Suppression of the human erythropoietin gene expression by the TR2 orphan receptor, a member of the steroid receptor superfamily.

A DNA response element, TR2RE-EPO (5'-TCTGACCTCTCGACCTAC-3') has been identified in the 3-minimal hypoxia-inducible enhancer of the human erythropoietin gene for the TR2 orphan receptor, an androgen-repressed transcription factor and a member of the steroid/thyroid hormone receptor superfamily. Electrophoretic mobility shift assay showed a specific binding with high affinity (Kd = 0.14 nM) between the TR2 orphan receptor and the TR2RE-EPO. Our data further indicated that this specific binding is not due to the homo-dimerization of the TR2 orphan receptor. In addition, reporter gene expression using chloramphenicol acetyltransferase assay demonstrated that the TR2 orphan receptor may suppress the expression of the chloramphenicol acetyltransferase activities via the TR2RE-EPO in the hypoxic/normoxic human hepatoma HepG2 cells. Finally, our in situ hybridization data also indicated that the TR2 orphan receptor and the erythropoietin transcripts can be co-expressed in mouse kidney and liver. Together, our data suggest that the human erythropoietin gene could represent the first human target gene regulated directly by the human TR2 orphan receptor.

Members of the steroid/thyroid hormone receptor superfamily are transcriptional factors that regulate the expression of target genes by binding to specific cis-acting sequences in the nuclei of animal cells (1). These nuclear receptors include receptors for steroid, thyroid, vitamin D 3 , vitamin A-derived hormones, and a large number of orphan receptors in which cognate ligands have not yet been identified (2,3). Numerous orphan receptors have been identified by low stringency hybridization screening and other cloning techniques (4, 5) (reviewed in Refs. 6 and 7). Thus, they share common modular architecture within the superfamily, including a variable Nterminal portion, a high degree of homology in the DNA-binding domain with two zinc fingers, and a putative ligand-binding domain at the C-terminal region. Consequently, physiological roles of orphan receptors have been postulated and subjected to speculation since they were initially identified. More recent efforts exploring their potential functions have demonstrated the remarkable impact of the nuclear receptor superfamily. Novel ligands or activators for several orphan receptors have been identified, for instance, 9-cis-retinoic acid, 15-deoxy-⌬ 12,14 -prostaglandin J 2 , and melatonin (5-methyoxy-N-acetyltryptamine) for retinoid X receptor (RXR), 1 peroxisome proliferator-activated receptor ␥, and retinoid Z receptors ␣ and ␤ (8 -11). In addition, some orphan receptors are constitutive transactivators or repressors, such as the TR3 orphan receptor or COUP-TF I (12)(13)(14)(15)(16)(17). Certain orphan receptors and classical steroid receptors can be activated to regulate gene transcription by modulators, such as neurotransmitters (dopamine), or by internal changes in phosphorylation pathways (6). Several orphan receptors, however, may function as co-regulators of ligand-dependent receptors to modulate ligand-mediated signaling pathways at the protein or DNA level; for example, RXR heterodimerizes with respective receptors for retinoic acid (RAR), thyroid hormone, peroxisome proliferator-activated, and vitamin D 3 (18).
The human TR2 orphan receptor is one of the first orphan receptor identified that shares structural homology with members of the steroid/thyroid hormone receptor superfamily (4,19). The TR2 orphan receptor cDNAs were isolated from both human prostate and testis cDNA libraries using a probe homologous to a highly conserved DNA-binding domain common to steroid hormone receptors. The TR2-11 orphan receptor cDNA encodes a protein of 603 amino acid residues with a calculated molecular mass of 67 kDa. We also identified a distinct set of cDNAs, named the human TR4 orphan receptor, from human prostate and testis cDNA libraries (20). The amino acid sequence of the TR4 orphan receptor is closely related to that of the TR2 orphan receptor. This high homology between the TR2 and TR4 orphan receptors highlights a unique subclass within the steroid/thyroid hormone receptor superfamily. Northern blot analysis showed that the TR4 orphan receptor could be detected in many tissues in humans and mice (20,21). The expression of the TR4 orphan receptor transcripts in the human kidney is significantly more than that in the liver. Recently, the rat TR2 orphan receptor cDNAs encoding 590 amino acids were isolated from rat prostate cDNA library (22). In addition, the genomic locus of the TR2 orphan receptor gene has been mapped to the human chromosome 12q22. 2 More recently, we demonstrated that the TR2 orphan receptor may modulate the activation of both RAR and RXR hormone response elements (HREs). This suggested that the TR2 orphan receptor may be a master regulator in the retinoic acid signal transduction pathway (23). Moreover, we have identified a TR2 orphan receptor response element (TR2RE-SV40) in the transcriptional initiation site of the SV40 major late promoter (24). This DNA response element contains a direct repeat of AG-GTCA consensus motif, and the TR2 orphan receptor may function as a repressor for SV40 gene expression.
