Thyroid Hormone Receptor (cid:98) 2 Promoter Activity in Pituitary Cells Is Regulated by Pit-1*

There are three known thyroid hormone receptor (TR) isoforms that arise from two distinct (cid:97) and (cid:98) gene loci. TR (cid:97) 1 and TR (cid:98) 1 mRNAs are found in many tissues, whereas mRNA for the N-terminal TR (cid:98) 2 variant derived from the (cid:98) locus is readily detectable only in the pituitary gland and derived cell sources such as GH3 soma-totropes and TtT-97 thyrotropes. We previously isolated the genomic region governing expression of the TR (cid:98) 2 isoform in thyrotropes and showed that transcription arose from multiple origins within a 400-base pair (bp) region. We now report that the region extending 500 bp upstream of the putative AUG codon (A is (cid:49) 1) contains six areas of interaction with the pituitary-specific transcription factor Pit-1. In addition there are separate areas that bind other factors present in thyrotrope cells. Promoter deletions revealed that removal of regions containing the Pit-1 sites at (cid:50) 456 to (cid:50) 432, (cid:50) 149 to (cid:50) 127, and (cid:50) 124 to (cid:50) 102 progressively decreased TR (cid:98) 2 promoter activity in thyrotropes. A more proximal footprinted area from (cid:50) 65 to (cid:50) 19, which accounted for the remaining promoter activity, contained sites that inter-acted with recombinant Pit-1; however, extracts of TtT-97 thyrotropes, which express Pit-1, footprinted potential binding sites for Pit-1. This report documents that the region important for TR (cid:98) 2 promoter activity in thyrotropes does interact with Pit-1 as as with other proteins in thyrotropes. We also demonstrate that this genomic region also supports promoter activity in GH3 somatotrope cells and that the activity in both cell types is dependent on the presence of areas that interact with Pit-1. Finally cotransfection experiments with (cid:97) -TSH cells

tropes as constructs lacking them by deletion or mutation were not stimulated by Pit-1.
The effects of thyroid hormone (T3) are dependent on its interaction with high affinity nuclear receptor molecules that are related to those that mediate the effects of the steroid hormones, retinoids, and vitamin D (1). Thyroid hormone receptors (TRs) 1 arise from two separate genomic loci (␣ and ␤) (2). Translation of alternately spliced transcripts from the ␣ locus gives rise to TR␣1, a hormone binding isoform that regulates T3-responsive genes and ␣2, a C-terminal variant, that does not bind T3 (3,4) and may act as an antagonist of T3 response (5). In contrast, the ␤ locus gives rise to two receptor isoforms (TR␤1 and TR␤2) as a result of transcription directed by two separate promoter regions and subsequent splicing of two different N termini onto the same DNA and hormone binding regions (2,6). TR␤1 expression is widespread (7), and although TR␤2 immunoreactivity has been reported in a variety of tissues (8), its mRNA is detectable by Northern blot analysis only in the pituitary gland and cell sources derived from it (6,9). Following its original description in rat GH3 somatotrope tumor cells and demonstration of its restricted expression to the pituitary gland (6), our laboratory cloned TR␤2 cDNA from mouse thyrotropic TtT-97 tumor tissue (9). Childs et al. (10) subsequently demonstrated by in situ hybridization that transcripts encoding TR␤2 colocalized almost exclusively with cells in the pituitary gland that stained for thyroid-stimulating hormone (TSH) and growth hormone (GH). Differentiation of these two pituitary cell types has been shown to be dependent on the transcription factor Pit-1 (11,12). Because of its restricted expression to the pituitary, we reasoned that the promoter region that regulates transcription of TR␤2 may be under the control of pituitary-specific factors such as Pit-1. We recently cloned a mouse genomic fragment containing the region immediately upstream of the TR␤2 coding region and showed that it exhibited the properties of a promoter by preferentially directing expression of luciferase fusion constructs in TR␤2-expressing TtT-97 cells when compared with non-expressing ␣-TSH cells (13). The TR␤2 promoter region contained several motifs that could be potential binding sites for Pit-1. This report documents that the region important for TR␤2 promoter activity in thyrotropes does interact with Pit-1 as well as with other proteins present in thyrotropes. We also demonstrate that this genomic region also supports promoter activity in GH3 somatotrope cells and that the activity in both cell types is dependent on the presence of areas that interact with Pit-1. Finally cotransfection experiments with ␣-TSH cells that express neither TR␤2 nor Pit-1 establish that Pit-1 is capable of reconstituting TR␤2 promoter activity to a level approaching that observed with expressing TtT-97 thyrotropes and that the activation is dependent on the interaction of Pit-1 at specific sites.

