Characterization of the response element and DNA binding properties of the nuclear orphan receptor germ cell nuclear factor/retinoid receptor-related testis-associated receptor.

Recently, we have reported the cloning of the germ cell-specific, nuclear orphan receptor germ cell nuclear factor (GCNF)/RTR. In this study, we characterize the RTR response elements by an electrophoretic mobility shift assay/polymerase chain reaction-based, DNA binding site selection strategy. RTR binds with the greatest affinity to response elements containing TCA(AG(G/T)TCA)2 (consensus RTR response element; conRTRE), to which it binds as a homodimer. RTR is also able to bind as a monomer to a single core motif TCAAG(G/T)TCA, albeit with a lower affinity. Mutation analysis supports the specific requirements of the 5'-flanking sequence and the core motif of the RTRE for optimal binding of RTR. An RTR-specific antiserum (RTR-Ab2) was raised that causes supershift of the RTR-conRTRE complex in EMSA. Based on the sequence of the conRTRE, we located a putative RTRE, referred to as P2-RE, in the 5' promoter-flanking region of the mouse protamine 2 gene, which is induced during the same stage of spermatogenesis as RTR. The ability of RTR-Ab2 to cause a supershift of an RTR-RTRE complex with nuclear extracts from different tissues correlated with the tissue- and development-specific expression of RTR. Transfection of RTR in CV-1 cells was unable to cause RTRE-dependent transactivation of a CAT reporter gene; however, an RTR-VP16 fusion protein could induce transactivation through several RTREs, including P2-RE.

The nuclear receptor superfamily constitutes a class of liganddependent transcription factors that regulate gene expression during many biological processes, including development, cellular proliferation, and differentiation (1)(2)(3)(4)(5). This superfamily includes receptors for steroid hormones, retinoids, and vitamin D, and an increasing number of orphan receptors for which the ligand has not yet been identified (6 -9). Members of this superfamily share a common modular structure consisting of four major domains (6, 10 -13). The DNA-binding domain is the most conserved among the nuclear receptors and is composed of two "zinc finger" motifs that play a role in DNA recognition and protein-protein interactions (10,14). The ligand-binding domain at the C terminus is involved in several functions including ligand binding, receptor dimerization, nuclear transloca-tion, and transcriptional activation (11,(13)(14)(15)(16). The functions of the N-terminal region and the hinge domain that separates the DNA-binding and ligand-binding domains are still poorly understood. Recent studies have indicated that in certain instances the amino-terminal domain contains a transactivation function (17) and that the hinge domain contain sites that interact with co-repressors (18).
Nuclear receptors control the transcription of target genes by binding to DNA sequences referred to as hormone response elements (14,16). Most members of this superfamily bind, as homodimers or heterodimers, to cis-acting DNA sequences that contain two half-site core motifs of RGGTCA configured in a direct repeat, a palindrome, or an inverted palindrome separated by spacers of different lengths (2,7,14,16,19,20). A number of orphan receptors, including members of the ROR/ RZR family (21)(22)(23), NGF1-B/Nur77 (24), and FTZ-F1/SF-1 (24 -26) bind as monomers to hormone response elements consisting of a single core motif, RGGTCA, preceded by an AT-rich sequence.
Recently, we reported the cloning of the novel nuclear orphan receptor RTR from a mouse testis cDNA library (27). This receptor has also been cloned from rat and has been named germ cell nuclear factor (GCNF 1 ; Ref. 28). This orphan receptor is highly expressed in the testis and in particular in round spermatids (27). These observations suggest that GCNF/RTR functions as an important transcriptional factor in the regulation of gene expression during a very specific stage of spermatogenesis.
