extent than to the wild type MBP_TRE (Fig. 4B, compare lanes 10 and 12 with lanes 2 and 4). The same results were obtained when the mutation introduced in the 5' half-site consisted of random nucleotides (Fig. 4B, compare MBP 5'm lanes 9-12 with MBP 5'm/N lanes 13-16). This ruled out a possible artifact due to the nature of the mutations and clearly demonstrated that the TR1 monomer specifically binds to the 3' half-site. The importance of the three nucleotides flanking the AGGACA core hexamer of this half-site in 5' was then assessed by mutating them within the full-length MBP_TRE (MBP agt, Fig. 4A). These mutations abolished detectable monomer binding and thus demonstrated that the nonanucleotide CTGAGGACA is required in the wild-type MBP_TRE for TR1 monomer binding (Fig. 4B, lanes 17-20). Moreover, the formation of TR/RXR heterodimer complexes onto MBP agt dramatically decreased, indicating that heterodimers also make specific contacts with these nucleotides. In summary, these mutagenesis experiments demonstrated that (i) only the 3' extended half-site of MBP_TRE is capable of binding the TR1 monomer and (ii) this specific half-site also strongly participates in the binding of TR/RXR heterodimer.

View larger version (63K): [in this window] [in a new window]   Fig. 4.   Specific binding of TRs on the 3' half-site of MBPTRE. A, sequences of wild type and mutated MBPTRE. The arrows underline hexameric half-sites, and regions enclosed in shaded boxes correspond to mutated sequences. B, EMSAs using in vitro translated TR1 and TR1 incubated in the presence or absence of Sf9-expressed recombinant RXR, as indicated. The radiolabeled probes are as indicated below each panel. For comparison, lanes 1-4 correspond to the observed complexes bound onto the wild type TRE, run with the same extracts and on the same gel as lanes 9-12. In lanes 2 and 4, heterodimers appear as a doublet, whose fainter and faster migrating complex is composed of TR and a protein from Sf9 extract, most likely USP (72).

Analyses of the Transcriptional Activity of TR1 Monomer-- In a first set of experiments, we introduced the mutation corresponding to MBP 5'm into the wild-type MBP promoter, creating MBP256/5'm, and used this reporter gene in transfection assay of NIH3T3 cells. This reporter gene bearing only a half-site TRE exhibited a weak but significant response to T3 (Fig. 5). The T3 response by both TR1 and TR1 was affected by the mutations and corresponded to the reduction of the efficiency of the TR/RXR heterodimer binding onto MBP 5'm (see Fig. 4B). Thus, the correlation between the intensity of heterodimer binding and the transactivation level suggested that, in transfected cells, the binding of heterodimers onto the half-site rather than the monomer formation, which was not affected by the mutation, may account for the residual response.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Functional analyses of the TR1 monomer binding site. NIH3T3 cells were cotransfected with the wild type MBP256 or with the mutation MBP256/5'm reporter plasmid. pSG5, pSG5TR1, and pSG5TR1 (3 µg/dish) were cotransfected in the presence or absence of 100 nM T3 as indicated. pCMV (0.25 µg/dish) was included as an internal control. CAT activity levels were standardized to -galactosidase expression. The basal activity of the reporter plasmid in the absence of receptor and T3 was set to 1. Results are the mean ± S.E. of three independent experiments.

View larger version (36K): [in this window] [in a new window]   Fig. 6.   TR1 and TR1 expressed in Xenopus oocytes mediates a T3 response. A, in vitro transcribed mRNAs (lanes 1-12) corresponding to TR1 or TR1 and/or human RXR were injected into Xenopus oocyte cytoplasm (mRNA). Whole cell extracts were prepared as described under "Materials and Methods" and analyzed by EMSA with the radiolabeled MBPTRE probe (lanes 1-6) and the MBP5'm probe (lanes 7-12). B, B', C, and D, Xenopus oocytes were microinjected into the cytoplasm with TR1 mRNA (3 ng/oocyte), TR1 mRNA (3 ng/oocyte), with or without RXR mRNA (3 ng/oocyte) as indicated at the top of each panel. After an overnight incubation, nuclei were microinjected with MBP256 (B, B'), MBP256/5'm (C), or ME315 (D) CAT reporter genes (1 ng of single-stranded DNA/oocyte). Oocytes were further incubated 24 h in the presence (+) or absence () of 100 nM T3. Primer extension analyses of total RNA from 10 oocytes/sample were performed with a radiolabeled CAT primer for the reporter gene activities together with the H4 primer as internal control.