MATERIALS AND METHODS
Experimental Animals-TtT-97 thyrotropic tumor propagation and maintenance in hypothyroid male LAF 1 mice have been previously described (14). All tumor bearing mice used in these studies were treated in accordance with the National Institutes of Health guidelines for animal use and care. All protocols were reviewed and approved by the University of Colorado Health Sciences Center Committee on Use and Care of Animals.
Cell Culture-Monolayer cultures of GH3 cells (ATCC CCL 82.1) or ␣-TSH cells (15) in suspension were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. To maximize expression of the TR␤2 isoform, which is down-regulated by thyroid hormone (9), cells were incubated in the same medium containing charcoal-stripped serum that lacked detectable levels of T4 and T3 for 48 h before preparation of nuclear extracts or transfection experiments.
Construction of 5Ј Deleted and Mutated TR␤2 Luciferase Fusion Plasmids and Transient Transfection Assays-The TR␤2 promoter luciferase fusion plasmids extending 3Ј to ϩ40 and with 5Ј deletion points at Ϫ2064, Ϫ572, Ϫ465, Ϫ204, and Ϫ77 were constructed using convenient restriction sites as described previously (13). Additional deletions with 5Ј extents at Ϫ152, Ϫ121, and Ϫ25 were prepared using a polymerase chain reaction-based strategy to generate amplified fragments with these end points. For this purpose the following TR␤2 5Ј sense oligonucleotides, with a SmaI site incorporated (underlined) to facilitate subsequent subcloning, were synthesized: 5Ј-CAGCCCGGGCTG-GTGGTGTTTATTCAT-3Ј; 5Ј-CAGCCCGGGTTTCATGTGTATGTATG-3Ј; and 5Ј CAGCCCGGGTAGAACCTGAACCTGGAT-3Ј. Each sense strand primer was used together with a 3Ј antisense oligonucleotide (5Ј-GCCTTTCTTTATGTTTTTGGCG-3Ј) complementary to a sequence just within the luciferase coding sequence to generate the desired fragments by amplification from the luciferase vector containing the TR␤2 sequence from Ϫ572 to ϩ40. Polymerase chain reaction for 30 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min was performed. Following gel purification and digestion with SmaI and HindIII, the new 5Ј deleted promoter fragments were inserted between the same sites of the promoterless plasmid pA 3 LUC (20), and end points were verified by sequencing.
Site-directed mutagenesis of Pit-1 binding consensus sequences was carried out in the context of the Ϫ204 to ϩ40 TR␤2 promoter fragment excised from pA 3 LUC and inserted into pSELECT (Promega, Madison, WI) as described previously (25). Specifically, 32-bp oligonucleotides were synthesized with 12 bp on either side of the AT-rich Pit-1 e or f site, which was altered to a GC-rich NotI recognition site. Mutagenesis was performed according to the supplier's instructions described in the Altered Sites System (Promega), and subsequent digestion of miniprep DNA with NotI aided in screening for the presence of the mutations, which were verified by sequencing. The double mutant (Pit-1 sites e and f) was generated using the oligonucleotide used to mutate the Pit-1 f site, except that pSELECT containing the Pit-1 e mutation was used as the starting plasmid. Mutated TR␤2 promoter fragments were re-excised with KpnI and HindIII and reinserted into pA 3 LUC between the same sites.
Transient transfection by electroporation was carried out essentially as described previously (14). Specifically, 20 g of TR␤2 promoterluciferase plasmid DNA, together with 1-3 g of pCMV␤-gal (Clontech, Palo Alto, CA) as an internal transfection efficiency control, were transfected into 5-10 million freshly dispersed TtT-97 tumor cells or 5 million GH3 cells. An RSV-directed luciferase vector (pA 3 RSV400LUC) was transfected in parallel allowing comparison of TR␤2 promoter activity between different cell types. Following incubation at 37°C for 18 -24 h, cells were harvested and freeze-thaw extracted, and supernatants were assayed for both luciferase and ␤-galactosidase as previously reported (19). For experiments where the effect of exogenously supplied Pit-1 was assessed, 3 million ␣-TSH cells were transfected with the TR␤2LUC and ␤-galactosidase plasmid DNAs as above and either 10 g of pCMVPit-1 or the same vector lacking the Pit-1 cDNA sequence (pCMV), both of which were constructed as outlined in a previous report (21).