In this study, we have analyzed the characteristics of the RTR response element (RTRE) more precisely by a DNA binding site selection strategy that is based on a combination of polymerase chain reaction (PCR) and electrophoretic mobility shift assay (EMSA) and by mutation analysis. Based on the consensus RTRE sequence obtained, we located a putative RTR response element (P2-RE) in the 5Ј promoter-flanking region of the mouse protamine 2 gene that is induced at the same stage of spermatogenesis as GCNF/RTR (29,30) and demonstrated that GCNF/RTR specifically interacts with this DNA sequence. We provide evidence showing that GCNF/RTR can mediate transactivation of a reporter gene through either the consensus RTRE or P2-RE. These results suggest that the protamine 2 gene may belong to a subset of genes that are regulated by GCNF/RTR during the postmeiotic phase of spermatogenesis.

EXPERIMENTAL PROCEDURES
Plasmids-The plasmid pBSK-RTR containing the full-length mouse RTR cDNA in pBluescript II SK(Ϫ/ϩ) (Stratagene) has been described (27). The expression construct pZeoSV-RTR was made by inserting the XbaI-KpnI fragment into the SpeI-KpnI site of the pZeoSV vector (Invitrogen). The plasmid pGEM3Z-RTR, which was used for in vitro translation of RTR, was created by inserting the XbaI-KpnI fragment of pBSK-RTR containing the full-length RTR into the expression vector pGEM3Z (Promega). The CAT reporter gene constructs were created by inserting three copies of the consensus RTRE (conRTRE) (TCAAGGT-CAAGGTCATAA) 3 , the half-site RTRE (GGAATCAAGGTCATCT) 3 , P2-RE (TAGAAGTTCAAGGTCATTC) 2 TAGAAGTTCAAGGTCAT or a single copy of conRTRE into the HindIII and BamHI sites of pBLCAT5 (31) to create the reporter plasmids of (conRTRE) 3 -CAT, (1/2-RTRE) 3 -CAT, (P2RE) 3 -CAT, and (conRTRE) 1 -CAT. The pSG5-VP16-RTR chimeric expression plasmid that encodes a fusion protein consisting of the VP16 activation domain and the full-length coding region of the RTR was created as follows. pGEM3Z-RTR was cut first with EclXI and XhoI enzymes to obtain the full-length RTR fragment, which was then inserted into the Bsp120 I and XhoI sites of pBluescript II SK to create pBSK-RTR-2. pSG5-VP16-RTR was obtained by inserting the KpnI-XhoI fragment of pBSK-RTR-2 into the same sites of pSG5-VP16. The pSG5-VP16 plasmid was obtained from Dr. J. Lehmann (Glaxo-Wellcome) and described previously (32). Expression of fusion proteins VP16-RTR was verified by in vitro translation and its subsequent characterization by polyacrylamide gel electrophoresis and EMSA. pGEM3Z-RTR 406 , which encodes an RTR truncated at amino acid 406, was created by inserting stop codons at the NsiI site.
Generation of Anti-RTR Antiserum-A peptide DSDHSSPGNRAS-ESNQPSC (RTR-pep2) encoding amino acids 189 -206 of RTR was synthesized and cross-linked via maleimide to KLH using the Imject kit from Pierce. The antigen was mixed 1:1 with Freund's adjuvant and injected subcutaneously into New Zealand White rabbits as described (33). Immunization boosts were given every 4 weeks. Antisera were tested by immunoprecipitation using protein A-agarose as described (33) and in EMSA using in vitro synthesized RTR. The obtained antiserum RTR-Ab2 was used at a 200-fold dilution.
Nuclear Extract Preparation-Nuclear extracts from brain, kidney, testis, and liver of 3-month-old CD-1 mice were prepared as described (34,35). Nuclear extracts from juvenile mice at day 17 after birth were also prepared.