The TR1 Monomer Interacts Neither with N-CoR Nor with SRC-1-- The interaction of TR with co-repressors such as N-CoR or SMRT was shown to participate in the transcriptional repression mediated by TR in the absence of its ligand (25, 26), while the co-activator SRC-1 plays a role in the positive enhancement of gene expression mediated by liganded TR (27). Both factors are present in NIH3T3 cells and thus should be involved in TR1 monomer activity, if there is any. Using EMSAs, we thus assessed the specific interacting properties of DNA-bound TR1 complexes with these co-factors. Because the high molecular weight of N-CoR and SRC-1 precluded their use as full-length proteins in EMSA, we expressed the mouse N-CoR domain known to interact with nuclear receptors (25) as a GST fusion protein (Fig. 7A). Interestingly, as shown in Fig. 7, TR1 bound as a monomer on MBP_TRE as well as on MBP 5'm was not able to bind N-CoR, since no supershift was observed (Fig. 7B, lanes 1-7). In contrast, the DNA-bound TR/RXR heterodimer was supershifted in the presence of N-CoR, due to the formation of a ternary complex. T3 addition led to the disruption of this higher order complex and to the reappearance of the heterodimer (Fig. 7B, lanes 8-10).

View larger version (40K): [in this window] [in a new window]   Fig. 7.   TR1 monomer bound to MBPTRE does not interact with N-CoR. A, schematic representation of the murine N-CoR cDNA. Two amino-terminal repressing domains (RDI, nucleotides 117-1053; RDII, nucleotides 2373-3165) are represented by shaded boxes, and the C-terminal interacting domain is shown by a black box (ID, nucleotides 6834-7017). The N-CoR region from nucleotide 6715 to 7618, encompassing the interaction domain with TR, was subcloned into the pGEX2T vector. The GST fusion protein was expressed in E. coli and used in EMSA. B, EMSAs using in vitro translated TR1 incubated in the presence or absence of GST-N-CoR (1 µg of partially purified protein), 100 nM T3, and Sf9-expressed recombinant RXR as indicated. The radiolabeled probe was MBPTRE (lanes 1-3 and 8-10) or MBP 5'm (lanes 4-7). A specific interaction of DNA-bound TR/RXR with N-CoR is released by T3, which also provoked a small downshift of the TR/RXR complex.

View larger version (42K): [in this window] [in a new window]   Fig. 8.   TR1 monomer bound to MBPTRE does not interact with SRC-1. A, schematic representation of the SRC-1 cDNA. The different regions of the SRC-1 cDNA (A, nucleotides 67-617; B, nucleotides 617-1259; C, nucleotides 1259-2330; D, nucleotides 2330-3391) were subcloned into the pGEX 5 × 3 vector. The GST fusion proteins were expressed in E. coli and used in EMSA. B, EMSAs using in vitro translated TR1 incubated in the presence or absence of GST-SRC-1 clone A or clone B (1 µg of partially purified protein) in the presence or absence of 100 nM T3 and RXR, as indicated. The radiolabeled probes were MBPTRE (lanes 1-8) or MBP 5'm (lanes 9-10).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In this study, we report on the differential regulation of the MBP promoter by TR1 and TR1 in mammalian cells. We show that the weakness of the T3 response mediated by TR1 correlates with its ability to form monomers on the MBP_TRE. This TR1 monomer interacts neither with the N-CoR nor with the SRC-1 domains known to interact with the receptor. These functional characteristics may have profound implications with respect to the T3 regulation of the MBP gene during development.

Binding of TR Monomer Has Specific Sequence Requirements-- Binding of TR as a monomer was first described using the synthetic TREpal as probe (40). Later, the observation that nucleotides flanking the core hexamer might affect TRE activity (41) was at the basis of an unbiased search for an optimal DNA binding sequence for TR1 monomer, which resulted in the octameric sequence 5'-TAAGGTCA (42). Further work with diverse synthetic octameric sequences determined the 5'-(T/C)(A/G)AGGTCA-3' motif as consensus half-site for the TR1 monomer (43). Among the natural elements, the rat growth hormone, the chick lysozyme, and the rat - and -myosin heavy chain genes also contain positive TREs comprising an optimal or suboptimal octamer sequence to which TR binds as monomer (23, 44-49). In the natural elements that we analyzed, the MBP_TRE sequence contains the nonamer CTGAGGACA whose disruption in the 5' part or in the core sequence results in the loss of TR1 monomer binding (see Fig. 4B). In contrast, the sequence of the two half-sites that composes the ME_TRE and taken as nonamers are TTGGGGTTA and GGGAGGACA. None of them can bind TR1 monomer in the conditions used in the present work, indicating that TR1 monomer binding has strict requirements covering at least 8 or 9 nucleotides. Some reports, however, indicated that purified E. coli-expressed TR may bind as a monomer to ME_TRE (44, 45, 48). Two nonexclusive hypotheses can explain the apparent discrepancies with our own work. First, the amount of TR expressed in E. coli may be much higher than that obtained in reticulocyte lysate, allowing the detection of low affinity complexes. Second, post-translational modifications that do not occur in E. coli could be important for TR binding properties (50) as shown for c-erbA2, which binds as monomer only after mutation of two phosphorylation sites (51).

