Negative regulation of the gene for the preprothyrotropin-releasing hormone from the mouse by thyroid hormone requires additional factors in conjunction with thyroid hormone receptors.

To gain additional insights into the negative gene regulatory action by triiodothyronine (T3), we isolated a 2-kilobase pair 5′-flanking region of the mouse preprothyrotropin-releasing hormone (ppTRH) gene and characterized the DNA elements mediating inhibitory regulation by T3 in the promoter region. In GH4C1 cells, the expression of the 2-kilobase pair mouse ppTRH 5′-flanking region fused to the luciferase reporter gene occurred by transfection and was significantly suppressed by T3. In contrast, T3 suppression was not observed in T3 receptor (T3R)-deficient CV-1 cells, suggesting that T3Rs were required for the negative regulation. Cotransfected mouse T3R α1, β1, and β2 possessed indistinguishable potency for the negative regulation. Deletion analysis localized the element mediating the negative regulation to the region between −83 and +46, and the sequence downstream of the transcription start site (TSS) between +12 and +46 was found to be essential for the inhibitory regulation. In mobility shift assays, only T3R monomers bound to the element containing a T3 response element half-site at −57. No apparent T3R binding was observed to the element downstream of TSS. Neither the T3 response element half-site nor the element downstream of the TSS confer T3 suppression individually in heterologous promoters. These results indicate that the negative regulation of murine ppTRH gene by T3 might be mediated by the cooperation of T3R monomers with unknown factor(s) interacting with the element downstream of the TSS.

To gain additional insights into the negative gene regulatory action by triiodothyronine (T 3 ), we isolated a 2-kilobase pair 5-flanking region of the mouse preprothyrotropin-releasing hormone (ppTRH) gene and characterized the DNA elements mediating inhibitory regulation by T 3 in the promoter region. In GH 4 C 1 cells, the expression of the 2-kilobase pair mouse ppTRH 5-flanking region fused to the luciferase reporter gene occurred by transfection and was significantly suppressed by T 3

. In contrast, T 3 suppression was not observed in T 3 receptor (T 3 R)-deficient CV-1 cells, suggesting that T 3 Rs
were required for the negative regulation. Cotransfected mouse T 3 R ␣1, ␤1, and ␤2 possessed indistinguishable potency for the negative regulation. Deletion analysis localized the element mediating the negative regulation to the region between ؊83 and ؉46, and the sequence downstream of the transcription start site (TSS) between ؉12 and ؉46 was found to be essential for the inhibitory regulation. In mobility shift assays, only T 3 R monomers bound to the element containing a T 3 response element half-site at ؊57. No apparent T 3 R binding was observed to the element downstream of TSS. Neither the T 3 response element half-site nor the element downstream of the TSS confer T 3 suppression individually in heterologous promoters. These results indicate that the negative regulation of murine ppTRH gene by T 3

might be mediated by the cooperation of T 3 R monomers with unknown factor(s) interacting with the element downstream of the TSS.
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 reg-ulated 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 (T 3 ) (5-7). These results indicated that the ppTRH gene belongs to a family of T 3 -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 T 3 at the level of gene transcription (8 -15).
Thyroid hormone receptor (T 3 R) 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 (T 3 REs), located in the 5Ј-flanking regions of T 3 -responsive genes (16,17). To date, the negative T 3 REs (nT 3 REs) in several gene promoters have been characterized (8 -15). In the TSH␤ subunit and glycoprotein ␣ subunit genes, the nT 3 REs reside near the TATA box and contain a single hexamer sequence matching perfectly or loosely to the consensus core half-site motifs for T 3 RE (8 -13). It has therefore been speculated that T 3 Rs binding to the inhibitory T 3 REs mediate transcriptional suppression by steric interference with the transcription initiation machinery. In contrast, the nT 3 RE in epidermal growth factor receptor gene is localized between Ϫ112 and Ϫ76, and it contains a single TRE half-site-like motif (GGGACT) which overlaps with the Sp1 binding site (15). The nT 3 RE weakly binds T 3 R homodimers, and an addition of nuclear extracts from HeLa cells augments T 3 R binding with formation of T 3 R/T 3 R 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 (TGCCTTATT-AGGAAGGCA) is activated by unliganded TR␣1, but not by TR␤1, and the effect is reversed by an addition of T 3 (19). The T 3 RE in the Rous sarcoma virus promoter binds TR␣ homodimers and RXR/TR␣ heterodimers (19). These results indicate that T 3 Rs are capable of exerting inhibitory regulation of gene transcription through structurally divergent nT 3 REs, which seems to be analogous to transcriptional activation by T 3 through a variety of positive TREs, such as the palindromic T 3 RE (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 T 3 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 T 3 Rs 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 T 3 R monomers to the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) D86548.