Preparation of Nuclear and Bacterial Extracts, Probe Production, and DNase I Protection Analysis-Nuclear extracts were prepared from enzymatically dispersed TtT-97 thyrotropic tumors, ␣-TSH and GH3 cells as described previously (16,17). Bacterial extracts expressing recombinant rat Pit-1 protein as a fusion with glutathione S-transferase were prepared according to a previously published procedure (18). Probes corresponding to the TR␤2 promoter regions from Ϫ572 to Ϫ77, from Ϫ210 to ϩ102 and from Ϫ77 to ϩ176 were prepared by cloning end-filled SpeI to NcoI, TaqI to HpaII, and end-filled NcoI to HinfI fragments into the SmaI, BstBI, and SmaI sites of pGEM7zf(ϩ), respectively. For analysis of Pit-1 mutations, the Ϫ204 to ϩ40 wild type and mutated regions were excised from pSELECT with SmaI and HindIII and inserted between the same sites of pGEM7zfϩ. Subsequent excision of the TR␤2 fragments with EcoRI (AATT-5Ј overhang) and MluI (CGCG-5Ј overhang) allowed the resulting fragments to be selectively end-filled using avian myeloblastosis virus reverse transcriptase and either 32 P-labeled dATP and dTTP or dCTP and dGTP. Radiolabeled TR␤2 probes were allowed to interact with 20 g of bovine serum albumin (no extract), 120 g of bacterial Pit-1 extract protein, or 100 -300 g of pituitary cell nuclear extract protein under defined conditions, and DNase I digestion and analysis on 5% polyacrylamide-8 M urea gels was carried out as previously reported (19).

RESULTS
Pit-1 binds to Multiple Sites within the TR␤2 Promoter Region Important for Expression in Thyrotropes-We previously reported that the TR␤2 genomic region upstream of the putative AUG codon when fused to a luciferase reporter directed luciferase expression in TtT-97 thyrotropic tumor cells that express endogenous TR␤2 mRNA (13). In contrast, when similar constructs were transfected into ␣-TSH cells, a thyrotropederived cell line devoid of detectable TR␤2 mRNA, 4 -5-fold less luciferase activity resulted (13). This functional data together with the observation that this 5Ј region also contained sites of transcriptional initiation in thyrotrope cells (13) strongly suggested that the 5Ј TR␤2 region contained the promoter elements important for expression in thyrotrope cells. Deletional analysis demonstrated that the proximal 5Ј area comprising 465 bp upstream of the AUG codon (designated ϩ1) to position ϩ40 was sufficient to support promoter activity in thyrotropes and that inclusion of larger fragments extending as far as 2.9 kilobases upstream did not further enhance promoter activity (13). An examination of the sequence of this region revealed the presence of multiple AT-rich sequence motifs with homology to binding sites for the pituitary-specific transcription factor Pit-1 as well as GC-rich regions resembling sites of interaction for the more widely expressed transcription factor Sp1. Because ␣-TSH cells do not contain detectable Pit-1 protein (18) and also poorly support transfected TR␤2 promoter activity, we investigated the possibility that Pit-1 was important for the expression of TR␤2 in TtT-97 thyrotrope cells. First we wished to determine if the AT-rich motifs did indeed bind Pit-1. DNase I protection analyses using labeled fragments encompassing the entire 500-bp promoter region incubated with either bacterially expressed Pit-1 or nuclear extracts from Pit-1 expressing TtT-97 tumors demonstrated that the functionally important region contained six areas of interaction with the recombinant Pit-1 preparation (Fig. 1, A-D, open boxes). All of these sites, with the exception of one from Ϫ412 to Ϫ387, were also protected by Pit-1 containing TtT-97 nuclear extracts (filled boxes). In addition to the Pit-1 sites, at least four other regions (also designated by filled boxes) were protected by thyrotrope nuclear extracts but not by bacterial Pit-1 protein.