EMSA-The SP6 RNA-coupled reticulocyte lysate system (Promega) was used to synthesize RTR protein. Oligonucleotide probes for EMSA were end-labeled with [␥-32 P]ATP by T4 polynucleotide kinase (Promega). Approximately 0.2-0.5 ng (50,000 cpm) of the probe was used in a binding reaction with 2 l of RTR programmed reticulocyte lysate or 1-2 l of nuclear extract (10 -20 g of protein) in a buffer containing 20 mM HEPES (pH 7.9), 50 mM KCl, 2.5 mM MgCl 2 , 1 mM DTT, and 10% glycerol. To prevent nonspecific binding, 1 g of poly(dI-dC), 1 g of salmon sperm DNA, and 0.5 g of nonspecific oligonucleotide were included in the reaction buffer. The programmed lysates/nuclear extracts were first incubated with reaction buffer for 10 min at room temperature and then in the presence of the radiolabeled probe with or without competitor for 30 min. As a control, probes were incubated with the same amount of unprogrammed lysate. For supershift assays, 1 l of RTR-Ab2 (1:200 final dilution) was preincubated with the in vitrotranslated RTR for 30 min at room temperature prior to the addition of radiolabeled probe. The RTR-nucleotide complexes were separated on 5% nondenaturing polyacrylamide gels (National Diagnostics) containing 0.5 ϫ TBE.
Selection of RTR Binding Sites-To select for the DNA binding sites of RTR, a similar procedure was used as reported previously (21). A mixture of 70-bp DNA fragments was synthesized by PCR using the random oligomer 5Ј-CGCGGATCCTGCAGCTCGAGN 30 GTCGACAAG-CTTCTAGAGCA-3Ј as a template and 5Ј-CGCGGATCCTGCAGCTCG-AG-3Ј (primer 1) and 5Ј-TGCTCTAGAAGCTTGTCGAC-3Ј (primer 2) as the forward and reverse primers, respectively. In a separate selection, the oligomer 5Ј-CGCGGATCCTGCAGCTCGAGN 7 GGTCAAGGTCA-N 7 GTCGACAAGCTTCTAGAGCA-3Ј was used as template. The PCR amplification was carried out using 20 pmol of oligomer, 100 pmol of 32 P-end-labeled forward primer, and 100 pmol of reverse primer for three cycles under the following conditions: 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C for each cycle. The double-stranded mixed DNA fragments generated were purified and incubated with in vitro synthesized RTR, and complexes were analyzed by EMSA. A band corresponding to the RTR-RTRE complex was excised, and the DNA was eluted in TE buffer. Recovered DNA was amplified by PCR for 15 cycles and used for EMSA analysis with RTR under the conditions described above. This procedure was repeated four times. In the fifth round, PCR products were cloned into TA vector (Invitrogen). The inserts from individual white colonies were amplified and used as competitors in EMSA. DNA that competed in EMSA effectively was subjected to sequence analysis. The sequences of 37 independent clones were analyzed.
Transactivation Assay-CV-1 cells were plated in 6-well dishes containing minimal essential medium and 10% fetal bovine serum. The next day, cells were transfected in Opti-MEM (Life Technologies, Inc.) with the expression vector pSG5-VP16-RTR or pZeoSV-RTR and pBLtk-CAT reporter DNA under the control of various response elements using lipofectamine (Life Technologies) as described (19,23). The plasmid ␤-actin-LUC was used as an internal control to correct for differences in transfection efficiency. The reporter constructs used were (conRTRE) 3 CAT, (1/2-RTRE) 3 -CAT, (P2-RE) 3 -CAT, and (conRTRE) 1 -CAT. Cells were collected 48 h after transfection, and CAT activity was determined with an enzyme-linked immunosorbent assay kit (Boehringer Mannheim) following the manufacturer's instructions.

RESULTS
Identification of the Consensus DNA Binding Sequence for RTR-The amino acid sequence in the C-terminal region of the first zinc finger of nuclear receptors is key in classifying their DNA binding specificity (11,12,14). In RTR, this sequence is CEGCKG, suggesting that RTR, like the retinoid, thyroid hormone, vitamin D, and peroxisome proliferator activating receptors, recognizes response elements containing the consensus core motif AGGTCA (2,14,16). On this basis, we screened a series of well characterized natural and synthetic response elements that contain the consensus core motif RGGTCA configured either as a direct repeat, palindrome, or inverted palindrome spaced by different numbers of nucleotides for RTR binding. These results showed that RTR was able to bind to a DR0 sequence (not shown).
