INTRODUCTION

Thyrotropin-releasing hormone (TRH)¹ is the hypothalamic tripeptide stimulating thyrotropin (TSH) synthesis and secretion in the anterior pituitary gland (1). TRH is derived from a large precursor protein, preproTRH (ppTRH), by posttranslational processing and enzymatic modification (2). The expression of the ppTRH gene in the parvocellular subdivision of the paraventricular nucleus in the hypothalamus is negatively regulated by thyroid hormones (3, 4). Recently, we and others demonstrated that the promoter activities of human and rat ppTRH genes were directly suppressed by triiodothyronine (T₃) (5, 6, 7). These results indicated that the ppTRH gene belongs to a family of T₃-responsive genes, including the  and  subunits of TSH genes, -myosin heavy chain gene, and the epidermal growth factor receptor gene, whose expression are negatively regulated by T₃ at the level of gene transcription (8, 9, 10, 11, 12, 13, 14, 15).

Thyroid hormone receptor (T₃R) is the ligand-dependent transcriptional factor that activates or inhibits gene transcription basically by binding to the specific cis-acting DNA elements, so-called thyroid hormone response elements (T₃REs), located in the 5-flanking regions of T₃-responsive genes (16, 17). To date, the negative T₃REs (nT₃REs) in several gene promoters have been characterized (8, 9, 10, 11, 12, 13, 14, 15). In the TSH subunit and glycoprotein  subunit genes, the nT₃REs reside near the TATA box and contain a single hexamer sequence matching perfectly or loosely to the consensus core half-site motifs for T₃RE (8, 9, 10, 11, 12, 13). It has therefore been speculated that T₃Rs binding to the inhibitory T₃REs mediate transcriptional suppression by steric interference with the transcription initiation machinery. In contrast, the nT₃RE in epidermal growth factor receptor gene is localized between 112 and 76, and it contains a single TRE half-site-like motif (GGGACT) which overlaps with the Sp1 binding site (15). The nT₃RE weakly binds T₃R homodimers, and an addition of nuclear extracts from HeLa cells augments T₃R binding with formation of T₃R/T₃R auxially protein heterodimers in the EMSA (18). It has been recently demonstrated that the hormone-responsive element of the Rous sarcoma virus long terminal repeat consisting of a novel inverted palindrome with a 6-bp spacer (ATTAGG) is activated by unliganded TR1, but not by TR1, and the effect is reversed by an addition of T₃ (19). The T₃RE in the Rous sarcoma virus promoter binds TR homodimers and RXR/TR heterodimers (19). These results indicate that T₃Rs are capable of exerting inhibitory regulation of gene transcription through structurally divergent nT₃REs, which seems to be analogous to transcriptional activation by T₃ through a variety of positive TREs, such as the palindromic T₃RE (PAL), the direct repeat of the half-sites with a 4-base pair (bp) spacer (DR4), and the inverted palindrome with a 6-bp spacer (IP-6) (20). However, the detailed mechanism by which T₃ regulates gene transcription positively or negatively remains to be determined.

A detailed analysis of the human ppTRH gene promoter recently demonstrated that both the binding of T₃Rs as heterodimers with the 9-cis-retinoic acid receptors (RXR) to the element that contains a single TRE half-site located 60 bp upstream of the TSS and the binding of T₃R monomers to the two TRE half-sites positioning downstream of the TSS were required to exhibit full-inhibition of the human ppTRH gene promoter by T₃ (7). Moreover, one of the functional T₃R isoforms, TR1, preferentially mediates the negative regulation (7). These results provided direct evidence of the involvement of RXR and T₃R isoform specificity in the negative regulation of the human ppTRH gene promoter by T₃.

In order to gain further insight into the negative gene regulatory action by T₃, we cloned a 2-kb fragment of the 5-flanking region of the mouse ppTRH gene and sought to identify the cis-acting elements necessary for the negative regulation by T₃. The present results indicate that the mechanism involved in the inhibitory regulation of the mouse ppTRH gene promoter by T₃ is distinct from that of the human ppTRH gene.