‡ To whom correspondence should be addressed. Tel.: 272-20-8122; Fax: 272-20-8136. 1 The abbreviations used are: TRH, thyrotropin-releasing hormone; T 3 , triiodothyronine; T 3 R, T 3 receptor; T 3 RE, T 3 response element; nT 3 RE, negative T 3 response element; TSH, thyrotropin; bp, base pair(s); kb, kilobase pair(s); RXR, 9-cis-retinoic acid receptor; EMSA, electrophoretic mobility shift assay; TK, thymidine kinase; PAL, palindromic T 3 RE; TR, thyroid hormone receptor. two TRE half-sites positioning downstream of the TSS were required to exhibit full-inhibition of the human ppTRH gene promoter by T 3 (7). Moreover, one of the functional T 3 R isoforms, TR␤1, preferentially mediates the negative regulation (7). These results provided direct evidence of the involvement of RXR and T 3 R isoform specificity in the negative regulation of the human ppTRH gene promoter by T 3 .
In order to gain further insight into the negative gene regulatory action by T 3 , we cloned a 2-kb fragment of the 5Ј-flanking region of the mouse ppTRH gene and sought to identify the cis-acting elements necessary for the negative regulation by T 3 . The present results indicate that the mechanism involved in the inhibitory regulation of the mouse ppTRH gene promoter by T 3 is distinct from that of the human ppTRH gene.

MATERIALS AND METHODS
Cloning of the 5Ј-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 32 Plabeled 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 5Ј-untranslated region (HS2.0) and 5Ј-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.).
Plasmid Construction-Using HS2.0 and SB2.0 fragments, a series of plasmids containing various lengths of the promoter and 5Ј-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, pA 3 Luc, 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.
Cell Culture, Transient Transfection, and Luciferase Assay-GH 4 C 1 , GH 3 , 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 g/ml) (Life Technologies, Inc.), and amphotericin B (0.25 g/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 g of reporter constructs. For cotransfection experiments, 150 ng of expression vectors for the mouse TR␣1, ␤1, or ␤2 driven by Rous sarcoma virus promoter (pRSVmTR␣1, ␤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 T 3 (10 nM) or all-transretinoic acid (1 M). 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 l of a buffer containing 1% Triton X-100, 25 mM glycylglycine, 15 mM MgSO 4 , 4 mM EGTA, and 1 mM dithiothreitol. Luciferase activities were quantitated with 300 l 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 K 2 HPO 4 , 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/g protein.
Electrophoretic Gel Mobility Shift Assay-The double-stranded oligonucleotides described above were radiolabeled by fill-in reaction of the 5Ј overhangs with [␣-32 P]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 TR␤1 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 [ 35 S]methionine and cysteine (Expre 35 S 35 S, 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 l with 1 g 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 l 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 l of the specific antibody raised against a synthetic peptide corresponding to the N-terminal region of the TR␤1 (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 ϫ TBE buffer (20 ϫ TBE: 1 M Tris, 1 M boric acid, and 20 mM EDTA-Na 2 ) at 250 V for 80 min at room temperature. Autoradiography was carried out for 16 -48 h with an intensifying screen at Ϫ80°C.