Erythropoietin (EPO) is an essential survival and growth factor for the erythrocytic progenitor cells in the bone marrow (reviewed in Refs. [25][26][27]. This glycoprotein hormone containing 165 amino acids, with a molecular mass of 30.4 kDa, is synthesized mainly in the kidney and fetal liver in response to hypoxia in mammals (28 -30). The mechanism for the induction of the EPO gene expression by the lack of oxygen is only partially understood. EPO deficiency is the primary cause of anemia in chronic renal failure. The human EPO gene has been cloned and expressed in vitro in mammalian cell cultures (31,32). The cis-acting elements of the human EPO gene responsible for hypoxic induction were identified in both the 5Ј-promoter and 3Ј-flanking regions (33)(34)(35)(36). This 3Ј-enhancer is a highly conserved region located 120 base pairs (bp) downstream of the polyadenylation site among several species, and contributes a 4 -14-fold induction of reporter gene expression in a wide variety of cell lines (37). Recent studies have further narrowed down this enhancer to a 50-bp element consisting of at least three transcriptional factor binding sites (38). The first site, a highly conserved 9-bp near the 5Ј-end of the minimal enhancer, can be bound by a 120-kDa hypoxia-inducible factor from nuclear extracts. The second one, containing three CA repeats, has not yet been characterized. However, the last one, located at the 3Ј-end of the minimal enhancer, consists of a direct repeat of AGGTCA consensus motif separated by 2 bp (35). This site has been suggested as a potential DNA response element for a few orphan receptors, such as the hepatic nuclear factor 4, COUP-TF I, and TR2 orphan receptors (39). We are interested in knowing if the TR2 orphan receptor may bind specifically to this EPO enhancer and play a role in the control of EPO gene expression. In the present study, we developed poly-and monoclonal anti-TR2 orphan receptor antibodies as probes to explore the possible consequences of this specific binding between the TR2 orphan receptor and the EPO enhancer. Our data demonstrate that the TR2 orphan receptor may function as a repressor in EPO gene regulation. Thus, the EPO gene could represent the first identified human target gene regulated by the TR2 orphan receptor.

MATERIALS AND METHODS
Plasmid Constructions-The full-length coding region of the TR2 orphan receptor cDNA was cloned under the control of the polyhedrin promoter in the pVL1393 baculovirus transfer vector (Invitrogen), named pVL-TR2. The TR2 orphan receptor coding region was released from the pBluescript-TR2-11 plasmid by the digestion of NarI/blunt and XbaI restriction enzymes. This 2.2-kb fragment was then subcloned at the SmaI and XbaI sites of the pVL1393 vector. Plasmid pET-TR2 consists of the DNA-binding domain of the TR2 orphan receptor cDNA under the control of the T7 promoter in the pET-14b prokaryotic expression vector (Novagen). A 1.4-kb fragment digested with MspI/blunt and EcoRI/blunt from the pBluescript-TR2-11 plasmid was subcloned into the NdeI/blunt sites of the pET-14b vector. Plasmid pSG5-TR2 contains the full-length coding sequence of the TR2 orphan receptor cDNA under the control of SV40 early and T7 promoters for in vivo and in vitro expression, respectively (24). Two C-terminal deletions of the TR2 orphan receptor, pSTR2 and p3STR2, were created for the present study. The baculoviral expression plasmid, pTR2SH, constitutes the full-length TR2 orphan receptor cDNA with the 5Ј-region replaced by a fragment from a polymerase chain reaction (Bam primer: 5Ј-AGCG-GATCCTCATGGCAACCATAGAAGA-3Ј, and Sac primer: 5Ј-TACT-GAGCTCTGGCAGGCTGT-3Ј) into the pAcSG-GP67-His-NTC vector (PharMingen). Plasmid pSTR2 was then constructed by the ligation of a 1.6-kb EcoRI fragment of the pTR2SH plasmid into the EcoRI site of the pSG5 vector (Stratagene). Plasmid p3STR2 was generated by the removal of the 0.5-kb BglII fragment from the pSTR2 plasmid. For the N-terminal truncation, an 0.8-kb XbaI/blunt-SacI fragment of the pET-TR2 plasmid was cloned at NarI/blunt and SacI sites of the pSG5 vector, and termed the pCTR2 plasmid. Plasmid pSVcatEJ consists of the minimal hypoxia-inducible human EPO enhancer at the BamHI site of the pCAT-promoter vector as described by Semenza and Wang (38). Plasmid pSG5-TR2/ARp/TR2 is a chimera containing the P (proximal) box of the human androgen receptor cDNA in the coding sequences of the TR2 orphan receptor cDNA (23).
