Transcription from the Thyroid Hormone-dependent Promoter of the Xenopus laevis Thyroid Hormone Receptor βA Gene Requires a Novel Upstream Element and the Initiator, but Not a TATA Box*

The thyroid hormone receptor (TR) β genes in Xenopus laevis are regulated by thyroid hormone in all organs of an animal during metamorphosis. This autoregulation appears to be critical for systematic transformations of different organs as a tadpole is transformed into a frog. To understand this autoregulation, we have previously identified a thyroid hormone response element in the hormone-dependent promoter of the X. laevisTRβA gene. We report here the detailed characterization of the promoter. We have now mapped the transcription start site and demonstrated the existence of an initiator element at the start site critical for promoter function. More important, our deletion and mutational experiments revealed a novel upstream DNA element that is located 125 base pairs upstream of the start site and that is essential for active transcription from the promoter. Promoter reconstitution experiments showed that this novel element does not function as an enhancer, but acts as a core promoter element, which, together with the initiator, directs accurate transcription from the promoter. Finally, we provide evidence for the existence of a protein(s) that specifically recognizes this element. Our studies thus demonstrate that the TRβA promoter has a unique organization consisting of an initiator and a novel upstream promoter element. Such an organization may be important for the ubiquitous but tissue-dependent temporal regulation of the gene by thyroid hormone during amphibian metamorphosis.

Thyroid hormone (T 3 ) 1 is the causative agent of amphibian metamorphosis, a process that systematically changes most, if not all, organs of a tadpole to prepare the animal for adult terrestrial life (1)(2)(3). The hormone is known to regulate the transcription of target genes through their nuclear receptors or thyroid hormone receptors (TRs) (4 -11). Thus, it is believed that T 3 induces a cascade of gene regulation in each tissue or organ to effect the metamorphic transition (12). Many T 3 response genes have been isolated from various metamorphosing tadpole tissues, and their developmental expression profiles have implicated potential roles during metamorphosis (12)(13)(14). Among the T 3 response genes are the TR genes themselves. Two TR␣ and two TR␤ genes have been isolated from Xenopus laevis (15,16), whereas only one TR␣ gene and one TR␤ gene have been cloned from Rana catesbeiana (17,18). Consistent with their roles in mediating T 3 effects, all TR genes are expressed during metamorphosis and can be up-regulated by T 3 treatment of premetamorphic tadpoles (17)(18)(19)(20). In particular, the Xenopus TR␤A genes have been shown to be directly regulated by T 3 at the transcriptional level (21)(22)(23)(24). This T 3 regulation appears to be mediated mostly by a thyroid hormone response element (TRE), consisting of two near-perfect repeats of AGGTCA separated by 4 bp. Interestingly, promoter studies using transient transfection assays in frog tissue culture cells failed to identify any other elements necessary for the TR␤A promoter due to the lack of information on the transcription start site (22,23).
Promoters recognized by RNA polymerase II generally contain a TATA box and/or an initiator that directs specific transcription initiation. In many genes, the TATA element is the primary core element responsible for positioning the basal transcription machinery on the promoter (25)(26)(27). However, many other genes lack a TATA element and, instead, contain an initiator. The initiator encompasses the transcription start site and is sufficient to position the basal transcription complex. This specific positioning of the basal transcription machinery at a promoter by a TATA and/or initiator element allows basal transcription, which can be enhanced by transcription activators.
To determine the nature of the TR␤A promoter, we have now mapped its start site by introducing the TR␤A promoter into Xenopus oocytes and analyzing the resulting transcript by primer extension and by PCR cloning of the 5Ј-end of TR␤A mRNA from tadpoles. Deletion and mutational studies demonstrated a unique nature of the TR␤A promoter, consisting of a novel promoter element and an initiator, thus different from the two major classes of RNA polymerase II promoters mentioned above. We further show that specific proteins exist to recognize the novel promoter element, thus likely allowing specific transcription initiation and activation by TRs.

MATERIALS AND METHODS
Plasmid Constructs-The wild-type pTR␤A promoter construct was generated by cloning a 1.9-kilobase EcoRI fragment containing 1.6 kilobases of TR␤A promoter sequence and ϳ0.3 kilobase of chloramphenicol acetyltransferase gene sequence from plasmid pCAT-WT (22) into pBluescript II KS(Ϫ) (Stratagene). To make a construct of the promoter with a 5Ј-and/or a 3Ј-deletion, a 5Ј-primer (bearing a HindIII restriction site and located at an appropriate position for the desired deletion) and a 3Ј-primer (bearing a BglII restriction site and located at a downstream position for the desired deletion) were used to PCRamplify a promoter fragment from the wild-type template. The amplified fragment was then cloned into the wild-type plasmid after removing the promoter sequence by HindIII and BglII digestions. * 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.