RESULTS
Characterization of the 5Ј-Flanking Region of the Mouse Pre-proTRH 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 5Јflanking region, three exons interrupted by two introns, and an entire 3Ј-untranslated region. 2 To study the regulation of the mouse ppTRH gene promoter by T 3 , we further characterized a 2-kb 5Ј-flanking region. To date, the nucleotide sequences of 5Ј-flanking regions of the rat and human ppTRH genes have been reported up to Ϫ494 and Ϫ243 bp, respectively, from the position of TSS (32,33). We have therefore sequenced the longest 5Ј-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 Ϫ1174 (ACTTGCAT) and Ϫ635 (ATT-TGCCT). The consensus half-site sequences for the T 3 RE (AG-GTCA or TGACCT) were found at three separate positions in different arrangements. An inverted palindrome with a 5-bp spacer (TGGCCTCTCCAAGGTCA, designated as IP-5) was found at position Ϫ1876, which resembled the alignment of the positive T 3 RE identified in chick lysozyme gene promoter (TGACCCCAGCTGAGGTCA) (35). An octameric T 3 RE (TGAC-CTCA, designated as TRE1/2) was found at position Ϫ57. Moreover, a palindrome with a 6-bp spacer (AGGTAATGCCTCTG-ACCT, designated as PAL6) was identified at position ϩ102, which overlapped with the first exon-intron junction and resembled the alignment of T 3 RE identified in herpes simplex virus thymidine kinase promoter (AGGTGACGCGTGTGGC-CT) (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 Ϫ208. The overall sequence homologies to the 5Ј-flanking regions of rat and human ppTRH gene were 80.9% and 50.8%, respectively.
Negative Thyroid Hormone Response Element in the Mouse ppTRH Gene Resides near the Transcription Start Site-We first examined expression of the reporter construct, in which the Ϫ1893/ϩ127 fragment (including the promoter, 5Ј-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 GH 3 and GH 4 C 1 cells. These pituitary tumor cell lines have been known to express functional T 3 Rs endogenously and are utilized extensively to study negative regulation of promoter activities by T 3 (10 -12). When chimeric constructs were transfected into two cell lines in parallel, approximately 10 times higher luciferase activities were observed in GH 4 C 1 cells than in GH 3 cells (data not shown). We therefore used GH 4 C 1 cells for further experiments. As shown in Fig. 2, 10 nM T 3 suppressed luciferase activity of the longest construct (Ϫ1893/ϩ127) about 2.2-fold. In contrast, T 3 did not influence TK promoter activities significantly (Fig. 6). More- over, 1 M all-trans-retinoic acid did not influence the mouse TRH promoter activities (data not shown), indicating that the inhibitory effect of T 3 on the mouse ppTRH gene promoter was specific. Deletion of the upstream sequence from Ϫ1893 to Ϫ255, which contained IP-5 motif (Ϫ254/ϩ127), did not change the repression of the promoter activities by T 3 . In contrast, deletion of the sequence between ϩ12 and ϩ127 abrogated T 3 inhibition despite the length of the upstream DNA sequence (Ϫ1893/ϩ11, Ϫ1079/ϩ11, and Ϫ254/ϩ11), the data suggesting that the element between ϩ12 and ϩ127 was necessary for the negative regulation. Deletion of the PAL6 motif (Ϫ254/ϩ87) did not significantly reduce T 3 suppression compared with the Ϫ254/ϩ127 construct, indicating that PAL6 was not involved in the inhibitory regulation. Finally, the shortest construct (Ϫ83/ ϩ46) was significantly suppressed by T 3 . These data imply that the promoter-proximal element between Ϫ83 and ϩ46 contains the DNA element necessary for the inhibitory regulation by T 3 in the mouse ppTRH gene promoter. TR␣1, ␤1, and ␤2 Possess Equal Potency for the Inhibitory Effect by T 3 on the Mouse ppTRH Gene Promoter-To study whether transcriptional inhibition is mediated by T 3 Rs and whether T 3 R isoform-specificity is involved in the negative regulation of the mouse ppTRH gene promoter by T 3 , we performed cotransfection experiments using CV-1 cells, which are known to be deficient for endogenous T 3 Rs. PALTK-Luc, in which two copies of the palindromic T 3 RE were ligated upstream of the TK promoter, was used as a positive control. As shown in Fig. 3A, 10 nM T 3 did not influence the basal promoter activities of PALTK-Luc or TRH-Luc (Ϫ1893/ϩ127) reporter constructs. Cotransfected mouse TR␤1 repressed basal promoter activities of PALTK-Luc approximately 3-fold without its ligand, and the addition of T 3 activated the transcription about 30-fold. In contrast, unliganded TR␤1 stimulated the basal expression of TRH-Luc reporter approximately 2-fold, and 10 nM T 3 reversed this activation. These ligand-independent and -dependent effects by T 3 Rs were consistent with those observed by us with human ppTRH promoter in a neuroblastoma cell line (6). These results suggest that cotransfected T 3 Rs play pivotal roles in mediating stimulatory or inhibitory effects of T 3 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 T 3 -mediated inhibitory potency among three functional isoforms of the mouse TR␣1, ␤1, and ␤2 (Fig. 3B).