These included an extension of a Pit-1 footprinted area from Ϫ522 to Ϫ545 distally to position Ϫ561, from Ϫ201 to Ϫ183, from Ϫ65 to Ϫ56, and from ϩ9 to ϩ25. boxes designated T-1 through T-4). Closer examination of the Pit-1 protected area from Ϫ49 to Ϫ19 (Fig. 1D) reveals that the thyrotrope extract footprint does not exactly coincide with that of bacterially expressed Pit-1, being slightly displaced distally by 3-4 bps (Ϫ53 to Ϫ23). This difference in protection may suggest the presence within thyrotrope cells of a factor with similar sequence recognition as Pit-1 but that has a greater affinity than Pit-1 for the Pit-1 g, h, or i site and exhibits different binding characteristics. We therefore designated the thyrotrope footprint in this area as T-3A to distinguish it from the area of recombinant Pit-1 interaction. The other non-Pit-1 thyrotrope footprints do not bear any similarity to known transcription factor recognition sites with the exception of T-3, which contains a variant Sp1 motif (GGGCTGG) that has been shown in the promoter of the CD14 gene to bind Sp1 and mediate lymphocyte-specific expression (22). Interestingly, the thyrotrope extract protected T-4 region from ϩ9 to ϩ25 encompasses a cluster of transcription start sites that are downregulated by T3 in thyrotrope cells (13). However, a consensus T3-response element, indicative of a TR binding site, is not present within the sequence of the footprinted T-4 area.
Deletion Analysis of the TR␤2 Promoter in Thyrotropes-Because Pit-1 has been shown to be important for the expression of other pituitary genes particularly those in thyrotropes (TSH␤) and somatomammotropes (GH and PRL) (23-25), we evaluated the role of the Pit-1 binding sites in TR␤2 promoter activity in thyrotrope cells. Previous studies from our laboratory (13) have shown that deletion of the region from Ϫ572 to Ϫ465, which contains the Pit-1 a site, had no effect on TR␤2 promoter activity in transfected thyrotrope cells (also shown in Fig. 2A). However, when the region from Ϫ465 to Ϫ204 (containing Pit-1 sites b/c and d) was removed, a 50% decrease in promoter activity was observed (shown in Fig. 2A with additional determinations included). Further deletion to position Ϫ77, which removes the Pit-1 e and f sites, accounted for the majority of the remaining promoter activity in thyrotropes. To more specifically define the relative contributions of the proximal Pit-1 sites and the areas protected by thyrotrope extracts, we created three additional deletion mutants using a polymerase chain reaction strategy. These new 5Ј constructs, which terminate at Ϫ152, Ϫ121, and Ϫ25 together with the previously described deletions (see Fig. 1E) remove, in a progressive fashion, T-2, the Pit-1 e site, the Pit-1 f site, and finally the T-3/T-3A region, which encompasses the Sp1 motif and the Pit-1 g, h, and i sites. Fig. 2A shows the results of such a systematic deletional strategy on TR␤2 promoter activity in transfected TtT-97 thyrotropes cells. Removal of the T-2 area had no effect, whereas loss of the Pit-1 e site resulted in a 60% reduction in promoter activity from the already decreased level observed with the promoter fragment deleted to Ϫ204. Subsequent deletion to Ϫ77, which removes the Pit-1 f site, further decreases activity to approximately 20% of the activity of the Ϫ204 construct. Finally, deletion to Ϫ25, which removes the remaining Pit-1 sites, results in a promoter construct with no measurable luciferase expression above the promoterless pA 3 LUC control. Thus, removal of regions of Pit-1 interaction except Pit-1 a has a significant effect on TR␤2 promoter activity in thyrotrope cells.