¶ These two authors contributed equally to this work. ʈ To whom correspondence should be addressed: Lab. of Molecular Embryology, NICHD, NIH, Bldg. 18T, Rm. 106, Bethesda, MD 20892-5431. Tel.: 301-402-1004; Fax: 301-402-1323; E-mail: Shi@helix.nih. gov. 1 The abbreviations used are: T 3 , thyroid hormone; TR, thyroid hormone receptor; TRE, thyroid hormone response element; bp, base pairs; To make a specific mutation in the UPE or Inr region, a specific primer bearing the desired mutation was used to prime second strand DNA synthesis on a single-stranded wild-type construct using a mutagenesis kit (Amersham Pharmacia Biotech). The resulting DNA was transformed into Escherichia coli to obtain the mutant promoter construct. To place the UPE into a short inactive promoter (pTRp5Ј-1), a double-stranded oligonucleotide bearing the wild-type UPE (made of  5Ј-AGC TTT AAA GTA CCC CCT CTT GTA AAA TAT AAG GAT ATT  ATA A-3Ј and 5Ј-AGC TTT ATA ATA TCC TTA TAT TTT ACA AGA  GGG GGT ACT TTA A-3Ј) or a truncated UPE (mUPE) (made of 5Ј-AGC  TTG TAA AAT ATA AGG ATA TTA TAA-3Ј and 5Ј-AGC TTT ATA ATA  TCC TTA TAT TTT ACA-3) and containing HindIII overhangs was cloned into HindIII-digested pTRp5Ј-1. All clones were verified by sequencing.
Finally, a double-stranded oligonucleotide containing either the Gal4-binding site (consisting of 5Ј-AGC TTC GGA GGA GAG TCT TC-C GA-3 and 5Ј-AGC TTC GGA AGA CTC TCC TCC GA-3) (34) or a TRE (22) and HindIII overhangs was cloned into HindIII-digested pTRp5Ј-1 and pTRp5Ј-7 to generate chimeric promoters. The clones containing one or three copies of the Gal4-binding site or TRE were selected based on DNA sequencing.
The double-stranded DNA of pTR␤A promoter constructs was prepared using a QIAGEN kit as described by the manufacturer. The single-stranded DNA of pTR␤A promoter constructs was prepared from phagemids induced with VCS M13 as described (28).
Microinjection of Xenopus Oocytes-The preparation of Xenopus stage VI oocytes and the microinjection procedure were essentially as described (29). The TR␤A promoter plasmid DNA was injected (23 nl/oocyte) either as single-stranded DNA (1.15 ng/oocyte) or doublestranded DNA (2.3 ng/oocyte) into the nuclei (germinal vesicle) of the oocytes, and the indicated amounts of mRNAs for Xenopus TR␤A and RXR␣ were injected (27 nl/oocyte, 50 ng/l) into the oocyte cytoplasm. The mRNAs were usually injected 6 h before the injection of DNA. For transcription analysis, ϳ20 oocytes were injected for each sample to minimize the variations among oocytes and injections. The injected oocytes were incubated at 18°C overnight in MBSH buffer (29) supplemented with antibiotics (50 units/ml ampicillin and streptomycin) and then collected for transcription analysis.
Preparation of mRNA in Vitro-The pSP64(A)-xTR␤A, pSP64(A)-xRXR␣, and pSP64(A)-Gal4-VP16 plasmids (30,31) were linearized with EcoRI, and in vitro transcription was performed using an SP6 Message Machine kit (Ambion) as described by the manufacturer. A typical reaction with ϳ0.5 g of linearized template in a 20-l reaction yielded 10 -15 g of capped mRNA. The mRNAs were resuspended in diethyl pyrocarbonate-treated water at a final concentration of 50 ng/l and injected into the cytoplasm of groups of oocytes (27 nl/oocyte). After incubation overnight, the oocytes were collected. We examined the relative levels of expression of the TR␤A and RXR␣ receptors in the oocytes by coinjection of [ 35 S]methionine and mRNA (31), which indicated that similar amounts of receptors were produced when equal concentrations of mRNAs were injected. Therefore, in all experiments with the injection of TR␤A and RXR␣, equal amounts of TR␤A and RXR␣ mRNAs were mixed to give a final concentration as indicated for each.