To determine in an unbiased manner the consensus sequence of the response element that binds RTR with high affinity (RTRE), we used a DNA-binding site selection strategy that is based on a combination of PCR and EMSA. A mixture of 70-bp oligonucleotides containing 30 random nucleotides flanked by two 20-bp primers was used as template (see "Experimental Procedures"; Fig. 1A). After four rounds of selection with in vitro synthesized RTR protein, a strong radiolabeled band consisting of RTR-oligonucleotide complexes was observed in EMSA. The PCR products generated after the fourth selection were cloned into the TA vector, and the sequence of 37 independent clones was analyzed as described under "Experimental Procedures" (Fig. 1A). From these sequences the percentage of A, G, T, and C at each position was calculated, and the RTRE consensus sequence was derived (Fig. 1B). These data demonstrated that RTR bound most effectively to a DR0 consisting of a direct repeat of the core motif AG(G/T)TCA. In addition, these results showed that a DR0 preceded by TCA (Ϫ9 to Ϫ7 position) was preferred for high affinity binding, indicating that the 5Ј-flanking region rather than the 3Ј-flanking region is important for optimal binding of RTR. The Ϫ10 position did not reveal a strong requirement for a particular nucleotide. In a second DNA binding site selection experiment, a mixture of 65-bp oligomers consisting of 5Ј-CGCGGATCCTGCAGCTCG-AGN 7 GGTCAAGGTCAN 7 GTCGACAAGCTTCTAGAGCA-3Ј was used as template. This experiment yielded the consensus sequence NNTCAAGGTCAAGGTCA and confirmed that TCA in positions Ϫ9 to Ϫ7 is necessary for optimal binding of RTR.
Based on the sequence analysis above, we synthesized the consensus RTR response element GGAATCAAGGTCAAGGT-CATCTAAT (conRTRE) ( Fig. 2A) and analyzed its binding to RTR in EMSAs. As shown in Fig. 2B, in vitro translated RTR formed a complex with conRTRE, while increasing concentrations of the unlabeled conRTRE competed effectively for the binding. Since conRTRE consists of two overlapping TCAAG-GTCA motifs, we next determined whether RTR was able to bind to a single core motif (M3-RTRE) consisting of the conR-TRE in which the upstream half-site was destroyed by mutating three different nucleotides ( Fig. 2A). Although this single core motif RTRE did compete with 32 P-conRTRE for RTR binding, it was not as effective as the conRTRE (Fig. 2B). These observations suggest that RTR can bind well to an RTRE containing one core motif but does so with a lower affinity than to an RTRE consisting of a DR0. A DR1-RTRE, identical to con-RTRE except that the two core motifs are separated by one nucleotide and a P0-RTRE containing a palindromic repeat of the core motif ( Fig. 2A), also did not compete as well for binding as the conRTRE (Fig. 2B). It is likely that the binding of RTR to DR1-RTRE and P0-RTRE takes place through interaction with the single motif TCAAGGTCA present in these elements. The apparent lower affinity of M3-RTRE, DR1-RTRE, and P0-RTRE for RTR is in agreement with the results obtained in Fig.   1, demonstrating that a DR0 is most optimal for RTR binding. In addition, these results suggest that RTR very likely binds to DR0 as a homodimer and as a monomer to the single motif TCAAGGTCA.
The hypothesis that RTR binds as a homodimer to conRTRE was further investigated by analyzing the binding of the fulllength RTR and RTR 406 , a truncated form of RTR. EMSA with either RTR or RTR 406 yielded one single shifted band that migrated at different positions. We then performed EMSA using a mixture of RTR and RTR 406 . As can be observed in Fig. 3, EMSA analysis of the binding of RTR and RTR 406 to 32 P-conRTRE yielded three shifted bands that would be consistent with the formation of an RTR homodimer, an RTR-RTR 406 heterodimer, and an RTR 406 homodimer. When EMSA was carried out with 32 P-M3-RE, which binds RTR or RTR 406 as a monomer, only two shifted bands were observed. These findings support the interpretation that RTR binds to conRTRE as a homodimer.