Electrophoretic Mobility Shift Assays-To examine whether T 3 Rs bind to the putative T 3 REs found in the 5Ј-flanking region of the mouse ppTRH gene, EMSA was carried out using in vitro translated TR␤1 and radiolabeled oligonucleotides containing IP-5 (Ϫ1893/Ϫ1862), TRE1/2 (Ϫ73/Ϫ47), or PAL6 motifs (ϩ97/ ϩ123). It has been reported that T 3 R 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 T 3 Rs bind to the element of the mouse gene (here designated as Site 5&6, ϩ14/ϩ47). An oligonucleotide containing the idealized palindromic T 3 RE (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 TR␤1, confirming the CV-1 cells were cotransfected with PALTK-Luc or mouse TRH-Luc (Ϫ1893/ϩ127) and an expression vector for the mouse TR␤1 in the presence or absence of 10 nM T 3 . The data represent mean Ϯ S.E. of triplicate determinants. The experiment was repeated twice with similar results. B, comparison of inhibitory potency of mouse T 3 R isoforms for regulation of the mouse ppTRH promoter by T 3 . CV-1 cells were cotransfected with mouse TRH-Luc and expression vectors for either mouse TR␣1, ␤1, or ␤2 in the presence or absence of T 3 . Each point of data is expressed as -fold repression in the presence of T 3 and represents the mean Ϯ S.E. from five experiments with triplicate determinants. Statistical analysis was performed by Duncan's multiple range test.
binding specificity. The apparent homodimer formation was observed also upon IP-5. In contrast, a faster migrating band representing T 3 R monomer-DNA complex was observed on TRE1/2 oligonucleotide, and this complex was also supershifted. As anticipated, no significant T 3 R binding was detected on Site 5&6 sequence as well as on PAL6 (data not shown). We next examined whether RXR/T 3 R 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 T 3 dissociated T 3 R homodimers, but not heterodimers, from the IP-5 (Fig. 4C). To further confirm that T 3 Rs bind to the TRE half-site at Ϫ57 and not to Site 5&6, we performed competition experiments to find whether RXR/T 3 R heterodimer formation on the PAL is inhibited by two overlapping fragments containing the TRE half-site (TRE1/2 and TRE 1/2Ј) and Site 5&6. As shown in Fig. 5B, addition of TRE1/2 and TRE1/2Ј 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 T 3 R-binding site. In contrast, Site 5&6 did not inhibit heterodimer formation, further confirming that T 3 Rs could not bind to this element (Fig. 5A).