Transcription Analysis-Transcription analysis by primer extension from the injected oocytes were performed essentially as described (29). Briefly, ϳ20 injected oocytes were collected for each sample, rinsed with 400 l of MBSH buffer, and then homogenized in 300 l of 0.25 M Tris (pH 8.0). To isolate RNA, 50 l of of RNazol reagent/oocyte was added to the sample, vortexed, and then incubated on ice for 15 min before centrifugation. The clean supernatant was transferred to a new tube and extracted once with an equal volume of phenol/chloroform. The RNA was then precipitated with 0.7 volume of isopropyl alcohol, rinsed with 70% ethanol, and dissolved in diethyl pyrocarbonate-treated water. For primer extension analysis, RNA from 1 or 2 oocyte eq was annealed with either the end-labeled chloramphenicol acetyltransferase primer (5Ј-GGTGGTATATCCAGTGATTTTTTTCTCCAT-3Ј, located just downstream of the TR␤ promoter sequence) or primer I (5Ј-ATCCTTATAAACGGTGAGTAGTGATGTACT-3Ј, located at ϩ109 to ϩ80) in 10 l of 0.4 M KCl at 65°C for 10 min, 55°C for 25 min, and 42°C for 5 min. Thirty microliters of reverse transcription mixture (67 mM Tris-HCl (pH 8.3), 8 mM MgCl 2 , 5 mM dithiothreitol, 1 mM dNTP mixture, 1 unit of RNasin, and 10 units of Superscript II) were then added. The reaction was incubated at 42°C for 1 h and then stopped by ethanol precipitation. The products were analyzed on a 6% sequencing gel and visualized by autoradiography. As an internal control, a histone H4 antisense primer (5Ј-GGCTTGGTGATGCCCTGGATGTTATCC-3Ј) was included in the primer extension reaction to quantify the endoge-nous H4 mRNA level.
PCR Cloning of the 5Ј-End of the TR␤A mRNA-The anchor PCR cloning procedure was performed according to Frohman et al. (32) with slight modifications. Ten micrograms of stage 64 tadpole RNA were reverse-transcribed as described above with primer II (5Ј-AAAAGC-CATGAATATCCTGTA-3Ј, ϩ136 to ϩ116). The cDNAs were isolated by phenol extraction, phenol/chloroform extraction, and ethanol precipitation and were resuspended in 1ϫ Tris/EDTA. One-fifth of the cDNAs were precipitated again with ethanol in an ammonium acetate buffer and resuspended into 20.6 l of 1ϫ Tris/EDTA. For tailing, 2.4 l of 2.5 mM dATP, 6 l of 5ϫ tailing buffer (Life Technologies, Inc.), and 15 units of terminal deoxynucleotidyltransferase (Life Technologies, Inc.) were added, and the mixture was incubated for 12 min at 37°C and heated for 15 min at 65°C. The reaction mixture was diluted to 500 l in 1ϫ Tris/EDTA, and 5-l aliquots were used for amplification in 50 l of PCR mixture (5 l of 10ϫ Taq polymerase buffer with MgCl 2 (Promega), 0.8 l of 25 mM dNTP, 0.65 l of 0.1 g/l (dT) 17 adapter, 0.8 l of 0.1 g/l adapter, 0.8 l of 0.1 g/l primer I, and 2.5 units of Taq DNA polymerase (Promega)) (for adapter primer sequence, see Ref. 32). Using a DNA thermal cycler (Perkin-Elmer), the mixture was denatured at 94°C for 4 min, annealed at 42°C for 2 min, and extended at 72°C for 30 min before 40 cycles of amplification using a step program (94°C, 40 s; 55°C, 2 min; and 72°C, 3 min), followed by a 30-min final extension at 72°C. PCR products were cloned with the Original TA cloning kit (Invitrogen). Individual clones were isolated and sequenced with a T7 Sequenase Version 2.0 DNA sequencing kit (Amersham Pharmacia Biotech). Seven independent TR␤A cDNA clones were obtained that had only a few base changes (different among different clones) compared with the TR␤A genomic clone. These changes were likely derived from PCR errors due to the use of Taq polymerase and/or sequence polymorphisms. However, all had their 5Ј-end at position ϩ1 or ϩ3.