The TR␤2 5Ј Region Also Functions as a Promoter in Somatotrope Cells-To determine if the same regions that govern TR␤2 expression in thyrotropes were also functional in somatotrope cells, we carried out similar transfection experiments with GH3 cells. Initial experiments showed that when normalized to a luciferase plasmid directed by an RSV promoter transfected in parallel, the TR␤2 promoter construct from Ϫ572 to ϩ40 expressed at an equivalent or higher level in GH3 cells when compared with TtT-97 thyrotropes (data not shown). Fig.  2B shows the results of a 5Ј deletion analysis in GH3 cells performed with the same constructs described earlier for TtT-97 thyrotropes. In agreement with the results of our previous transfections into thyrotropes (13), sequences upstream of Ϫ465 were dispensible for TR␤2 promoter activity in somatotropes. However, in contrast to the situation in thyrotropes, deletion of the region from Ϫ465 to Ϫ204 containing the Pit-1 b, c, and d sites, which resulted in a 50% decrease in thyrotropes ( Fig. 2A), had no effect on promoter activity in GH3 cells implicating a role for this region specific to thyrotrope cells. When the additional deletions, which systematically remove the more proximal Pit-1 sites, were tested in GH3 cells, the activity pattern produced was qualitatively similar to that seen with thyrotrope cells, except that a greater decrease in promoter activity (75% as opposed to 60%) was observed as a result of removing the Pit-1 e site.
Somatotrope Cell Extracts Interact with the Regions Containing Pit-1 Sites in the TR␤2 Promoter Region-Because the regions of the promoter governing TR␤2 expression in somatotropes differed both qualitatively and quantitatively from those being utilized in thyrotropes, we wished to determine if somatotrope cell nuclear extracts displayed a different protection pattern of the TR␤2 5Ј region than that observed with thyrotrope extracts. Fig. 3A shows that GH3 extracts were equally capable of protecting the Pit-1 b/c site despite no decrease in promoter activity in GH3 cells as a result of its removal by deletion to position Ϫ204 (Fig. 2B). As expected from the trans- fection deletion data presented in Fig. 2B, GH3 extracts were able to clearly protect Pit-1 sites e through i in a manner indistinguishable from recombinant Pit-1 (Fig. 3B), indicating that it is probably Pit-1 protein present in the nuclear extracts that is generating the footprints. These data suggest that binding of Pit-1 to these sites accounts for their prevalent contribution to TR␤2 promoter activity in GH3 cells as well as TtT-97 cells. This is supported by the observation that similar extracts from ␣-TSH cells, which lack Pit-1 protein, do not protect the Pit-1 e and f sites (Fig. 3B). Expansion of the more proximal region containing the Pit-1 g/h and i sites by footprinting a different fragment (Fig. 3C) demonstrated that the pattern of protection of the proximal area of interaction with GH3 extracts more closely resembles that generated by recombinant Pit-1 and does not manifest the distally displaced footprint seen with TtT-97 extracts. This suggests that somatotrope cells lack the factor(s) present in the thyrotrope cells that interacts at this proximal promoter area.

Binding of Pit-1 at the e and f Sites Is Required for TR␤2 Promoter Activity in GH3 Cells-The 5Ј deletion studies in both
TtT-97 and GH3 cells suggested that the region between Ϫ152 and Ϫ77, which contains the Pit-1 e and f sites, is important for TR␤2 promoter activity in both cell types. To further investigate the role of Pit-1 binding to these sites, we mutated the AT-rich consensus binding motifs within the footprinted areas and examined the consequences on both Pit-1 interaction and TR␤2 promoter activity. Fig. 4 shows that altering either or both of the Pit-1 motifs resulted in loss of binding at those sites of both recombinant Pit-1 as well as pituitary cell nuclear extracts. However, disruption of Pit-1 binding at one of the sites did not appear to affect its ability to interact at the other, ruling out possible cooperativity between the Pit-1 e and f sites. The effects of these mutations on TR␤2 promoter activity are shown in Fig. 5. When Pit-1 was no longer able to bind at the Pit-1 e site, promoter activity was decreased to 50% of the unmutated Ϫ204 construct. However, when binding to the more proximal Pit-1 f site was disrupted, either alone or in conjunction with the Pit-1 e site mutation, promoter activity was more dramatically lowered to only 20 -25% of the intact promoter construct to a level exhibited by the Ϫ77 construct, which has both Pit-1 e and f deleted. These results emphasize the importance of both the Pit-1 e and f sites for the expression of TR␤2 in pituitary cells but demonstrate the more dominant role of the more proximal f site. Similar decreases in TR␤2 promoter activity were also seen in TtT-97 thyrotropes as a result of the Pit-1 e and f site mutations (data not shown). We have also mutated the AT-rich Pit-1 consensus sequences within the b/c and i sites with no decrease in TR␤2 promoter activity in either pituitary cell type.