MATERIALS AND METHODS

Cloning of the 5-Flanking Region of the Mouse PreproTRH Gene

A mouse TT2 cell genomic library (kindly provided by T. Aizawa) (21) was screened by the standard method described elsewhere using a ³²P-labeled mouse ppTRH cDNA previously isolated in this laboratory (22). The isolated clone was mapped by restriction endonuclease digestions in combination with Southern blot analyses. Two overlapping DNA fragments containing the promoter region and a part of the 5-untranslated region (HS2.0) and 5-untranslated region, the first exon and a part of the first intron (SB2.0) excised with HindIII and SalI or SalI and BamHI digestion, respectively, from the isolated clone were subcloned into pGEM 4Z (Promega). The nucleotide sequence for both strands of these clones were determined by the dideoxy chain termination method (23) using Sequenase Version 2 (U. S. Biochemical Corp.). Nucleotide sequences were analyzed using a computer program, GENETYX (Software Development Co., Ltd.).

Identical oligonucleotides with HindIII linkers at 5 and 3 ends were used for the construction of heterologous promoters and EMSA. The nucleotide sequences of the upper strands of these oligonucleotides were as follows (underlines indicate TRE half-site motifs, and lowercase letters indicate HindIII linker sequences): PAL, 5-AGCTATCGCGA-3; IP-5, 5-agctTAGACCGGCTCCA GCAGGa-3; TRH1/2, 5-agctTGCCCCTCCCCGCCACA-3; TRE1/2, 5-agcttCCCGCCACAGGGGCCGCTGTCTCGA-3; Site 5&6, 5-agctTGGATTCTGGAGAGCCTTGCAGACTCTACCCAGCCA-3; PAL6, 5-agCTTGCTGC CTCGGGA-3.

Using HS2.0 and SB2.0 fragments, a series of plasmids containing various lengths of the promoter and 5-untranslated regions of the mouse ppTRH gene were constructed in pGEM 4Z or 11Zf (Promega) by appropriate restriction enzyme digestions or amplification by polymerase chain reactions. The DNA fragments excised from these pGEM vectors were subsequently inserted into the multiple cloning sites of a reporter plasmid, pA₃Luc, which contains the firefly luciferase cDNA as a reporter (kindly provided by W. M. Wood) (24). The heterologous promoters, IP-5-, TRE1/2-, Site 5&6-, and PAL6-TKLuc plasmids were constructed by ligation of the double-stranded oligonucleotides described above into the unique HindIII site of the pT109Luc, which possesses the minimal promoter of Herpes Simplex virus thymidine kinase gene (25). The orientation and number of inserts in these reporter constructs were verified by dideoxy sequencing. Heterologous constructs that possessed a single copy of an individual oligonucleotide in correct orientation were used for transfection experiments.

GH₄C₁, GH₃, and CV1 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, penicillin (100 units/ml), streptomycin (100 µg/ml) (Life Technologies, Inc.), and amphotericin B (0.25 µg/ml) (Sigma). Cells were split 24 h before transfection into 60-mm tissue culture dishes in subconfluence. Transient transfection was performed in triplicate plates in all experiments by the calcium phosphate precipitation method using Cellphect (Pharmacia Biotech Inc.) with 3 µg of reporter constructs. For cotransfection experiments, 150 ng of expression vectors for the mouse TR1, 1, or 2 driven by Rous sarcoma virus promoter (pRSVmTR1, 1, and 2 kindly provided by W. M. Wood) (26, 27) were transfected with TRH reporter constructs. Glycerol shock was performed 16 h after transfection for 2 min except for CV-1 cells, and the cell culture medium was changed to Dulbecco's modified Eagle's medium without phenol-red supplemented with 10% fetal bovine serum treated with AG1-X8 resin (Bio-Rad) and activated charcoal (Sigma) to remove thyroid and steroid hormones. Cells were incubated for an additional 48 h with or without T₃ (10 nM) or all-trans-retinoic acid (1 µM). Luciferase assays were carried out as described previously (6, 28). In brief, cells were rinsed twice with 5 ml of phosphate-buffered saline and harvested with 400 µl of a buffer containing 1% Triton X-100, 25 mM glycylglycine, 15 mM MgSO₄, 4 mM EGTA, and 1 mM dithiothreitol. Luciferase activities were quantitated with 300 µl of cell extracts by integration of light readings over 10 s on a standard luminometer (Chiba Corning) after injection of 0.1 ml of an assay buffer containing 0.2 mM luciferin (Wako), 15 mM K₂HPO₄, 2 mM ATP, and 1 mM dithiothreitol. Protein concentrations were determined by the Bradford method using bovine serum albumin (Sigma) as a standard (29). Luciferase activities were normalized by protein concentration and expressed as light units/µg protein.