Promoter Analysis in Frog Oocytes Identifies a Minimal
Region Essential for TR␤A Transcription-The two X. laevis TR␤ genes (TR␤A and TR␤B) are regulated identically during development and by thyroid hormone (16,33). Genomic structure analyses have revealed that each TR␤ gene produces mRNAs with two alternative 5Ј-ends, i.e. having two different 5Ј-exons (exons a and b, respectively) (33). Although the relative locations of the two exons, i.e. whether exon b is located upstream of exon a or vice versa, have yet to be determined, both exons are independently transcribed (21,33). The expression from the promoter upstream of exon a is maintained at low but constitutive levels. In contrast, the promoter upstream of exon b is repressed in the absence of T 3 , but is activated to high levels when T 3 is present. We (22) and others (23) have previously analyzed this T 3 -inducible promoter (upstream of exon b) of the TR␤A gene and identified a strong TRE that mediates the strong autoinduction of the receptor gene (Fig. 1A). However, the start site of the TR␤A promoter had not been determined, possibly due to the low abundance of the mRNA in vivo or generated from transfection (22). Thus, it has not be possible to identify any elements other than the TRE that are necessary for reporter gene expression in transient transfection assays (22).
To determine the transcription start site, we microinjected a single-stranded reporter construct, which contained 46 bp of the previously reported exon b (16) and another 1.6 kilobases of upstream sequence, into Xenopus oocytes in the presence or absence of overexpressed TR/RXR heterodimers and/or T 3 . The single-stranded DNA is known to be replicated into the doublestranded form and assembled into chromatin through a replication-coupled assembly process, mimicking that in somatic cells (24,29). Primer extension analysis showed that there was a very low level of transcription from the promoter in the absence of T 3 and TR/RXR (Fig. 1B). Overexpression of TR/RXR in the absence of T 3 caused a further reduction of transcription.
In the presence of both T 3 and TR/RXR, the transcription was strongly activated, in agreement with the early transfection studies. Both the basal and T 3 -induced transcripts initiated from position ϩ1 (Fig. 1, A and B), 264 bp upstream of the 5Ј-end of the published exon b sequence (16). Thus, the previously published exon b sequence was derived from a partial mRNA clone truncated at the 5Ј-end, and the TRE is located ϳ262 bp downstream of the transcription start site (Fig. 1A).
To confirm that the start site identified in the oocyte is also used in tadpoles, the anchor PCR or rapid amplification of cDNA ends PCR method (32) was used to clone the 5Ј-end of TR␤A mRNA. For this purpose, stage 64 tadpole RNA was reverse-transcribed with a TR␤-specific primer and PCR-amplified with another specific primer and the adapter primer. The resulting cDNA was cloned, and individual clones were isolated and sequenced. Seven independent TR␤A cDNA clones were obtained. Three clones had their 5Ј-ends at position ϩ1, and four clones at position ϩ3 (Fig. 1A). These results indicate that position ϩ1 is also the start site in metamorphosing tadpoles when TR␤ genes are highly expressed (20,31).
As earlier transfection studies showed that TR␤A plasmid constructs with 5Ј-deletions passing site ϩ1 still gave strong reporter signal (Ref. 22, where position ϩ1 is equivalent to FIG. 2. Deletion analyses reveal that sequences upstream of ؊172 and downstream of ؉8 are not necessary for promoter function. Double-stranded promoter constructs containing the indicated regions of the promoter were microinjected into oocytes. After overnight incubation, the RNA was isolated and analyzed by primer extension with 32 P-labeled pCAT primer. Lanes 1-10 correspond to control without DNA and the constructs from top (wild-type (WT)) to bottom (pTRp11). The products derived from the TR␤ promoter are indicated by small arrowheads. The large arrowhead indicates the product generated from an endogenous oocyte RNA due to cross-hybridization by the pCAT primer, which served as a control for RNA isolation and primer extension (24). Promoter activities in oocytes were normally analyzed at least twice, and similar results were obtained (data not shown). An UPE is present at Ϫ151 to Ϫ125, and sequence of the Inr region is also shown. A TRE, consisting of two near-perfect repeats of AGGTCA separated by 4 bp, is present at 262 bp downstream of the start site. The sequences are from Ranjan et al. (22). The vertical arrows at ϩ1 and ϩ3 indicate the positions of the 5Ј-ends of the PCR clones of TR␤A cDNA from stage 64 tadpole RNA, with three and four independent clones, respectively. B, transcription start site mapping. A single-stranded wild-type TR␤ promoter construct was injected into oocytes with or without prior injection of the mRNAs for TR␤ and RXR␣. The promoter plasmid was replicated in vivo and assembled into chromatin. After overnight incubation in the presence or absence of 50 nM T 3 , the RNA transcribed from the chromatinized template was analyzed by primer extension using the pCAT primer. A sequencing ladder of the promoter plasmid with the same primer was used to determine the start site. position ϩ270 in this paper), we were interested to determine whether these constructs were able to direct accurate transcription in frog oocytes. For this purpose, we microinjected the same double-stranded plasmid constructs used in the earlier studies into frog oocytes. Since double-stranded plasmids do not undergo replication and consequently are assembled through a different pathway into a chromatin form that yields high levels of basal transcription (24, 29), it was not necessary to overexpress TR/RXR and/or to add T 3 to the oocyte culturing medium. Consistently, primer extension analysis of the RNA FIG. 3. Deletion analyses identify a minimal promoter region located at ؊154 to ؉7. 5Ј-Deletion (A) and 3Ј-deletion (B) constructs were microinjected into oocytes in double-stranded form, which allowed efficient transcription of the wild-type (WT) promoter in the absence of both T 3 and TR/RXR. After overnight incubation, the RNA was isolated and analyzed by primer extension with primer I for the TR␤ transcript and with the H4 primer for the endogenous histone H4 mRNA, which served as an internal control (indicated by arrowheads).