Pit-1 Is Sufficient to Reconstitute TR␤2 Promoter Activity in ␣-TSH Cells-To determine whether Pit-1 was capable of activating the TR␤2 promoter, we carried out experiments where Pit-1 was co-expressed with TR␤2 promoter constructs in ␣-TSH cells that express neither endogenous TR␤2 mRNA nor Pit-1 protein (13,18). Specifically we wanted to see if Pit-1 could reconstitute TR␤2 promoter activity to the level exhibited by TtT-97 cells that contain both TR␤2 message and Pit-1 protein detectable by Northern and Western blots, respectively (9,18). Fig. 6 shows that cotransfection of Pit-1 driven by the potent CMV promoter was able to stimulate a TR␤2 luciferase construct containing all of the Pit-1 sites (Ϫ572 to ϩ40) 4 -5fold, which was equivalent to the level previously seen in endogenously expressing TtT-97 thyrotrope cells (13). To see if the stimulation by Pit-1 was dependent on the presence of Pit-1

Pit-1 Regulates TR␤2 Promoter Activity
binding sites in the TR␤2 promoter fragment, experiments were carried out cotransfecting Pit-1 with the constructs described earlier, which have individual Pit-1 sites progressively deleted. The results of such an analyses is also presented in Fig. 6. Although deletion to Ϫ465, which removes Pit-1 a, results in a slight reduction in Pit-1 stimulation (5-3.5-fold), the Ϫ204 construct regains the 5-fold effect seen with the Ϫ572 deletion. However, when the Pit-1 e site is removed, a significant decrease in stimulation by Pit-1 to 2.5-fold is observed. Further deletion to Ϫ77, which eliminates Pit-1 f but still retains the g, h, and i, results in only minimal Pit-1 stimulation, which in some experiments was not significant. Further evidence that it is interaction of Pit-1 at the e and f sites that is primarily responsible for the stimulation by Pit-1 is presented in Fig. 7 where the TR␤2 constructs bearing mutations at either or both of these sites are impaired in their ability to be stimulated by Pit-1. Disruption of binding at the Pit-1 e site results in decreased Pit-1 stimulation from 4.4-to 2.9-fold, whereas when the f site is mutated, either alone or in conjunction with a mutated e site, Pit-1 has little or no effect on the constructs as is also seen with the Ϫ77 construct, which lacks both the e and f sites. A surprising finding was that the smallest construct (containing only 25 bp 5Ј upstream of the AUG codon), which is devoid of all Pit-1 sites, was actually inhibited by 50% in the presence of Pit-1 (Fig. 7). A similar inhibition was also seen for the RSV luciferase positive control and the promoterless pA 3 LUC plasmid, neither of which contain Pit-1 sites. No inhibition of a similar RSV promoter construct was reported by Mangalam et al. (26), whereas Steinfelder et al. (27) reported substantial inhibition of RSV luciferase activity by Pit-1 in 235-1 cells. A general inhibition of promoter activity independent of DNA binding by another homeodomain protein Msx-1 has recently been described (28). These transfection data in thyrotrope-derived ␣-TSH cells suggest that the Pit-1 e and f sites are the primary sites responsible for the stimulation of TR␤2 promoter activity in the presence of exogenously supplied Pit-1.

DISCUSSION
Pit-1 is a pituitary-specific transcription factor that plays an important role in pituitary development (29) and in the expression of several pituitary genes including GH and PRL (30,31), TSH␤ subunit (11,21,25,26), growth hormone releasing factor receptor (32), renin (33), and the Pit-1 gene itself (34). The results presented in this report demonstrate that expression of the pituitary-specific ␤2 TR isoform in thyrotrope and somatotrope cells is also dependent on Pit-1. Although T3-binding activity can be immunoprecipitated by TR␤2 antibodies from cell extracts from several extra-pituitary sources (8), readily detectable levels of TR␤2 mRNA expression are restricted to thyrotrope and somatotrope cells of pituitary origin (6,9,10). By gene transfer we have identified the regions of the murine TR␤2 promoter that are active in pituitary-derived GH3 somatotropes and also in cells derived from murine thyrotropic tumors. DNA-protein interaction studies show that this activity is correlated with the binding of Pit-1 and/or a factor with a similar sequence specificity present in extracts of these cells. These data in conjunction with the observation that cotransfection of a Pit-1 expression vector into Pit-1-deficient ␣-TSH cells results in a binding site-dependent activation of the TR␤2 promoter suggest that Pit-1 activates the murine TR␤2 promoter in transfected GH3 and TtT-97 cells and is most likely involved in regulating transcription from the endogenous TR␤2 genomic locus in a pituitary-specific fashion.