Preparation of Poly-and Monoclonal Antibodies-To obtain large quantities of the TR2 orphan receptor as antigens in raising specific poly-and monoclonal antibodies, we have employed both baculovirus and Escherichia coli expression systems. The recombinant baculoviruses overproducing the full-length TR2 orphan receptor were generated by homologous recombination after co-transfection of Spodoptera frugiperda 9 insect cells with the BaculoGold linearized baculovirus DNA (PharMingen) and the pVL-TR2 recombinant transfer plasmid by using the BaculoGold transfection kit as described previously (40). Several recombinant viruses were then plaque-purified by a one-step plaque assay. To produce the recombinant TR2 orphan receptor, S. frugiperda 9 cells infected with the recombinant baculovirus were harvested, lysed, and centrifuged as described previously (40). The protein extracts were separated on SDS-polyacrylamide gel electrophoresis, and the protein band containing the TR2 orphan receptor was isolated. To produce monoclonal antibody against the TR2 orphan receptor, the hybridoma technique was provided by PharMingen, and the procedures were as described previously (41).
The pET system is a powerful system to express recombinant proteins in E. coli (42). The expression of the TR2 orphan receptor from the pET system was performed according to the manufacturer's instruction (Novagen). Basically, plasmid pET-TR2 containing the DNA-binding domain of the TR2 orphan receptor cDNA with six consecutive histidine residues at the N terminus was transformed into the BL21(DE3)pLysS host strain (Novagen), and cultured in NZCYM (10 g NZ amine, 5 g NaCl, 5 g yeast extract, 1 g casamino acids, 2 g MgSO 4 ⅐7H 2 O (pH 7.5) per liter) medium until the OD 600 reached 0.6 with shaking at 37°C. Isopropyl ␤-D-thiogalactopyranoside was added to a final concentration of 0.5 mM, and the incubation was continued for an additional 3 h. E. coli were harvested by centrifugation at 7,000 rpm for 5 min at 4°C, and resuspended in one-fourth of the culture volume of the binding buffer (40 mM Tris-HCl (pH 7.9), 5 mM imidazole, and 0.5 M NaCl). Bacteria were lysed through one freeze/thaw cycle followed by homogenization (OMNI 2000, Omni International). The lysates were centrifuged at 3,000 rpm (Beckman GPR) for 20 min at 4°C. The cellular extractions were then either analyzed on SDS-polyacrylamide gel electrophoresis followed by Coomassie Blue staining, or purified by a onestep metal chelation chromatography (Novagen). For the production of polyclonal antibody, the purified peptide was emulsified with Freund's complete adjuvant and injected intradermally into New Zealand White rabbits (Medical School Animal Care Unit, University of Wisconsin-Madison) as described previously (41).
Coupled in Vitro Transcription and Translation-Circular plasmids containing the full-length TR2 orphan receptor cDNA and its N-and C-terminal truncations were in vitro transcribed and translated simultaneously in the TNT-coupled reticulocyte lysate system according to the manufacturer's instructions (Promega). Depending on the purpose of the experiment, the reactions were performed in the presence or absence of L-[ 35 S]methionine in the transcription-translation mixture. After protein synthesis, ZnCl 2 was added to a final concentration of 0.5 mM as previously reported (24,43). The in vitro translated products were then analyzed directly by either electrophoresis in SDS-12% polyacrylamide gel or electrophoretic mobility shift assay (EMSA).
Electrophoretic Mobility Shift Assay-EMSA was carried out as described previously (24). Double-stranded oligonucleotides corresponding to the human EPO gene nucleotide numbers 3481-3498 (5Ј-TCT-GACCTCTCGACCTAC-3Ј) were end-labeled by [␥-32 P]ATP as a probe (32) (GenBank TM accession no. M11319). For competition reactions, cold double-stranded oligonucleotides were mixed with the labeled probe prior to the addition to the reactions. For antibody supershift analysis, 1 l of either the polyclonal or the monoclonal anti-TR2 orphan receptor antibody was added into the reactions for 15 min at room temperature prior to loading on a 5% native gel.