Neither the T 3 R Half-site nor the Downstream Element Functioned Individually as nT 3 REs in Heterologous Promoters-To examine whether putative T 3 REs found in the mouse ppTRH 5Ј-flanking region independently confer transcriptional stimulation or inhibition by T 3 , 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 GH 4 C 1 cells. As shown in Fig. 6, 10 nM T 3 stimulated luciferase activities of the PALTK about 5-fold. In contrast, no significant stimulation by T 3 was

FIG. 4. Binding of in vitro translated T 3 R and RXR to the putative T 3 REs in the mouse ppTRH gene promoter.
A, mobility shift assays with oligonucleotides containing PAL, IP-5, TRE1/2, and Site 5&6. Radiolabeled oligonucleotides were incubated with in vitro translated TR␤1 in the presence and absence of unlabeled oligonucleotides or a specific antibody for TR␤1. Closed arrowheads indicate specific T 3 R-DNA complexes. Open arrowheads show nonspecific bindings determined by incubation with unprogrammed lysates in separate experiments. Closed arrowheads at the top, middle, and bottom indicate supershifted T 3 R-, T 3 R homodimer-and T 3 R monomer-DNA complexes, respectively. B, RXR/T 3 R hetrodimer binds to PAL and IP-5 oligonucleotides. SS indicates supershifted complex by incubation with a specific antibody for TR␤1. C, effect of T 3 on hetero-and homodimer bindings to the PAL and IP-5 oligonucleotides. Binding reaction was performed in the presence and absence of 100 nM T 3 . TRE1/2 (B). The indicated fold molar excess of cold oligonucleotides was added into binding reaction mixtures. observed in TK109-Luc. Although T 3 R homodimers and heterodimers bound to IP-5 in the EMSA, this element did not function as T 3 REs 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 T 3 , did not function independently as nT 3 REs in heterologous constructs, suggesting that the TRE half-site and the element downstream of the TSS cooperatively mediate the T 3 suppression, presumably in a position-dependent manner. Moreover, PAL6-TK-Luc did not show transcriptional activation or inhibition by T 3 .

DISCUSSION
In the present study, we characterized the 2-kb 5Ј-flanking region of the mouse ppTRH gene and localized its nT 3 RE in the promoter region. The 5Ј-flanking region of the mouse ppTRH gene contained T 3 RE-like motifs, which resembled previously characterized positive T 3 REs, at two separate positions, an inverted palindrome with a 5-bp spacer (IP-5) at Ϫ1876 and a palindromic TRE with a 6-bp spacer (PAL6) at ϩ102. 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 T 3 . 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 T 3 REs demonstrated that the spacing in IP response elements is also important to determine the binding characteristics of T 3 R homodimers and RXR/T 3 R heterodimers by EMSA (37). IP-5 motif found in the mouse ppTRH gene effectively bound both T 3 R homodimers and RXR/T 3 R heterodimers, although it mediated neither transcriptional activation nor inhibition by T 3 in heterologous promoters. These findings, taken together, suggest that RXR/T 3 R heterodimers bound to IP-5 in the mouse ppTRH gene promoter might be unable to interact effectively with other factors involved in T 3 R-mediated transcriptional regulation by its inappropriate steric conformation. Another putative T 3 RE, PAL6, was found at ϩ102 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 T 3 in GH 4 C 1 cells and binds T 3 R monomers and homodimers (36). However, no T 3 activation or repression in heterologous construct was observed on PAL6-TK Luc in the mouse ppTRH gene, consistent with its absence of T 3 R binding capacity in the EMSA. These results suggest that the sequence between two half-sites and/or 5Ј-and 3Ј-flanking sequences of the half-sites might be crucial to mediate T 3 action on the PAL6 motif.