The double-stranded oligonucleotides described above were radiolabeled by fill-in reaction of the 5 overhangs with [-³²P]dCTP (3000 Ci/mmol, Du Pont NEN) using a Klenow fragment of DNA polymerase I (Takara) and were purified using Sephadex G-25 columns (Boehringer Mannheim). Human TR1 and RXR were synthesized from non-linearized peA101 and pBS-RXR (kindly provided by R. M. Evans) (30, 31), respectively by in vitro translation using TNT-coupled rabbit reticulocyte lysates (Promega) according to the supplier's manual. Synthesis of proteins in expected molecular weights was confirmed by labeling of in vitro translated products with [³⁵S]methionine and cysteine (Expre³⁵S³⁵S, DuPont NEN), followed by SDS-polyacrylamide gel electrophoresis analysis (data not shown). Binding reactions were performed for 20 min at room temperature in a total volume of 20 µl with 1 µg of poly(dI-dC) (Pharmacia), 20 mM HEPES (pH 7.6), 50 mM KCl, 20% glycerol, 1 mM dithiothreitol, 100,000 cpm of purified probes, and 4 µl of in vitro synthesized reaction mixtures. For competitive experiments, 200-fold molar excess of cold oligonucleotides were included unless otherwise indicated. For supershift experiments, 2 µl of the specific antibody raised against a synthetic peptide corresponding to the N-terminal region of the TR1 (amino acid 62-82) was incubated with binding reaction mixtures for an additional 20 min at 4 °C (Affinity BioReagent). Binding reaction mixtures were loaded on 5% nondenaturing polyacrylamide gels and were separated in 0.5 × TBE buffer (20 × TBE: 1 M Tris, 1 M boric acid, and 20 mM EDTA-Na₂) at 250 V for 80 min at room temperature. Autoradiography was carried out for 16-48 h with an intensifying screen at 80 °C.

RESULTS

Characterization of the 5-Flanking Region of the Mouse PreproTRH Gene

By screening of a mouse genomic library with a mouse hypothalamic ppTRH cDNA probe, we isolated an approximately 10-kb mouse ppTRH gene possessing a 6-kb 5-flanking region, three exons interrupted by two introns, and an entire 3-untranslated region.² To study the regulation of the mouse ppTRH gene promoter by T₃, we further characterized a 2-kb 5-flanking region. To date, the nucleotide sequences of 5-flanking regions of the rat and human ppTRH genes have been reported up to 494 and 243 bp, respectively, from the position of TSS (32, 33). We have therefore sequenced the longest 5-flanking region of mammalian ppTRH genes (Fig. 1). Sequence analyses revealed that a putative TATA box (TATAA) was located in the position similar to that of the rat and human ppTRH gene promoters. A putative Sp1 binding site (GGGCGG) found in the human and rat TRH gene promoters was also conserved in the mouse gene 117 bp upstream of the TSS. No CCAAT box was found, as is the case with other species. Two octameric sequences homologous to the consensus binding site (ATTTGCAT) of POU homeodomain proteins (34) were found at positions 1174 (ACTTGCAT) and 635 (ATTTGCCT). The consensus half-site sequences for the T₃RE ( or ) were found at three separate positions in different arrangements. An inverted palindrome with a 5-bp spacer (CTCCA, designated as IP-5) was found at position 1876, which resembled the alignment of the positive T₃RE identified in chick lysozyme gene promoter (CAGCTG) (35). An octameric T₃RE (, designated as TRE1/2) was found at position 57. Moreover, a palindrome with a 6-bp spacer (TGCCTC, designated as PAL6) was identified at position +102, which overlapped with the first exon-intron junction and resembled the alignment of T₃RE identified in herpes simplex virus thymidine kinase promoter (CGCGTG) (36). Two TRE half-sites positioned downstream of the TSS in the human ppTRH gene (GGGTCC and TGACCT) (7) were not conserved in the mouse gene. A putative half-site sequence for the glucocorticoid response element (TGTTCT) found in the rat and human ppTRH gene promoters was also conserved in the mouse gene at position 208. The overall sequence homologies to the 5-flanking regions of rat and human ppTRH gene were 80.9% and 50.8%, respectively.