isolated from oocytes injected with the full-length TR␤A promoter (wild-type, Ϫ1336 to ϩ316) showed that the RNA with the expected size was efficiently transcribed in the oocytes (Fig.  2). However, many of the short constructs failed to produce specific transcripts (data not shown). To better define the promoter sequence, several additional deletion constructs were made around the start site and analyzed in oocytes. Constructs with 5Ј-deletions up to Ϫ276 or Ϫ172 (pTR6 and pTRp11, respectively) were still able to direct accurate transcription from position ϩ1, as were the constructs with 3Ј-deletions up to ϩ266, ϩ244, or ϩ8 (pTRp9, pTRp6, and pTRp10, respectively) (Fig. 2). On the other hand, those constructs with 5Ј-deletions passing Ϫ172 (pTRp4, pTRp5, pTRp7, and pTRp8) failed to produce any transcripts from the start site (ϩ1). Thus, these data suggest that the reporter activity detected by chloramphenicol acetyltransferase assay reported for the transient transfection experiments with some of the constructs (22) was most likely due to nonspecific transcription from the reporter plasmid, which was enhanced by the TR␤A DNA sequence inserted in front of the chloramphenicol acetyltransferase coding region through a yet unknown mechanism. Furthermore, these results suggest that the region from Ϫ172 to ϩ8 is sufficient for accurate basal transcription from the promoter.
Deletion and Mutational Analyses Reveal the Existence of a Novel DNA Element and an Initiator in the TR␤A Promoter-To further characterize the DNA sequences necessary for the basal promoter activity, additional 5Ј-and 3Ј-deletion constructs were made and injected into oocytes in the doublestranded form. Primer extension analysis showed that 5Ј-deletion up to Ϫ154 (pTRp5Ј-7) still yielded an active promoter, whereas a further deletion of 13 bp (pTRp5Ј-6) or more resulted in an inactive promoter (Fig. 3A). Thus, the region from Ϫ154 to Ϫ141 is an essential element for the activity.
Transcription studies of promoter constructs with additional 3Ј-deletions showed that the sequences downstream of ϩ7 (pTRp3Ј-1) were not necessary for promoter activity, whereas deletion to ϩ5 (pTRp3Ј-2) or Ϫ4 (pTR3Ј-3) abolished accurate transcription (Fig. 3B). These results suggest that the region around the start site is critical for the promoter, i.e. the presence of an initiator.
Site-directed mutagenesis was then carried out to further define the UPE around Ϫ140 and the Inr. Three site-directed mutational constructs were made, each changing 6 residues to the recognition site of NdeI. Two of them were in the upstream region around Ϫ140, i.e. the UPE, and one was near the start site, i.e. the Inr. Consistent with the deletion analyses above, mutating residues Ϫ148 to Ϫ143 abolished the promoter function (Fig. 4, construct UPE/m2). On the other hand, changing residues Ϫ130 to Ϫ125 had no effect (construct UPE/m1). Thus, the UPE lies within Ϫ151 to Ϫ130. Similarly, the mutation of residues ϩ6 to ϩ11 to CATATG (construct Inr/m1) inactivated the promoter, confirming that sequence up to ϩ7 is important for initiator function.