Six areas of interaction with bacterially expressed Pit-1 were detected within the proximal TR␤2 promoter region. Within these areas nine sequence motifs (Pit-1 a-i) were identified that closely resemble the Pit-1 consensus binding site sequence (A/T)(A/T)TATNCAT derived by Ingraham et al. (35). While Pit-1 d and e conform exactly to the consensus, the others vary by no more than two nucleotides from it. In the Pit-1 footprinted area from Ϫ456 to Ϫ432 b and c are two possible alignments of the Pit-1 binding site consensus sequence. Similarly, because of the inaccuracy in exactly assigning footprint boundaries, g or h cannot be excluded as contributing to the most proximal area of Pit-1 interaction. A similar ambiguous arrangement of three adjacent and overlapping Pit-1 consensus motifs are present in the proximal promoter of the rat Pit-1 gene from Ϫ63 to Ϫ41 (36). Deletional analyses demonstrated that the Pit-1 a site, which is protected by both recombinant Pit-1 and pituitary cell extracts, is not required for TR␤2 promoter activity in either thyrotrope or somatotrope cells. Furthermore, although its functional role is not known, the recombinant Pit-1 footprinted area, which contains the perfect consensus Pit-1 d site, is not protected by extracts from thyrotrope cells that contain Pit-1. However, the Pit-1 b/c, which is protected by both recombinant Pit-1 and pituitary cell extracts, can be mutated with no effect on TR␤2 promoter activity in thyrotropes. Examples of sites with high affinity for Pit-1 but that cannot be ascribed a functional role can be found in the proximal promoter (P4 site) and distal enhancer (D3 site) of the PRL gene (37,38), in the D1 region of the TSH␤ promoter (18,39), and in the upstream enhancer (Pit-1 a and b sites) of the Pit-1 gene itself (34). The reason for their lack of functional FIG. 6. Stimulation of TR␤2 promoter activity in ␣-TSH cells is dependent on areas of Pit-1 interaction. 10 g of the TR␤2 5Ј deletion plasmids used in Fig. 2 were transfected by electroporation into three million ␣-TSH cells together with either 10 g of pCMV (Ϫ) or a CMVPit-1 expression plasmid (ϩ). Cells were then incubated for 40 h at 37°C and harvested, and extracts were assayed for luciferase activity. The promoterless vector pA 3 LUC and pA 3 RSV400LUC were transfected similarly in parallel reactions. The results of independent determinations (n) are expressed as fold stimulation by Pit-1 relative to the value obtained by cotransfection with the pCMV plasmid Ϯ S.E. contribution is not known, but two recent reports suggest that the context of Pit-1 sites relative to other promoter elements (40) or whether they bind Pit-1 as a monomer or dimer (41) may determine their functional significance.
The most proximal area of thyrotrope protein interaction incorporating the T-3 and T-3A protected regions was shown by deletion analysis to contribute to TR␤2 basal promoter activity in both thyrotropes and somatotropes. However constructs from Ϫ77 to ϩ40 containing only these sites but lacking the more upstream Pit-1 sites were not appreciably stimulated by coexpression of Pit-1 in ␣-TSH cells. Interestingly, although GH3 extracts appear to footprint this proximal region in a fashion indistinguishable from that generated by bacterially produced Pit-1, interaction with thyrotrope extracts was thought not to be a result of the binding of Pit-1 present in these extracts because the protection pattern was not identical to that produced by the recombinant Pit-1 preparation. A possible explanation is that another factor with similar sequence recognition properties such as a related POU homeodomain family member may, as a result of greater abundance or affinity, be precluding Pit-1 in nuclear extracts from binding. In fact the sequence motif (Pit-1 i), which colocalizes with the extract protein footprint, more closely resembles a recognition site for a member of the octamer binding family, which preferentially recognizes the sequence ATTTGCAT. In this regard Oct-1 has been shown to have significant affinity for Pit-1 sites (42) and simultaneous occupation as well as functional cooperation between Pit-1 and the octamer factor Oct-1 at certain Pit-1 sites has been described for the PRL promoter (42,43). An intriguing possibility is that competition for the Pit-1 i site by another factor enables Pit-1 to bind upstream at the g or h sites resulting in the T-3 footprint seen only with nuclear extracts. However, protection of the T-3 area by ␣-TSH extracts, which lack Pit-1 protein (Fig. 3B), argues against Pit-1 accounting for the T-3 footprint. A functional role for the variant Sp1 site is also presently undefined. However, it is interesting that Sp1 has been implicated in the regulation of human GH and chorionic somatotropin gene expression by Pit-1 (44,45) as well as playing a key role in transcriptional regulation of the TR␤1 isoform (46).