DNA-Protein Binding Affinity Assay-DNA-protein binding affinity assay was performed as described previously (24) with modifications that will be described elsewhere. Briefly, the free probe and DNAprotein complexes resolved by EMSA were quantified by PhosphorImager (Molecular Dynamics). The dissociation constant (K d ) value was determined from the negative reciprocal of the slope of the line generated from the experimental data.
Cell Cultures and Transfections-The human hepatocellular carcinoma cell line HepG2 (American Type Culture Collection, HB-8065) was maintained in Dulbecco's modified Eagle's medium/F12 nutrient mixture (Life Technologies, Inc.) supplemented with penicillin (100 units/ml), streptomycin (100 g/ml), amphotericin (0.25 g/ml), and 5% heat-inactivated (56°C for 30 min) fetal bovine serum (Harlan). Cells were cultured in a humidified 5% CO 2 , 95% air incubator at 37°C. Transient transfection of the HepG2 cells plated at an initial density of 3 ϫ 10 5 /60-mm dish was carried out using a calcium phosphate-DNA precipitation method as described previously (44,45). To normalize the transfection efficiency, the ␤-galactosidase expression plasmid was cotransfected. In the hypoxia stimulation condition, a cobalt chloride (100 M)-containing environment was introduced 16 h after transfection, and the incubation was continued for an additional 24 h as described previously (30,46). The normalized ratio of chloramphenicol acetyltransferase (CAT) expression at 100 M of CoCl 2 or normal conditions was averaged over at least three independent experiments for each point, with error bars designating standard deviation, as described previously (39).
In Situ Hybridization-For sample preparation, embryos from C57BL/J mice (Harlan) were collected at 14.5 and 16.5 days of gestation. The middle of the day of the vaginal plug was considered as 0.5 day post coitum (dpc). Samples were fixed in 4% paraformaldehyde/phosphate-buffered saline, dehydrated in ethanol, cleared with xylene, and embedded in paraffin. Serial sections (6 -8 m) were collected on poly-L-lysine-coated slides, air dried, and stored at 4°C under desiccation. For probe preparation, the cDNA fragment that covers the N-terminal region of the mouse TR2 orphan receptor from the translation initiation site to nucleotide position 277 was subcloned in pBluescript SKϩ (Stratagene). Both [ 35 S]-uridine 5Ј-(␣-thio)triphosphate-labeled antisense and sense RNA probes were synthesized by T3 and T7 polymerases, respectively, according to manufacturer's instructions (MAXIscript, Ambion). The probes were partially degraded to 150 -300 bp by limited alkaline hydrolysis. Unincorporated nucleotides were removed by chromatography through a Sephadex G-50 column (Pharmacia Biotech Inc.).
In situ hybridization was performed mainly as described elsewhere. 3 Briefly, mouse embryo sections were deparaffinized, hydrated, and then treated with proteinase K (Boehringer Mannheim) at 20 g/ml for 7.5 min. After washing in phosphate-buffered saline, the sections were fixed in 4% paraformaldehyde, acetylated, dehydrated, and air-dried. These sections were hybridized at 52°C for 17 h with cRNA probes (the specific activity for both sense and antisense probes reached 1 ϫ 10 9 cpm/g). The probe (10 6 cpm) was included in the hybridization buffer containing 50% formamide for each slide. Washes were performed with high stringency (2 ϫ SSC, 50% formamide at 65°C (1 ϫ SSC ϭ 0.15 M NaCl, 0.015 M sodium citrate, pH 7.0)) before and after RNase digestion (20 g/ml for 30 min). Slides were dipped into Kodak NTB2 emulsion and exposed for 6 weeks. Subsequently, slides were developed in Kodak D19 developer, fixed, dehydrated, and mounted for light-field analysis.

RESULTS
Production of Anti-TR2 Orphan Receptor Antibodies-Using both baculovirus and E. coli expression systems, we were able to generate large quantities of the human TR2 orphan receptor for the production of poly-and monoclonal antibodies. As shown in Fig. 1A, the full-length coding region of the TR2 orphan receptor cDNA was constructed into the pVL1393 baculovirus transfer vector. The standard baculovirus expression system was performed and subjected to monoclonal antibody production as detailed under "Materials and Methods." At least seven anti-TR2 orphan receptor monoclonal antibodies were produced, including G204-218.48 and G163-23 with IgM and IgG1 isotopes, respectively. Monoclonal antibody G204-218.48 specifically recognizes both the purified TR2 orphan receptor (Fig. 1B, lane 4) and the recombinant TR2-baculovirus-infected cellular lysate (lane 3). However, this antibody will not recognize either the S. frugiperda 9 cellular lysate (lane 1) or the wild-type baculovirus-infected cellular lysate (lane 2). We also produced polyclonal anti-TR2 orphan receptor antibody (no. 1132) using the purified DNA-binding domain of the TR2 orphan receptor generated from the E. coli-pET system (data not shown).