Inspection of the UPE and Inr sequences revealed that they are complementary to each other with the exception of a few residues. This prompted us to examine whether the sequences or the complementarity between them is critical for the promoter. We first mutated 5-CCCCC-3Ј in the UPE to 5Ј-GCGGG-3Ј (UPE/m3) or 5Ј-GGGGG-3Ј in the Inr to 5Ј-CCCGC-3Ј (Inr/m2) and analyzed the promoter activities of the resulting constructs. Both were found to be inactive, although slight activity was detected for UPE/m3 (Fig. 4). This is consistent with the deletion and mutational studies for UPE above and with the fact that initiators require the sequences both upstream and downstream of the start site (26, 27). We then introduced both the UPE and Inr mutations into a single promoter construct. The rationale was that these mutations would maintain the GC content in the mutant promoter compared with the wild-type promoter. In addition, we reestablished the complementarity between the UPE and Inr, and the potential duplex formed between the UPE and Inr would have similar stability. However, the resulting promoter still failed to direct FIG. 4. Mutational studies further define the UPE and the Inr. Site-directed mutagenesis was performed to generate the mutant constructs. The promoter constructs were injected into oocytes, and transcripts were analyzed as described in the legend Fig. 3. The arrowhead indicates the internal control. Note that the mutations at Ϫ148 to Ϫ143 (lane 3) and Ϫ144 to Ϫ140 (lane 4) in the UPE and at Ϫ3 to ϩ2 (lane 6) and ϩ6 to ϩ1 (lane 5) in the Inr abolished the promoter function, whereas the mutation at Ϫ130 to Ϫ125 (lane 2) in the UPE had no effect. The double mutation at Ϫ144 to Ϫ140 and at Ϫ3 to ϩ2 (lane 7), which restored the partial complementarity between the UPE and Inr, was also inactive, indicating that the sequences, not the complementarity, of the UPE and Inr are important for promoter function. WT, wild-type. accurate transcription (Fig. 4, construct UPE/Inr/m). Thus, the sequences of the UPE and Inr are important for promoter function, which is further supported by the detection of proteins that specifically recognize UPE and Inr sequences (see below).
The Novel DNA Element (UPE) Is Not an Enhancer, but Constitutes an Essential Part of the Basal Promoter-The inability of the UPE deletion and mutational constructs to support transcription suggests that the UPE may function as either a basal promoter element or an enhancer. In the latter case, the activity of the basal promoter may be too weak to be detected by the primer extension assay (once the UPE is mutated) and should be rescued by adding a different enhancer. To distinguish between these two possibilities, we reintroduced the UPE or a truncated UPE in both orientations into a deletion construct that had no UPE and tested the transcriptional activity of the resulting constructs (Fig. 5). The results showed that the truncated UPE failed to rescue the promoter function (Fig. 5, constructs 5 and 6), as expected. Interestingly, the UPE was able to rescue the promoter when placed in the same orientation as in the wild-type promoter (Fig. 5, construct 4), but it failed to do so when placed in the opposite orientation (construct 3). Thus, the UPE functions in an orientation-dependent manner, thus most likely as a promoter element, not as an enhancer. Consistently, when we placed the UPE in either orientation into a construct containing the enhancerless SV40 early promoter, we found that it failed to alter the pro-moter activity (data not shown).
To provide further evidence that the UPE is a basal promoter element, we replaced the UPE with other known enhancer elements and analyzed the activity of the resulting promoter constructs. For this purpose, we placed one or three copies of the binding site for the yeast transcription factor Gal4 (34) or one or three copies of the TRE (22) into pTRp5Ј-7, an active promoter (Fig. 3), or pTRp5Ј-1, an inactive promoter (Fig. 3). Both pTRp5Ј-1 and pTRp5Ј-7 also contain the TRE at ϩ262 (Fig. 1A). The transcriptional activity of the constructs was analyzed in the presence of Gal4-VP16 or TR/RXR in the frog oocytes through mRNA (Gal4-VP16 or TR/RXR) injection before the injection of the promoter constructs. The results showed that, as expected, TR/RXR heterodimers were able to enhance the transcription of the active promoter pTRp5Ј-7 in the presence of thyroid hormone (Fig. 6, lanes 1 and 2; data not shown). In contrast, they had no effect on the inactive promoter pTRp5Ј-1 (lanes 3 and 4; data not shown). Similarly, Gal4-VP16 could enhance the transcription of pTRp5Ј-7 with either one or three copies of the Gal4-binding sites (lanes 9 -12) while failing to affect pTRp5Ј-1 bearing the same sites (lanes 5-8). Thus, constructs lacking the UPE are inactive promoters, even in the presence of multiple recognition elements of the strong activator Gal4-VP16. The results further support that the UPE is an essential core promoter element of the TR␤A gene.