TR/RXR Can Bind an Extended Half-site-- Surprisingly, the nonamer present in the MBP_TRE, to which TR1 monomer binds, also allowed the formation of TR/RXR heterodimer but not of TR/TR homodimer complexes (see Fig. 4B). This suggests that, since TR and RXR heterodimerize in solution (13), the strong and specific anchorage of TR onto its half-site allows weaker or nonspecific DNA contacts by RXR. This is indeed substantiated by methylation interference assays, which showed that RXR contacts the 5' half-site of a DR4 with rather weak bonds, as compared with those of TR with the 3' half-site (52, 53). This ability of a heterodimer to bind to an extended half-site may also explain some unexpected observations of TR/RXR binding to direct repeats elements with a variety of spacing lengths between the half-sites (54, 55). This is reminiscent of the RXR/Nurr1 heterodimer, which can trigger a 9-cis-retinoic acid responsiveness through an 8-base pair monovalent site (56). Activation by TR through an octamer element has been previously reported and was attributed to the monomer activity (57). Herein, we demonstrated that TR/RXR heterodimers are probably responsible for the extended half-site activity that we detected.

DNA-bound TR Monomer Does Not Interact with NCoR or SRC-1-- Recent studies on the mechanism of action of nuclear receptors identified factors with which they interact (58). More precisely, two co-repressors N-CoR and SMRT interact with the unliganded TR, while TR interaction with the co-activators SRC-1 and p300/CBP is T3-dependent (25, 26, 28, 59, 60). These interactions were characterized using yeast two-hybrid systems or pull-down assays and showed that TR or TR/RXR interacted with these co-factors in solution. Herein, we demonstrated by EMSA that TR monomer bound to MBP_TRE interacts neither with GST-N-CoR nor with GST-SRC-1. The DNA-bound heterodimer, in contrast, forms a ternary complex with N-CoR in the absence of T3 (herein and Ref. 25) and interacts with SRC-1 in presence of T3. While this manuscript was in preparation, Zamir et al. (61) also showed that DNA-bound TR monomer does not interact with GST-NCoR.

Functional and Structural Differences of TR1 and TR1-- So far, all T3-responsive genes analyzed in transfections exhibit an equal or slightly better responsiveness to TR1 than to TR1 as emphasized herein with the ME gene. Besides a T3-independent TR1-specific activation of the pcp-2 gene (66), the MBP promoter is the first example of a clear contrast between TR1- and TR1-mediated T3 responsiveness. In parallel, previous EMSAs performed on various synthetic and natural elements indicated that TR1 more readily homodimerizes than TR1 and that TR1 binds more easily as monomer than TR1 (23, 40, 44, 46, 67). In this respect, the binding pattern of TR1 and TR1 to the MBP_TRE is paradigmatic.

Transcriptional Regulatory Patterns Mediated by TR Differ in NIH3T3 Cells and in Xenopus Oocytes-- Xenopus oocytes provide a model in which RXR has not been detected (39). The results we obtained in this model are surprising. First, without co-injecting RXR mRNA and in the absence of T3, both TR isotypes led to a strong repression of reporter genes carrying various TREs: an IP6, a half-site, or a DR4. That a TR monomer could be responsible for such an activity was not supported by the fact that TR1 onto MBP_TRE and TR1 or TR1 onto the ME_TRE could not form monomer. The same reasoning applies to the involvement of homodimers that were exclusively seen with the MBP256. Second, in this model, no differences between TR1 and TR1 were observed. One hypothesis would be that, in oocytes, TR irrespective of the complex that it forms on DNA is able to recruit partners possibly different from those known in mammalian cells. They are unlikely to be RXR homologues, since no RXR-like activity was detected in the oocyte extracts obtained after TR mRNA injection (Fig. 6A; see also Ref. 39). Little is known yet about the co-repressors and co-activators that are present in Xenopus oocytes. Their analysis could be highly valuable to understand the TR mechanism of action in these cells.

TR1 and TR1 in the Regulation of the MBP Gene during Brain Development-- In rat embryos from E14 onwards, TR1 mRNA is detectable at high levels during all developmental stages of the brain, whereas TR1 mRNA is low or undetectable. The insensitivity of MBP to T3 regulation at that time suggests that TR1 may indeed occupy the MBP_TRE sequence as monomer during development. After birth, TR1 mRNA expression peaks during the first 3 weeks (2, 7, 8). This correlates with high myelinating activity and high T3-dependent MBP expression. Our experiments suggest that after birth, the nonproductive promoter occupancy by TR1 monomer is challenged by the rise in TR1 concentration. In this model, the relative intracellular concentration of TR and RXR is an important determinant of the hormonal response as kinetic studies emphasize the higher affinity and better stability of the TR/RXR heterodimer complex versus homodimer and monomer (herein and Refs. 40 and 54). Three RXR isotypes are expressed in mammals, and all are able to form heterodimers with TR (18). RXR and RXR transcripts have been detected by in situ hybridization in the central nervous system of the mouse embryo from day 10.5 onwards and in the adult (18, 70). Thus, both TRs and RXRs are present during the myelination period; however, their relative levels are not known. It must also be kept in mind that the interaction of TRs with the MBP_TRE in neural tissue may involve other tissue-specific factors in addition to RXR, which might result in a transactivation pattern different from that we have observed in NIH3T3 cells. In view of the rather high level of TR1 expression in the developing brain, associated with the high efficiency with which TR1 monomers bind to the MBP_TRE, the physiological relevance of TR1 monomers deserves further consideration.