Mutational Analysis of RTRE-As shown in Fig. 2B, RTR is able to bind to a single core motif (1/2-RTRE). To define more precisely the requirements for high affinity RTR binding, the effect of several mutations in the 5Ј-flanking sequence of 1/2-RTRE on the ability of these response elements to compete with 32 P-1/2-RTRE for RTR binding was examined (Fig. 4A). As shown in Fig. 4B, when one or both of the adenosines at the Ϫ6 and Ϫ7 positions (H-M2, H-M3, and H-M4) were replaced by cytidine, the ability of the oligonucleotides to compete with 32 P-1/2-RTRE for RTR binding was greatly diminished (lanes 9 -11, 12-14, and 15-17). A single mutation of the C into A at position Ϫ8 (H-M8) also reduced dramatically the ability to compete for RTR binding (Fig. 4C, lanes 31-33). A single mutation of the T into C (H-M5) or G (H-M7) at position Ϫ9 slightly reduced the ability to compete with 32 P-1/2-RTR ( Fig.   FIG. 1. Characterization of the RTRE consensus sequence. To determine the consensus sequence of the response element that binds RTR with high affinity (RTRE), a DNA binding site selection strategy was followed that was based on a combination of PCR and EMSA as described under "Experimental Procedures." A mixture of 70-bp oligonucleotides containing 30 random bases flanked by two 20-bp primers (primers 1 and 2) was used as template. A, sequence of the 30-nucleotide-long fragments that bind RTR with high affinity identified in 37 independent clones. Boldface letters indicate the sequence of RTRE.  28 -30, respectively). The latter is in agreement with the lower requirement for a specific nucleotide at this position as shown in Fig. 1, although T appears to be the preferred nucleotide. Mutations at the positions Ϫ10, Ϫ11, and ϩ1 (H-M1 and H-M6, respectively) did not significantly alter the affinity for RTR (lanes 6 -8 and 21-23). Several mutations (H-M10 through H-M13) within the core motif of the 1/2-RTRE had a dramatic effect on the binding of RTR (Fig. 4D,  lanes 41-52). The oligonucleotide H-M9, which contains T instead of G at position Ϫ4, competed equally well for binding as the 1/2-RTRE in agreement with obtained consensus sequence (Fig. 1).
Supershift with RTR-specific Antiserum-An antibody, RTR-Ab2, was raised against a specific RTR peptide and tested for its ability to bind RTR. This antiserum, but not the preimmune serum, could immunoprecipitate in vitro translated, 35 S-labeled RTR, which migrated as a protein with a molecular mass of about 58 kDa (Fig. 5A, lanes 2 and 3). The addition of the immunizing peptide blocked the immunoprecipitation of RTR (Fig. 5A, lanes 4 and 5). These results demonstrate that this antiserum is able to recognize RTR protein specifically. The RTR-Ab2 antiserum was then used in a supershift assay. As shown in Fig. 5B, the RTR-RTRE complex migrates more slowly when the reaction is carried out in the presence of RTR-Ab2 (lane 2), whereas no change in migration was observed in the presence of preimmune serum (lane 3). The presence of increasing amounts of the immunizing peptide inhibited the supershift of the RTR-RTRE complex (lanes 4 and 5).