The TR2 Orphan Receptor Binds Specifically to the 3Ј-Minimal Hypoxia-inducible Enhancer of the Human EPO Gene-We have identified a DNA response element (TR2RE-SV40) containing an imperfect direct repeat of AGGTCA consensus motif in the transcriptional initiation site of the SV40 major late promoter for the TR2 orphan receptor (24). Since one of the transcriptional factor binding sites of the 3Ј-enhancer of the human EPO gene contains a core recognition motif similar to that of the TR2RE-SV40 (39), we were interested in determining if the TR2 orphan receptor might bind specifically to this site and play a role in the control of EPO gene expression. The TR2 orphan receptor was expressed in a coupled in vitro transcription and translation system from the pSG5-TR2 plasmid (data not shown). Double-stranded oligonucleotides corresponding to the human EPO gene nucleotide numbers 3481-3498 located within the 50-bp minimal hypoxia-inducible enhancer were synthesized (38). Both the in vitro expressed TR2 orphan receptor and the DNA element were used in the EMSA as shown in Fig. 2. A specific DNA-protein complex was revealed when the TR2 orphan receptor was incubated with the DNA probe (lane 3, arrowhead). This complex could be eliminated in the presence of 500-fold molar excess of unlabeled oligonucleotides (lane 4). Moreover, the polyclonal anti-TR2 orphan receptor antibody could further supershift this DNAprotein complex (lane 6, arrow). In contrast, the monoclonal anti-TR2 orphan receptor antibody (G163-23) can abolish such a specific complex (lane 7). These data demonstrated that the TR2 orphan receptor can specifically bind and form a single complex with this DNA element of the 3Ј-EPO enhancer.
To determine the DNA-protein binding affinity between the TR2 orphan receptor and the EPO enhancer, we performed Scatchard binding analysis by the EMSA. As shown in Fig. 3, constant amounts of the TR2 orphan receptor (60 ng) were incubated with different concentrations of the DNA probe (0.4 -12.8 ng). DNA-protein complexes were resolved in the EMSA (Fig. 3A). Scatchard plot analysis resulted in a single binding component for the specific DNA-protein complex with a dissociation constant (K d ) of 0.14 nM and B max of 0.005 nM (Fig. 3B). These results fit well into the range of K d for classical steroid receptors and their HREs.
Domain Architecture of the TR2 Orphan Receptor in the Recognition of the 3Ј-EPO Enhancer-Amino acid comparisons of orphan receptors with classical steroid receptors indicate that they share the same modular architecture (6,7). To investigate the possible interaction of each domain in the TR2 orphan receptor in the recognition of the EPO enhancer, we generated several deletion mutants of the TR2 orphan receptor and tested their binding properties in the EMSA. Four differ-ent variants including the intact TR2 orphan receptor, two C-terminal, and one N-terminal truncations were constructed into the same pSG5 vector (Fig. 4A). These plasmids were transcribed and translated in vitro to produce proteins with expected molecular masses of 67, 61, 33, and 62 kDa, respectively (Fig. 4B, lanes 2-5). In contrast, the mock-translated control expressed no detectable product (lane 6). We then surveyed the interaction between these variants and the EPO enhancer by the EMSA. A specific DNA-protein complex was seen as before in Fig. 2 between the intact TR2 orphan receptor and the probe (Fig. 4C, lanes 1-4, medium arrowhead). The polyclonal anti-TR2 orphan receptor antibody could further supershift this specific DNA-protein complex (lane 5, large arrowhead). The extreme C-terminal truncated TR2 orphan receptor (STR2) with the deletion of 71 amino acid residues behaved in a manner similar to that of the intact TR2 orphan receptor (lanes 6 -8). However, the binding affinity between this truncation and the DNA element was weaker than that of the intact TR2 orphan receptor. On the other hand, the Cterminal (3STR2) and N-terminal (CTR2) variants are capable of forming smaller DNA-protein complexes which, however, appear to be able to co-migrate with one of the nonspecific complexes (lanes 9 -14, small arrowhead). Interestingly, we detected the original DNA-protein complexes only when the DNA probe was incubated with the intact TR2 orphan receptor and one of its truncations (lanes 15-23). Taken together, these results indicated strongly that homodimerization does not occur in this reaction for the TR2 orphan receptor.