We first examined expression of the reporter construct, in which the 1893/+127 fragment (including the promoter, 5-untranslated regions, and a part of the first intron of the mouse ppTRH gene) was fused to the firefly luciferase cDNA, using transient transfection into GH₃ and GH₄C₁ cells. These pituitary tumor cell lines have been known to express functional T₃Rs endogenously and are utilized extensively to study negative regulation of promoter activities by T₃ (10, 11, 12). When chimeric constructs were transfected into two cell lines in parallel, approximately 10 times higher luciferase activities were observed in GH₄C₁ cells than in GH₃ cells (data not shown). We therefore used GH₄C₁ cells for further experiments. As shown in Fig. 2, 10 nM T₃ suppressed luciferase activity of the longest construct (1893/+127) about 2.2-fold. In contrast, T₃ did not influence TK promoter activities significantly (Fig. 6). Moreover, 1 µM all-trans-retinoic acid did not influence the mouse TRH promoter activities (data not shown), indicating that the inhibitory effect of T₃ on the mouse ppTRH gene promoter was specific. Deletion of the upstream sequence from 1893 to 255, which contained IP-5 motif (254/+127), did not change the repression of the promoter activities by T₃. In contrast, deletion of the sequence between +12 and +127 abrogated T₃ inhibition despite the length of the upstream DNA sequence (1893/+11, 1079/+11, and 254/+11), the data suggesting that the element between +12 and +127 was necessary for the negative regulation. Deletion of the PAL6 motif (254/+87) did not significantly reduce T₃ suppression compared with the 254/+127 construct, indicating that PAL6 was not involved in the inhibitory regulation. Finally, the shortest construct (83/+46) was significantly suppressed by T₃. These data imply that the promoter-proximal element between 83 and +46 contains the DNA element necessary for the inhibitory regulation by T₃ in the mouse ppTRH gene promoter.

TR1, 1, and 2 Possess Equal Potency for the Inhibitory Effect by T₃ on the Mouse ppTRH Gene Promoter

To study whether transcriptional inhibition is mediated by T₃Rs and whether T₃R isoform-specificity is involved in the negative regulation of the mouse ppTRH gene promoter by T₃, we performed cotransfection experiments using CV-1 cells, which are known to be deficient for endogenous T₃Rs. PALTK-Luc, in which two copies of the palindromic T₃RE were ligated upstream of the TK promoter, was used as a positive control. As shown in Fig. 3A, 10 nM T₃ did not influence the basal promoter activities of PALTK-Luc or TRH-Luc (1893/+127) reporter constructs. Cotransfected mouse TR1 repressed basal promoter activities of PALTK-Luc approximately 3-fold without its ligand, and the addition of T₃ activated the transcription about 30-fold. In contrast, unliganded TR1 stimulated the basal expression of TRH-Luc reporter approximately 2-fold, and 10 nM T₃ reversed this activation. These ligand-independent and -dependent effects by T₃Rs were consistent with those observed by us with human ppTRH promoter in a neuroblastoma cell line (6). These results suggest that cotransfected T₃Rs play pivotal roles in mediating stimulatory or inhibitory effects of T₃ on the ppTRH gene transcription in CV-1 cells. However, in contrast to the human ppTRH gene regulation (7), no significant difference was observed in T₃-mediated inhibitory potency among three functional isoforms of the mouse TR1, 1, and 2 (Fig. 3B).