Identification of an RTRE in the Promoter Regulatory Region of the Protamine 2 Gene-Previous observations demonstrated that RTR mRNA is highly expressed in testis and was associated with a specific stage of germ cells differentiation (27,28). The highest level of GCNF/RTR mRNA was detected in late round spermatids (27). 2 The association of GCNF/RTR mRNA expression with late round spermatids suggests that GCNF/ RTR regulates gene expression during maturation of round spermatids to elongated spermatids. This stage of differentiation is associated with many changes in gene expression (29,36). We, therefore, examined the 5Ј-flanking promoter regions of several genes, including protamine 2, reported to be induced at this stage of differentiation for the presence of RTREs. Based on the sequence of the conRTRE, we identified a putative RTRE in the regulatory region of the mouse protamine 2 gene between nucleotides Ϫ213 and Ϫ228 (30) (Fig. 6A). This RTRE, referred to as P2-RE, consists of a DR0 and is almost identical to the conRTRE (Fig. 6A). Northern analysis using poly(A) ϩ RNA isolated from mouse round spermatids confirmed that the RTR and protamine 2 genes are highly expressed at this stage of spermatogenesis (not shown).
The putative P2-RE was synthesized and analyzed for RTR binding in EMSA. As shown in Fig. 6B, P2-RE could compete with 32 P-conRTRE for RTR binding almost as well as the unlabeled conRTRE itself. We then performed EMSA with 32 P-P2-RE in the absence or presence of a 5-, 25-, and 100-fold excess of unlabeled P2-RE or P2-mRE, which is mutated at two positions (Fig. 6A). These experiments demonstrated that 32 P-P2-RE forms a complex with RTR (Fig. 6C, lane 2) and that unlabeled P2-RE competed effectively with this binding (lanes 3-5), while P2-mRE had a greatly reduced ability to compete with 32 P-P2-RE for RTR binding (Fig. 6C, lanes 6 -8). These results show that the P2-RE present in the promoter regulatory region of the protamine 2 gene can function as a high affinity binding site for RTR and as such could play a role in the regulation of the expression of this gene.
Specific RTRE Binding of RTR in Testis Nuclear Extracts-Since RTR is highly expressed in testis, we examined whether proteins in nuclear extracts from testis were able to bind to the P2-RE. When testis nuclear extracts were incubated with 32 P-P2-RE and analyzed by EMSA, multiple DNA-protein complexes were observed (Fig. 7). This is not surprising, since a number of other nuclear receptors, including SF-1/FTZ-F1 (24 -26), are expressed in the testis and able to bind to the same response element. For example, SF-1/FTZ-F1 has been reported to bind TCAAGGTCA (24), which is contained in the P2-RE. We used the RTR-Ab2 antiserum to determine if any of these DNA-protein complexes contained the RTR receptor and were supershifted by the antiserum. As shown in Fig. 7A, the RTR-Ab2 caused a supershift of one of the DNA-protein complexes (lane 2). To determine the specificity of the interaction of nuclear proteins with 32 P-P2-RE, we analyzed nuclear extracts from mouse liver, kidney, and brain that do not express detectable levels of RTR (27,28) in a supershift assay. In contrast to nuclear extracts from testis, no supershift was observed with nuclear extracts from brain (lanes 3 and 4), kidney (lanes 5 and 6), and liver (lanes 7 and 8) even at longer exposures (not shown). These findings show that the presence of RTR demonstrated by the RTR-Ab2-induced supershift in EMSA correlates well with the tissue-specific expression pattern of RTR.
The expression of RTR is developmentally regulated. 3 RTR is induced when germ cells differentiate into spermatids, a process that in mice is initiated at day 22 after birth. We therefore compared the binding of proteins of nuclear extracts isolated from testes of juvenile (17-day-old) and adult (3-month-old) mice in a supershift assay (Fig. 7B). In contrast to the nuclear extracts from testes of adult mice, no supershift was observed with RTR-Ab2 using nuclear extracts from testes of juvenile mice. These results are in agreement with the observed development-dependent expression of RTR.