Suppression of EPO Gene Expression by the TR2 Orphan Receptor via the 3Ј-EPO Enhancer-To determine whether the TR2 orphan receptor can play any modulatory role in EPO gene expression via the interaction with the enhancer DNA element, we carried out the CAT assay with the co-transfection of the TR2 orphan receptor expression vectors and CAT reporter plasmids into human hepatoma HepG2 cells under hypoxic and normoxic conditions. As shown in Fig. 5, the co-transfection of expression vectors containing either the full-length TR2 orphan receptor cDNA (lanes 2 and 5) or the chimeric TR2 orphan receptor (TR2/ARp/TR2) cDNA (lanes 3 and 6), and reporter plasmids consisting of either the parent pCAT-promoter ( lanes  1-3) or pSVcatEJ (lanes 4 -6) demonstrated that the TR2 orphan receptor can cause repression in EPO expression. In the presence of the EPO enhancer element, the TR2 orphan receptor could repress the transcriptional CAT activity to 68% under the normoxic conditions (lane 5). Moreover, CAT activity would be further suppressed to 23% by the TR2 orphan receptor in the presence of cobalt chloride. In contrast, this repression could be eliminated when the chimeric TR2/ARp/TR2 orphan receptor, a construct replacing only the P box region of the DNA-binding domain in the TR2 orphan receptor with that of the human androgen receptor, was applied (lane 6). These results sug- FIG. 4. Domain architecture of the TR2 orphan receptor in the recognition of the 3-EPO enhancer. A, schematic structure of various truncations of the TR2 orphan receptor. Plasmid pSG5-TR2 contains the full-length TR2 orphan receptor coding region, whereas plasmids pSG5-STR2, pSG5-3STR2, and pSG5-CTR2 represent two C-terminal and one N-terminal truncations of the TR2 orphan receptor, respectively. The DNA-binding domain (DBD) is included in these constructs. Each number shows the amino acid residue number within the TR2 orphan receptor cDNA. Molecular masses of the intact TR2 orphan receptor, two C-terminal, and one N-terminal truncations of the TR2 orphan receptors are indicated. B, analysis of the in vitro expressed TR2 orphan receptor and its variants in SDS-12% polyacrylamide gel electrophoresis. Lanes 1 and 6 show 14 C-labeled methylated protein standards and mock-translated products, respectively. Lanes 2-5 display the intact TR2 orphan receptor, two C-terminal, and one N-terminal truncations, respectively. C, binding of the binary mixture of the TR2 orphan receptor and its variants to the EPO enhancer. Lane 1 displays the DNA probe alone. Binding reaction mixtures incubated with the probe and mock-translated product (lane 2), the intact TR2 orphan receptor (lanes 3-5 and 15-23), the extreme C-terminal truncated TR2 orphan receptor (lanes 6 -8 and 15-17), the C-terminal deleted TR2 orphan receptor (lane 9 -11 and 18 -20), or the N-terminal truncated TR2 orphan receptor (lanes 12-14 and 21-23), in the presence of unlabeled oligonucleotides (lanes 4, 7, 10, 13, 16, 19, and 22), or polyclonal anti-TR2 orphan receptor antibody (lanes 5, 8, 11, 14, 17, 20, and 23) are shown. The retarded complexes are indicated by small, medium, and large arrowheads for the DNA-truncated TR2 orphan receptor complexes, the DNA-intact TR2 orphan receptor complexes, and DNAprotein-antibody complexes, respectively.
FIG. 5. The TR2 orphan receptor represses the EPO gene expression by the CAT assay. Human hepatoma HepG2 cells were co-transfected with expression vectors containing either the full-length TR2 orphan receptor expression plasmid (pSG5-TR2, lanes 2 and 5) or the chimeric TR2 orphan receptor expression plasmid (pSG5-TR2/ARp/ TR2, lanes 3 and 6), and reporter plasmids consisting of either the parent reporter pCAT-promoter plasmid (lanes 1-3) or pSVcatEJ plasmid (lanes 4 -6) in normoxia (open bar) or hypoxia (closed bar). Plasmid pSVcatEJ contains a 50-bp fragment of the minimal hypoxia-inducible enhancer in the pCAT-promoter vector (38). All CAT assays were standardized for the level of ␤-galactosidase activity. The normalized ratio of relative CAT activity at hypoxia (100 M of CoCl 2 ) or normoxia is shown. Each value represents the average of at least three independent experiments with the error bar designating standard deviation. gested that the TR2 orphan receptor might repress the EPO gene expression via the interaction between the DNA-binding domain of the TR2 orphan receptor and the 3Ј-EPO enhancer.