To examine whether T₃Rs bind to the putative T₃REs found in the 5-flanking region of the mouse ppTRH gene, EMSA was carried out using in vitro translated TR1 and radiolabeled oligonucleotides containing IP-5 (1893/1862), TRE1/2 (73/47), or PAL6 motifs (+97/+123). It has been reported that T₃R monomers bound to two TRE half-sites located downstream of the TSS in the human ppTRH gene promoter (7). Although these TRE half-sites were not conserved in the mouse gene, we examined whether T₃Rs bind to the element of the mouse gene (here designated as Site 5&6, +14/+47). An oligonucleotide containing the idealized palindromic T₃RE (PAL) was used as a positive control for EMSA. Nonspecific binding was assessed by incubation with unprogrammed reticulocyte lysates. As shown in Fig. 4A, a faint band corresponding to the TR-homodimer formation was observed on PAL, and this homodimer-DNA complex was diminished by addition of 200-fold molar excess of unlabeled oligonucleotides. Moreover, the band was supershifted by incubation with a specific antibody for human TR1, confirming the binding specificity. The apparent homodimer formation was observed also upon IP-5. In contrast, a faster migrating band representing T₃R monomer-DNA complex was observed on TRE1/2 oligonucleotide, and this complex was also supershifted. As anticipated, no significant T₃R binding was detected on Site 5&6 sequence as well as on PAL6 (data not shown). We next examined whether RXR/T₃R heterodimers bind to these oligonucleotides. As shown in Fig. 4B, strong heterodimer formation was observed on PAL. A heterodimer band with a slightly faster mobility than that formed on PAL was also detected on IP-5 with an intensity similar to that of homodimer band (Fig. 4B). Neither heterodimer formation nor RXR binding was observed on TRE1/2, Site 5&6, or PAL6 (data not shown). The presence of 100 nM T₃ dissociated T₃R homodimers, but not heterodimers, from the IP-5 (Fig. 4C). To further confirm that T₃Rs bind to the TRE half-site at 57 and not to Site 5&6, we performed competition experiments to find whether RXR/T₃R heterodimer formation on the PAL is inhibited by two overlapping fragments containing the TRE half-site (TRE1/2 and TRE 1/2) and Site 5&6. As shown in Fig. 5B, addition of TRE1/2 and TRE1/2 efficiently inhibited heterodimer binding to PAL in the manner similarly to unlabeled PAL (Fig. 5A), indicating that the overlapping 15-bp sequence in the two competitors (CCCGCTGACCTCACA) possessed a T₃R-binding site. In contrast, Site 5&6 did not inhibit heterodimer formation, further confirming that T₃Rs could not bind to this element (Fig. 5A).

Competition of RXR/T₃R heterodimer bindings to the radioactively labeled PAL by unlabeled PAL, Site 5&6 (A), TRE1/2 and TRE1/2 (B).

To examine whether putative T₃REs found in the mouse ppTRH 5-flanking region independently confer transcriptional stimulation or inhibition by T₃, heterologous promoters were constructed in which oligonucleotides carrying an IP-5, TRE1/2, Site 5&6, or PAL6 motif were fused in front of the TK promoter, and were transiently transfected into GH₄C₁ cells. As shown in Fig. 6, 10 nM T₃ stimulated luciferase activities of the PALTK about 5-fold. In contrast, no significant stimulation by T₃ was observed in TK109-Luc. Although T₃R homodimers and heterodimers bound to IP-5 in the EMSA, this element did not function as T₃REs in the heterologous construct. The TRE1/2 and Site 5&6 motifs, which were located in the sequence necessary for the negative regulation of the mouse ppTRH gene promoter by T₃, did not function independently as nT₃REs in heterologous constructs, suggesting that the TRE half-site and the element downstream of the TSS cooperatively mediate the T₃ suppression, presumably in a position-dependent manner. Moreover, PAL6-TK-Luc did not show transcriptional activation or inhibition by T₃.