Specific Recognition of the UPE by Xenopus Proteins-We FIG. 5. Promoter reconstitution suggests that the UPE functions as a basal promoter element. The truncated promoter (construct 1, Ϫ154 to ϩ316) containing the minimal promoter sequence was transcriptionally active (lane 1), whereas a short construct (construct 2, Ϫ63 to ϩ316) was inactive (lane 2). Placing the UPE back into construct 2 in the wild-type orientation (construct 4), but not in the opposite orientation (construct 3), restored the promoter activity (lanes 4 and 3, respectively). In contrast, placing a truncated UPE back into construct 2 in either orientation (constructs 5 and 6) did not rescue the promoter (lanes 5 and 6, respectively). Oocyte injection and transcript analyses were done as described in the legend to Fig. 3. The arrowhead indicates the internal control.
FIG. 6. Promoter reconstitution experiments confirm that the UPE is essential for basal promoter function. One or three copies of the Gal4 DNA-binding sites were inserted upstream of an active promoter (pTRp5Ј-7, Ϫ154 to ϩ316) or inactive promoter (pTRp5Ј-1, Ϫ63 to ϩ316), both containing the TRE at ϩ262. The original and new constructs were microinjected in the double-stranded form into oocytes with or without prior injection of TR/RXR in the presence of T 3 or Gal4-VP16 mRNA. After overnight incubation, the RNA was isolated and analyzed as in the legend to Fig. 3. The arrowhead and arrow indicate the internal control and TR␤ products, respectively. The TR␤ product migrated to a similar location as a product derived from nonspecific priming of endogenous RNA. Due to the use of 32 P-labeled primer I with a lower specific activity and lower levels of TR␤ promoter DNA used in this experiment, the signal for the TR␤ decreased relative to the nonspecific product. Thus, a signal derived from nonspecific endogenous RNA was observed for the inactive promoter (pTRp5Ј-1, lanes 3-8; compared with Fig. 3A, lane 8). The inactive promoter and its derivatives were not affected by either TR/RXR or Gal4-VP16 despite the presence of their binding sites. On the other hand, TR/RXR (lanes 1 and 2) and Gal4-VP16 (lanes 9 and 10) were able to activate the functional promoter containing their binding sites. used gel mobility shift assay to determine whether proteins exist to recognize the UPE. We isolated protein extracts from Xenopus oocytes and from a Xenopus tissue culture cell line that is known to be able to regulate endogenous TR␤ gene expression in a thyroid hormone-dependent manner, just like in tadpoles. 2 When a 32 P-labeled UPE oligonucleotide was mixed with the extract, several complexes were formed (Fig. 7). The same results were obtained when oocyte extract was used (data not shown). All of them could be competed away by the unlabeled UPE itself, but the major one could not be competed by the truncated UPE (mUPE) (Fig. 7) or by a mutant UPE bearing the mutations that inactivate its promoter activity (data not shown).
Consistent with the sequence similarity between the UPE and Inr, the Inr oligonucleotide could compete away the major complex. On the other hand, a mutated Inr failed to compete away this complex (Fig. 7). Thus, this complex represents the binding of the UPE by a sequence-specific DNA-binding protein, which also recognizes the Inr, and an intimate correlation exists between the binding of the sequence-specific DNA-binding protein to the UPE and Inr and the TR␤A promoter activity. DISCUSSION Thyroid hormone receptors mediate the biological effects of thyroid hormone. Perhaps the most dramatic T 3 -dependent process is amphibian metamorphosis, during which an aquatic tadpole is transformed into a terrestrial frog. As expected, the TR genes are highly expressed during amphibian metamorphosis (17-20, 31, 35-38). Interestingly, the expression of TR␤ genes has been shown to be regulated by T 3 temporally in a tissue-dependent manner that correlates with tissue-specific changes during development, even though they are regulated by T 3 ubiquitously in all organs (31, 37, 39 -42), and this regulation is directly mediated by TRs themselves (22,23). We have shown here that this spatial and temporal autoregulation appears to involve, in addition to a TRE, an initiator and a novel promoter element located ϳ140 bp upstream of the start site.