The deletion and mutation analyses as well as the Pit-1 coexpression experiments presented here demonstrate that interaction by Pit-1 at the footprinted areas containing the Pit-1 e and f motifs appears to be critical for TR␤2 promoter activity in cells of both thyrotrope and somatotrope origin, although their contribution seems to differ somewhat between the two cell types. The requirement for auxilliary factors that can influence the behavior of Pit-1 in a cell-specific manner has been reported. These include the estrogen receptor (47,48) and an ETS factor (49) for regulation of the PRL promoter, the zinc finger protein Zn-15 for GH expression (50), and an as yet unidentified factor that functionally cooperates with Pit-1 to activate the TSH␤-subunit promoter in thyrotropes (25). Differences between pituitary gland and thyrotrope tumors in the relative usage of transcription start sites (13) further suggest that expression from the TR␤2 promoter may be under the control of different promoter elements in thyrotropes and somatotropes. The ␣-subunit of the glycoprotein hormones has been shown to be dependent on different promoter sequences for its expression in pituitary thyrotropes and gonadotropes (reviewed in Ref. 51).
The importance of Pit-1 as a pituitary-specific transcription factor was first established by its ability to activate GH and PRL promoter constructs in heterologous cells, which do not express the endogenous GH or PRL genes (26). We report here that not only does Pit-1 stimulate an exogenously transfected TR␤2 promoter activity in ␣-TSH cells, in which the endogenous TR␤2 gene locus is silent (13), but the extent of stimulation approaches a level equivalent to reconstituting the TR␤2 promoter activity observed in expressing TtT-97 thyrotropes. In related studies where Pit-1 has been shown to activate other pituitary gene promoters in heterologous cells, reconstitution to the level seen when the same promoter constructs are transfected into pituitary-derived cells, which contain endogenous Pit-1 is not achieved. These include the GH promoter in CV-1 cells (50) and the PRL promoter in HeLa cells (52) and the TSH␤ promoter in ␣-TSH cells (18). In fact, stimulation of TSH␤ promoter activity in ␣-TSH cells requires coexpression of Pit-1 with Pit-1 T, a recently described thyrotrope-specific splice variant of Pit-1 (21).
The mutational analyses presented here revealed that disruption of Pit-1 binding at certain sites (e and f) had a greater effect on Pit-1 stimulation than removal of other sites by deletion (a-d). A similar heirarchy of importance of sites within the PRL and GH promoters that confer Pit-1 activation in HeLa cells cotransfected with Pit-1 expression vectors has also been reported (26). However, contrary to what is reported here for the TR␤2 promoter, the most proximal site in both the PRL and GH genes was the most critical. Another perhaps more analogous situation to the TR␤2 promoter is to be found in the Pit-1 gene itself where the more proximal of two Pit-1 sites lies downstream of the transcriptional start site (53). In this case, however, binding of Pit-1 results in an autologous down-regulation of Pit-1 expression.
In summary, we have demonstrated a requirement for Pit-1 in expression of the ␤2 isoform of TR in cells of thyrotrope and somatotrope origin. We have shown that binding at certain Pit-1 sites is important for expression in both cell types, whereas others appear to be cell type-specific. The innate complexity of the TR␤2 promoter region with regard to multiplicity of transcriptional origins interspersed with functionally important factor binding sites makes it difficult to distinguish the critical cis-active elements responsible for expression from loss of promoter activity due to removal of transcriptional start sites. We believe that the promoter deletion and mutation approaches described in this report represent a promising beginning toward unraveling the complexities of pituitary cellspecific TR␤2 expression.