DISCUSSION

In the present study, we characterized the 2-kb 5-flanking region of the mouse ppTRH gene and localized its nT₃RE in the promoter region. The 5-flanking region of the mouse ppTRH gene contained T₃RE-like motifs, which resembled previously characterized positive T₃REs, at two separate positions, an inverted palindrome with a 5-bp spacer (IP-5) at 1876 and a palindromic TRE with a 6-bp spacer (PAL6) at +102. The natural inverted palindromic sequences with differential spacing have been found in the chicken embryonic myosin gene (IP-2), the human TR promoter (IP-5), the chicken lysozyme silencer element (IP-6), and myelin basic protein (IP-6) (37). Among these natural IP motifs, only the IP-6 in two chicken genes are able to mediate transcriptional activation by T₃. In direct repeat response elements, the length of the spacer region is well known to be critical to determine receptor specificity, as by the 3-to-5 rule (38). A recent detailed analysis of IP-type T₃REs demonstrated that the spacing in IP response elements is also important to determine the binding characteristics of T₃R homodimers and RXR/T₃R heterodimers by EMSA (37). IP-5 motif found in the mouse ppTRH gene effectively bound both T₃R homodimers and RXR/T₃R heterodimers, although it mediated neither transcriptional activation nor inhibition by T₃ in heterologous promoters. These findings, taken together, suggest that RXR/T₃R heterodimers bound to IP-5 in the mouse ppTRH gene promoter might be unable to interact effectively with other factors involved in T₃R-mediated transcriptional regulation by its inappropriate steric conformation. Another putative T₃RE, PAL6, was found at +102 in the present study, which overlapped with the first exon-intron junction. It has been reported that the PAL6 motif in the viral TK promoter mediates transcriptional activation by T₃ in GH₄C₁ cells and binds T₃R monomers and homodimers (36). However, no T₃ activation or repression in heterologous construct was observed on PAL6-TK Luc in the mouse ppTRH gene, consistent with its absence of T₃R binding capacity in the EMSA. These results suggest that the sequence between two half-sites and/or 5- and 3-flanking sequences of the half-sites might be crucial to mediate T₃ action on the PAL6 motif.

In the transfection study with CV-1 cells, cotransfected T₃Rs stimulated basal promoter activities of the mouse ppTRH gene without T₃, and an addition of T₃ repressed this basal stimulation, indicating that T₃Rs are required for the negative regulation of the mouse ppTRH gene promoter by T₃. Deletion analyses revealed that the nT₃RE of the mouse ppTRH gene was located in the promoter-proximal element between 83 to +46. The nucleotide sequence of the region upstream of the TATA box in the mouse gene was highly conserved when compared with those of rat and human ppTRH genes (Fig. 7). In this region, a perfectly matched TRE half-site (TGACCT) was conserved in all species, and the two flanking nucleotides 3 of this half-site were matched to the consensus sequence for the proposed octameric TRE half-site sequence, (T/C)(A/G)AGGTCA (39). In vitro translated T₃Rs bound to this octameric half-site of the mouse gene exclusively as monomers in the EMSA. Furthermore, two overlapping oligonucleotides containing this half-site were able to compete with RXR/T₃R heterodimer binding to the palindromic T₃RE. These results indicate that T₃R monomers binding to the octameric TRE half-site was involved in the negative regulation of the mouse ppTRH gene promoter by T₃. In contrast, it has been reported that RXR/T₃R heterodimers binding to this element in the human ppTRH gene promoter play an important role in T₃ suppression (7). Several differences in the surrounding sequence of this TRE half-site might alter the preference for the binding of T₃R monomers or RXR/T₃R heterodimers to this element. With respect to the region downstream of the TSS, the nucleotide sequence was less conserved between the human and rodent ppTRH genes (Fig. 7). Two TRE half-sites positioned at +14 (GGGTCC) and +37 (TGACCT) in the human gene, which mediate T₃ suppression in cooperation with the TRE half-site upstream of the TATA box, were not conserved in rodent ppTRH genes. Although no significant binding of T₃Rs was observed in the corresponding region of the mouse gene by EMSA, deletion analysis revealed that the DNA element between +12 and +46 was necessary for the negative regulation by T₃. Unexpectedly, heterologous constructs containing either the TRE1/2 or the downstream element placed in front of the TK promoter did not independently confer negative regulation by T₃. It has been reported that a 17-bp motif (CGCCAAGTAAG) located at the 3 end of exon1 of the rat TSH  gene containing a single copy of a hexamer TRE half-site with some degeneracy mediates T₃ inhibition in an orientation- and position-independent manner when fused to the TKCAT reporter in GH₃ cells (40). The nT₃RE of rat TSH bound in vitro synthesized T₃R monomers (40). In contrast, it has been demonstrated that T₃R monomers also bound to the octameric T₃RE half-site functioning as positive T₃REs in Drosophila SL-3 cells when fused to the TKCAT reporter gene (39). The present results showed that the binding of T₃R monomers to the TRE half-site in the mouse ppTRH gene promoter was not sufficient for T₃ suppression in GH₄C₁ cells. These results, taken together, suggest that the negative regulation of the mouse ppTRH gene promoter by T₃ could be mediated by the cooperation of T₃R monomers that bind to the TRE half-site and a novel protein binding to the element downstream of the TSS in a position-dependent manner. Interestingly, the basal promoter activity of the mouse ppTRH gene increased approximately 5-fold by deletion of the element downstream of the TSS. Moreover, nuclear proteins extracted from GH₄C₁ cells bound to the element with differential mobilities from those of T₃R monomers, homodimers, and T₃R/RXR heterodimers formed on the PAL by EMSA.³ It has been reported that T₃ binding to bacterially expressed T₃Rs increased monomeric T₃R-DNA interaction and increased the mobility of the monomer-DNA complexes, suggesting the liganded T₃R monomers undergo a T₃-induced comformational change (41). T₃ may cause a conformational change in T₃R monomers binding to the TRE half-site upstream of the TSS, resulting in stabilization of some repressor protein that binds to the proximal downstream element through direct or indirect interaction, thereby repressing transcription of the mouse ppTRH gene. It has become increasingly evident that a large number of both cellular and viral genes utilize elements that are located downstream of the TSS for transcriptional regulation (42, 43, 44). The characterization of DNA-binding protein(s) interacting with the downstream element might address the detailed mechanism for negative regulation of the mouse ppTRH gene by T₃.