The specificity of the supershift was further examined with preimmune serum and competition with unlabeled P2-RE (  Transcriptional Activation through RTRE and RTR-To examine the transactivation activity of RTR, an RTR expression plasmid and (conRTRE) 3 -CAT reporter DNA were cotransfected into CV-1 cells and assayed for CAT activity. No increase in CAT reporter gene activity could be measured compared with cells transfected with (conRTRE) 3 -CAT reporter DNA alone (Fig. 8A). One of the reasons for the lack of transcriptional activity may be the absence of the proper ligand required for the activation of RTR. To study the interaction of RTR with RTRE in cells, we examined the ability of VP16-RTR, which encodes a fusion protein consisting of the activating domain of VP16 and the full coding region of RTR, to cause transactivation of CAT through (conRTRE) 3 . As shown in Fig. 8A, pSG5-VP16-RTR enhanced CAT activity about 5-fold, while no activation was observed in cells cotransfected with pSG5-VP16. We then analyzed the transactivation through several other response elements. Transactivation through (P2-RE) 3 was just as effective as that through (conRTRE) 3 , while transactivation through (1/2-RTRE) 3 and (conRTRE) 1 was greatly diminished (Fig. 8B). The latter may be due to the possibility that their orientation is not the most optimal for promoting interactions with the protein TFIID complex at the TATA-box. These observations support the hypothesis that binding of RTR to RTREs, including conRTRE and P2-RE, can mediate changes in transcriptional activation of target genes.  (lanes 3, 4, and 5, etc., respectively). Lane 1, unprogrammed lysate was used. Lane 2, EMSA without competitors. the core motif RGGTCA (28). 4 In this study, we characterized in an unbiased manner the consensus sequence of the RTRE by an EMSA/PCR-based strategy that selects for oligonucleotides with highest affinity for RTR from a pool of degenerate oligonucleotides (primer 1-N 30 -primer 2). The conRTRE, TCAAGG-(T/T)CAAG(G/T)TCA, calculated from 37 independent sequences revealed the following. 1) RTR binds with highest affinity to a direct repeat containing the core motif AG(G/ T)TCA. 2) There is a strong requirement for specific nucleotides in the 5Ј-flanking region for optimal RTR binding, while no strong requirements were revealed for the 3Ј-flanking region.
3) A single core motif, AGGTCA, preceded by TCA (defined as 1/2-RTRE) is sufficient for RTR binding; however, the apparent affinity of RTR for 1/2-RTRE is significantly lower than that for conRTRE. It is likely that the moderate competition for RTR binding displayed by DR1-RTRE and P0-RTRE occurs through the 1/2-RTRE present in these response elements. Mutational analysis of the 5Ј-flanking region of 1/2-RTRE supported the importance of A nucleotides at the Ϫ6 and Ϫ7 positions, a C at position Ϫ8, and to a lesser degree T at the Ϫ9 position for optimal binding of RTR and confirms that the sequence TCAAGGTCA is necessary for optimal binding of RTR. The observed differences in the affinity of RTR to conRTRE and 1/2-RTRE are consistent with the hypothesis that RTR binds as a homodimer to conRTRE and as a monomer to 1/2-RTRE. EMSA analysis using the full-length RTR and a truncated form of RTR supports this concept. As expected for dimer formation, this EMSA with 32 P-conRTRE yielded three shifted bands representing the homodimers (RTR) 2 and (RTR 406 ) 2 and the heterodimer RTR-RTR 406 . These results support the preliminary observations on the GCNF/RTR homodimerization (28).
Several other members of the steroid hormone superfamily have been reported to bind as monomers to single core motifs (21, 24 -26). Members of the ROR/RZR family have been shown to bind to a single core motif flanked at the 5Ј-end by a 6-bplong A/T-rich sequence (21,23). The sequence of the consensus half-site RTRE is identical to that reported for the SF-1/FTZ-F1 receptor (24 -26). Interestingly, comparison of the sequences in the T-and A-box of RTR with those of SF-1/FTZ (FTZ-F1 box) shows the presence of a highly homologous region in which 8 of 11 amino acids are identical (26,27). Since the FTZ-F1 box has been implicated in the recognition of the SF-1/FTZ response element (26), this similarity is consistent with observations showing that RTR and SF-1 bind to very similar response elements.