Co-expression of the TR2 Orphan Receptor and EPO in Fetal Mouse Liver, Kidney, and Brain-The kidney and the liver are the major organs known to produce EPO in mammals (29,47,48). Within the mammalian embryo, EPO is produced primarily by the fetal liver during midgestation, and by the kidney from late gestation to adulthood. Some evidence has also shown that the brain can produce EPO (49). Using in situ hybridization, Koury et al. (29) demonstrated that the cells containing EPO mRNA are localized in the interstitium between tubules in the renal inner cortex, whereas the EPO transcripts in the liver are found predominately in cells surrounding the central vein (50). We, therefore, wanted to determine whether the expression pattern of the TR2 orphan receptor is correlated with EPO production in the same cell types or tissues. As shown in Fig. 6A, our results demonstrated that the TR2 orphan receptor was expressed in mouse liver, kidney, brain, and other tissues at 14.5 dpc (midgestation stage). At 16.5 dpc, the signal of the TR2 orphan receptor in the liver was reduced significantly, but was still higher than background (compared to the level in the pancreas, which has been set as background) (Fig. 6, B and C). In contrast, the signals in the kidney and certain regions of the brain remained strong (Fig. 6B). At higher magnification, the labeling signal in kidney was found in the glomeruli, the tubuli, and the interstial tissues between tubules (Fig. 6D). Together, our results clearly demonstrate that the TR2 orphan receptor and EPO transcripts could be co-localized in the liver, kidney, and brain. DISCUSSION The mechanism for hypoxia or cobalt chloride triggering the increased production of the EPO mRNA and protein remains one of the major unsolved mysteries in EPO gene regulation (26,30). In this study, we have demonstrated that the human TR2 orphan receptor may suppress the human EPO gene expression via the 3Ј-transcriptional enhancer. This implies that FIG. 6. Localization of the TR2 orphan receptor transcripts in the mouse embryo. All in situ hybridization experiments were performed by the 35 S-UTP-labeled antisense mouse TR2 orphan receptor riboprobes. Exposure time was 6 weeks for all slides. Photomicrographs of autoradiograms are presented for sagittal sections of mouse embryos at 14.5 dpc (A) and 16.5 dpc (B-D). Tissues and organs with strong hybridization signals (dark areas in bright field) are labeled, for example, the brain (b). C, photoemulsion-dipped sections at higher magnification showed strong signals within the kidney (k) but weak within the liver (li), while the pancreas (p) served as a negative control. D, magnified sections in the kidney revealed signals in the developing glomeruli (gl), proximal tubules (pt), and interstitium between tubules (int). The bars represent 1 and 0.5 mm of length in panels A-C and D, respectively. the TR2 orphan receptor may function as a negative modulator in EPO gene regulation. It is interesting to note that androgenic steroids have been shown to increase the production of EPO in mammals (51,52). Prior to the introduction of recombinant human EPO in 1985, androgenic therapy was widely used by clinicians, and is still used occasionally (53). More recently, clinical trials have shown that replacement therapy with recombinant human EPO can benefit patients with anemia of chronic renal failure, myelodysplastic syndrome, acquired immunodeficiency syndrome, hemoglobinopathies, or malignancies (26,27,53). However, not all patients respond to or benefit from the treatment of recombinant human EPO (53). Recently, our data indicated that androgens can repress the expression of the TR2 orphan receptor mRNA in human prostate LNCaP cells and rat ventral prostate (4,22). It is possible that one of the potential androgenic effects for the induction of EPO expression may be indirectly involved in the suppression of the TR2 orphan receptor-mediated repression mechanism for EPO expression. Other possibilities include androgens directly increasing the EPO gene expression via the potential androgen response elements located at the 5Ј-or 3Ј-flanking region of the human EPO gene or stimulating the proliferation of erythrocytic progenitors in bone marrow (54 -56).
For several years, it was widely assumed that members of the steroid receptor superfamily were capable of binding to response DNA elements in three fundamentally different ways, monomeric, homodimeric, and heterodimeric categories (57). A monomeric receptor can bind to a single copy of a core recognition motif (such as NGFI-B); two receptors can bind to two copies of a core consensus sequence, resulting in either homoor heterodimers (such as glucocorticoid receptor and RAR-RXR, respectively). We initially tried to investigate whether the homodimerization occurs between the EPO 3Ј-enhancer and the TR2 orphan receptor. Our results, however, indicate that the TR2 orphan receptor and its different deletion variants may individually recognize the EPO enhancer element (Fig. 4). Furthermore, we were unable to detect any potential interaction between the full-length TR2 orphan receptor, RXR, and RAR using a direct repeat of the promoter of the cellular retinolbinding protein type II gene (58) as a probe (23). This again ruled out the possibility that the TR2 orphan receptor could form heterodimers with either RXR or RAR. Thus far, there is no sufficient evidence to show the TR2 orphan receptor forms either homo-or heterodimers by itself or with other receptors. Moreover, we observed similar phenomena for that of the TR4 orphan receptor, a close relative of the TR2 orphan receptor, during the study of gene expression on the SV40 major late promoter (43).