T₃Rs are encoded by two distinctive genes, c-erbA  and  (30, 45). Multiple isoforms of T₃Rs are generated from two genes by alternative promoter usage and alternative splicing of the primary transcripts in human, rat, mouse, and chicken (16, 17). Expressions of these T₃R isoforms are regulated in specific temporal and spacial patterns (46), and T₃ regulates their expression differentially in different tissues (47). TR2 was initially isolated as the specific isoform expressed exclusively in the pituitary gland, and its expression was negatively regulated by T₃ (47, 48). Therefore, TR2 has long been believed to be the specific isoform involved in the negative regulation by T₃ of the pituitary TSH  and common  subunit genes. TR2 was recently reported to be expressed in the paraventricular nucleus in the hypothalamus, where expression of the ppTRH gene is negatively regulated by T₃ (49). Thus, the possibility has been raised that this T₃R isoform plays a pivotal role in negative feedback action by T₃ in the hypothalamo-pituitary-thyroid axis. In the present study, we did not observe any significant functional difference among the three mouse T₃R isoforms in the negative regulation of the mouse ppTRH gene in CV-1 cells. These results were consistent with our previous findings in the negative regulation of the human ppTRH gene promoter using rat and human T₃R isoforms in a human neuroblastoma cell line (6). However, the results in our studies contradicted those obtained by others, demonstrating that TR1 preferentially exerts inhibitory regulation of human and rat ppTRH gene transcription by T₃ compared with TR1 in CV-1 cells and in primary culture of chick hypothalamic neurons, respectively (5, 7). The discrepancy of these results may be explained by three reasons: 1) the contents of endogenous T₃R and RXR may differ among these cell lines, 2) the mechanism involved in the negative regulation by T₃ appeared to be distinct between the human and mouse ppTRH gene, and 3) T₃R isoforms in different species may have distinctive functions by their structural differences.

In summary, the mouse ppTRH gene promoter was negatively regulated by T₃, probably through collaboration of T₃R monomers with an uncharacterized factor interacting with the proximal promoter-downstream element. The characterization of such DNA binding factor(s) may provide a new concept to elucidate the molecular mechanism by which the T₃/T₃R complex negatively regulates gene transcription.