In the transfection study with CV-1 cells, cotransfected T 3 Rs stimulated basal promoter activities of the mouse ppTRH gene without T 3 , and an addition of T 3 repressed this basal stimulation, indicating that T 3 Rs are required for the negative regulation of the mouse ppTRH gene promoter by T 3 . Deletion analyses revealed that the nT 3 RE of the mouse ppTRH gene was located in the promoter-proximal element between Ϫ83 to ϩ46. 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 3Ј of this half-site were matched to the consensus sequence for the proposed octameric TRE half-site sequence, (T/C)(A/G)AG-GTCA (39). In vitro translated T 3 Rs 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/T 3 R heterodimer binding to the palindromic T 3 RE. These results indicate that T 3 R monomers binding to the octameric TRE half-site was involved in the negative regulation of the mouse ppTRH gene promoter by T 3 . In contrast, it has been reported that RXR/T 3 R heterodimers binding to this element in the human ppTRH gene promoter play an important role in T 3 suppression (7). Several differences in the surrounding sequence of this TRE half-site might alter the preference for the binding of T 3 R monomers or RXR/T 3 R 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 ϩ14 (GGGTCC) and ϩ37 (TGACCT) in the human gene, which mediate T 3 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 T 3 Rs was observed in the corresponding region of the mouse gene by EMSA, deletion analysis revealed that the DNA element between ϩ12 and ϩ46 was necessary for the negative regulation by T 3 . 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 T 3 . It has been reported that a 17-bp motif (CGCCAGTG-CAAAGTAAG) located at the 3Ј end of exon1 of the rat TSH ␤ gene containing a single copy of a hexamer TRE half-site with some degeneracy mediates T 3 inhibition in an orientation-and position-independent manner when fused to the TKCAT reporter in GH 3 cells (40). The nT 3 RE of rat TSH␤ bound in vitro synthesized T 3 R monomers (40). In contrast, it has been demonstrated that T 3 R monomers also bound to the octameric T 3 RE half-site functioning as positive T 3 REs in Drosophila SL-3 cells when fused to the TKCAT reporter gene (39). The present results showed that the binding of T 3 R monomers to the TRE half-site in the mouse ppTRH gene promoter was not sufficient for T 3 suppression in GH 4 C 1 cells. These results, taken together, suggest that the negative regulation of the mouse ppTRH gene promoter by T 3 could be mediated by the cooperation of T 3 R 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 GH 4 C 1 cells bound to the element with differential mobilities from those of T 3 R monomers, homodimers, and T 3 R/RXR heterodimers formed on the PAL by EMSA. 3 It has been reported that T 3 binding to bacterially expressed T 3 Rs increased monomeric T 3 R-DNA interaction and increased the mobility of the monomer-DNA complexes, suggesting the liganded T 3 R monomers undergo a T 3 -induced comformational change (41). T 3 may cause a conformational change in T 3 R 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 T 3 .
T 3 Rs are encoded by two distinctive genes, c-erbA ␣ and ␤ (30,45). Multiple isoforms of T 3 Rs 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 T 3 R isoforms are regulated in specific temporal and spacial patterns (46), and T 3 regulates their expression differentially in different tissues (47). TR␤2 was initially isolated as the specific isoform expressed exclusively in the pituitary gland, and its expression was negatively regulated by T 3 (47,48). Therefore, TR␤2 has long been believed to be the specific isoform involved in the negative regulation by T 3 of the pituitary TSH ␤ and common ␣ subunit genes. TR␤2 was recently reported to be expressed in the paraventricular nucleus in the hypothalamus, where expression of the ppTRH gene is negatively regulated by T 3 (49). Thus, the possibility has been raised that this T 3 R isoform plays a pivotal role in negative feedback action by T 3 in the hypothalamo-pituitarythyroid axis. In the present study, we did not observe any significant functional difference among the three mouse T 3 R 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 T 3 R isoforms in a human neuroblastoma cell line (6). However, the results in our studies contradicted those obtained by others, demonstrating that TR␤1 preferentially exerts inhibitory regulation of human and rat ppTRH gene transcription by T 3 compared with TR␣1 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 T 3 R and RXR may differ among these cell lines, 2) the mechanism involved in the negative regulation by T 3 appeared to be distinct between the human and mouse ppTRH gene, and 3) T 3 R isoforms in different species may have distinctive functions by their structural differences.
In summary, the mouse ppTRH gene promoter was negatively regulated by T 3 , probably through collaboration of T 3 R 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 T 3 /T 3 R complex negatively regulates gene transcription.