To determine the ability of RTR to transactivate a CAT reporter gene that is under the regulation of an RTRE, CV-1 cells were transfected with an RTR expression plasmid and (RTRE) 3 -CAT. No transactivation of the CAT reporter gene was observed. However, the fusion protein RTR-VP16, consisting of the full-length RTR fused to the 80-amino acid-long activating domain of VP16 was able to cause RTRE-dependent transactivation of CAT. These results demonstrate that the fusion protein was able to bind to the RTRE through RTR and to cause transactivation of the reporter via the activation domain of VP16. The lack of transactivation by RTR could be attributed to several factors. First, the activation of RTR may require the presence of its ligand, which has not yet been identified. Alternatively, the presence of a ligand in the serum could repress the transactivation activity of RTR. We have found that neither the addition of lipid extracts isolated from testes nor the presence of delipidized serum had any effect on the ability of RTR to cause RTRE-dependent transactivation. through interactions with other nuclear proteins that function either as repressors or as co-activators (18,37). Some of these interactions may exhibit a high specificity. It is therefore possible that RTR needs a co-activator that is highly testis-specific and absent in CV-1 cells. Another possibility is that RTR functions as a negative transcriptional regulator.
Previous studies reported that the orphan receptor GCNF/ RTR is predominantly expressed in the testis and induced during a specific stage of spermatogenesis (27,28). The highest level of GCNF/RTR mRNA was detected in late round spermatids (27), 2 suggesting that GCNF/RTR regulates gene expression during maturation of round spermatids to elongated spermatids. This stage of differentiation is accompanied by an induction in the expression of many genes, including protamine 1 and 2, c-abl, transition nuclear proteins 1 and 2, c-pim-1, hsp-70.1 and histone 2b (29,36). To identify candidate target genes that are regulated by RTR, the reported promoter regions of several of these genes were examined for the presence of RTREs. The 5Ј promoter-flanking region of the mouse protamine 2 gene (30) was found to contain the sequence AGAAGT-TCAAGGTCAT, referred to as P2-RE, between nucleotides Ϫ213 and Ϫ228. This sequence is highly homologous to the conRTRE. We demonstrate by EMSA analysis that RTR is able to bind to this element almost as effectively as to the conRTRE. EMSA analysis of nuclear extracts isolated from testis, brain, liver, and kidney demonstrated that the RTR-specific antiserum was able to cause supershift of an RTR-DNA complex only with extracts from testis. We have demonstrated 3 that the expression of RTR is initiated at day 17 after birth at a time that mouse germ cells differentiate into spermatocytes and that it is highly expressed at day 22 when spermatids appear. The RTR-specific antiserum does not cause a supershift with nuclear extracts from testes of juvenile (17-day-old) mice in contrast to nuclear extracts from testes of adult (3-month-old) mice. These findings indicate that the observed supershift by the RTR-specific antibody correlates well with the tissue-specific and development-dependent expression of RTR. Comparison of this sequence between the mouse and rat protamine 2 gene shows that the mouse gene contains two core motifs (P2-RE) but that the rat gene contains only a single core motif. It is interesting to note that protamine 2 levels in rat testis have been reported to be only 5% of that in mouse (38). One could speculate that this may be related at least in part to the differences in the affinity of conRTRE and 1/2-RTRE for RTR.
In summary, in this study we provide evidence consistent with the conclusion that the orphan receptor GCNF/RTR is able to bind as a homodimer to a DR0 and as a monomer to a 1/2-RTRE. The sequence preceding the core motif is important for determining the affinity of GCNF/RTR to RTREs and consists of the consensus TCA. GCNF/RTR is unable to cause transactivation of a reporter gene through the conRTRE in CV-1 cells. This may be due to either the lack of its ligand or a proper co-activator. Alternatively, GCNF/RTR may function as a negative regulator of transcription. An RTRE was identified in the 5Ј promoter-flanking region of the protamine 2 gene that is induced during the same stage of spermatogenesis as GCNF/ RTR. This gene may belong to a subset of genes that are regulated by GCNF/RTR. Studies are in progress to obtain further evidence in support of this hypothesis.