Our data also suggest that suppression of EPO gene expression by the TR2 orphan receptor is accomplished in a DNA-dependent manner. It is noted that the HREs for steroid receptors are structurally related but functionally different (59). Based on the zinc finger model, five amino acid residues at the Cterminal region of the first zinc finger are designated as the P box which is important in protein-DNA interaction (59). The TR2 orphan receptor contains Glu-Gly-Cys-Lys-Gly amino acid sequences in the P box, whereas the human androgen receptor belonging to a different subfamily of the HREs consists of Gly-Ser-Cys-Lys-Val amino acid sequences in the same box. Our data showed that the TR2 orphan receptor may repress the EPO gene expression via the 3Ј-enhancer region (Fig. 5). In contrast, our data also showed such repression could be abolished when the chimeric TR2/ARp/TR2 orphan receptor, a construct replacing only the P box of the DNA-binding domain in the TR2 orphan receptor with that of the androgen receptor, was tested. These results highlight the importance and speci-ficity of the protein-DNA interaction during EPO regulation by the TR2 orphan receptor.
In addition to several hypoxia-inducible nuclear factors discovered thus far, a few orphan receptors have been reported to be involved in the key regulation of EPO gene expression via the 3Ј-enhancer (35,39). Based on our data described above, we hypothesize that the repression model of the TR2 orphan receptor in EPO gene transcription is similar to that of COUP family members (39). Thus, the TR2 orphan receptor may be one of the contributors, among the complicated network for binding the same DNA sequence of the EPO gene, and specifically compete with activators, such as transcriptional factors or orphan receptors (e.g. hepatic nuclear factor 4). The degree of inhibition could be dependent on which factor is in excess during hypoxia. As yet, we have no evidence that the TR2 orphan receptor can antagonize any positive regulator by competing for binding to the EPO enhancer. Therefore, the relative levels of the TR2 orphan receptor and other activators may control the switch of EPO production in response to hypoxia or cobalt.
It has long been known that EPO gene expression is highly tissue-specific and its expression levels increase in response to the severity of anemia (28 -30). By using in situ hybridization, Koury et al. (29) showed that EPO mRNA positive cells were not seen in any glomeruli, but were easily seen in tubular areas within the basement membrane in the anemic kidney. However, nonanemic kidneys even in the same section on the same slide exhibited no such labeling in either the cortex or medulla (29), indicating that EPO expression is very low under normal conditions and may be undetectable by some methodologies including in situ hybridization. In addition, cell types expressing EPO in the kidney and liver assayed by in situ hybridization have been well documented (29), and we, therefore, believe it is appropriate to compare directly the expression pattern of the TR2 orphan receptor with that of EPO as revealed by Koury's group.
Our in situ hybridization results showed that both the TR2 orphan receptor and EPO transcripts could be co-localized in the liver, kidney, and brain with the expression pattern correlated with the liver-to-kidney shift of EPO production during mouse development. In addition, the earliest TR2 orphan receptor expression in the mouse embryo was detected at 9.5 dpc, 3 while the ontogeny for EPO is not exactly defined. In transgenic mice with the EPO-null mutation, embryos died at about 13 dpc (60), suggesting that the timing for both EPO and the TR2 orphan receptor expression may be correlated. However, the tissue distribution of the TR2 orphan receptor in the mouse is much broader than that of EPO and is proposed to be position-specific, instead of tissue-specific. 3 Thus, the TR2 orphan receptor transcripts are highly expressed in most tissues undergoing early but not terminal differentiation. 3 One possible explanation is that the TR2 orphan receptor may have several target genes with EPO being just one of them. The physiological significance of interaction between the TR2 orphan receptor and EPO can be revealed by the overexpression of the TR2 orphan receptor in transgenic mice to see whether these transgenic mice develop an anemic phenotype.
In summary, our data indicate that the EPO gene is the first identified human target gene regulated by the TR2 orphan receptor. In addition, the TR2 orphan receptor may function as a repressor in the complicated EPO gene regulation. The finding and characterization of the TR2RE-EPO here may further help us to isolate more physiological target genes of the TR2 orphan receptor in the future.