Each of the TR␤ genes (TR␤A and TR␤B) in X. laevis is transcribed from two promoters based on transcript analysis (21,33). One of the promoters is constitutively expressed at low levels, and the other is thyroid hormone-inducible. We (22,33) and others (23) have shown previously by transient transfection that this T 3 -inducible promoter of the TR␤A gene contains a strong TRE (Fig. 1A), which mediates transcriptional repression by unliganded TR/RXR and activation by T 3 -bound TR/ RXR. However, these earlier studies failed to reveal any other DNA elements critical for promoter function due to the use of an indirect reporter assay and a lack of information on the transcription start site. By directly analyzing the transcripts derived from the promoter both in the absence and presence of TR/RXR and/or T 3 , we have shown here that the basal and T 3 -induced transcription starts at 262 bp upstream of the TRE. We have further defined a minimal promoter for accurate transcription that includes the sequence from Ϫ154 to ϩ7 relative to the start site. Within this promoter region, no TATA box is present, suggesting that the promoter belongs to the class of TATA-less promoters. Consistently, we have found that the promoter contains an essential initiator, a key feature of TATA-less promoters (26,27).
Sequence comparison reveals no obvious resemblance of the initiator to any groups of initiators identified so far (25,43,44). While transcription from TATA-less promoters usually starts from an A residue, primer extension analysis clearly indicate that a G residue is used. Thus, the TR␤ initiator is a novel element.
In addition to the initiator, our deletion and mutational analysis also revealed the existence of an important, novel UPE. The activity of the promoter depends on the presence of both the initiator and the UPE since it is abolished when either one is mutated or deleted. Several lines of evidence argue that the UPE functions as a novel core promoter element, but not as an enhancer. First, the UPE cannot be substituted by single or multiple TRE-or Gal4-binding sites, which, however, can mediate the strong transcriptional activation by liganded TR/RXR or Gal4-VP16, respectively, when the UPE is also present. Second, the function of UPE is orientation-dependent. In general, transcription enhancers function in an orientation-independent manner. The absolute requirement for both the initiator and UPE is exceptional since, in other TATA-less promoters, the initiator is sufficient to direct low levels of basal transcription, although the presence of binding sites for transcription factors like SP-1 can augment promoter activity.
The Xenopus TR␤A promoter bears some similarities to the human TR␤ promoter (45). Both promoters are autoregulated by TRs themselves, and TREs have been identified in the promoters. In addition, binding sites for transcription factor SP-1 or related factors are present in both genes and are important for human promoter function (46,47). Although the SP-1 sites are dispensable for Xenopus promoter function in tissue culture cells or oocytes (Fig. 2) (22), it cannot be ruled out that they may play a role in development.
Distinct differences, however, exist between the human and Xenopus TR␤ promoters. The human promoter has a TATA-like motif and Oct-1 elements (45)(46)(47). The Xenopus promoter lacks such elements. Instead, it contains an initiator element and an upstream promoter element.
A surprising feature of the Xenopus TR␤ promoter comes from the sequence similarity between the UPE and initiator (Fig. 1) (22). The UPE and initiator are orientated in opposite directions in the TR␤ promoters and thus could potentially form a heteroduplex. Interestingly, no transcription initiated FIG. 7. The UPE forms specific complexes with Xenopus proteins. Two nanograms of 32 P-labeled double-stranded oligonucleotide containing the UPE were mixed with protein extract from Xenopus tissue culture cells in the presence of 5, 10, 25, and 100 ng of the unlabeled UPE itself, a truncated UPE (mUPE), the Inr, or a mutated Inr (mInr). The resulting complexes were analyzed by gel mobility shift assay. The arrowhead points to the specific complex that was competed by UPE, but not by mUPE or mInr. The Inr was able to compete due its sequence similarity to UPE. from the UPE region, but in an opposite direction could be detected using a primer located upstream (5Ј to 3Ј direction is toward the UPE) of the UPE (data not shown), indicating that the sequence differences between the UPE and initiator and/or other sequences in the minimal promoter are important for the directionality of the promoter.
We have also tested whether the potential secondary structure formed due to the complementarity of the initiator and UPE is involved in promoter function. Reciprocal mutations that maintain the potential secondary structure and its GC content or stability failed to produce an active promoter, suggesting that the sequences, but not the potential secondary structures, are important for promoter function. Consistent with this, gel mobility shift experiments revealed the existence of a protein(s) that binds specifically to the UPE and initiator. The mutations/deletions that inactivate the UPE or initiator also abolish the ability to compete for binding. Thus, the binding by this protein(s) correlates with the transcription function of the UPE and initiator. The identity and nature of this protein(s) will be of particular interest for future studies on TR␤A promoter regulation.