Constitutive activation of retinoic acid receptor beta2 promoter by orphan nuclear receptor TR2.

The orphan nuclear receptor TR2 functions as a constitutive activator for the endogenous retinoic acid receptor beta2 (RAR(beta2)) gene expression in P19 embryonal carcinoma cells and for reporters driven by the RAR(beta2) promoter in COS-1 cells. The activation of RAR(beta2) by TR2 is mediated by the direct repeat-5 (DR5) element located in the RAR(beta2) promoter. Furthermore, cAMP exerts an enhancing effect on the activation of RAR(beta2) by TR2, which is mediated by the cAMP response element located in the 5'-flanking region of the DR5. The constitutive activation function-1 (AF-1) of TR2 is mapped to amino acid residues 10-30 in its N-terminal A segment. A direct molecular interaction occurs between CREMtau and TR2, detected by co-immunoprecipitation, which is mediated by the N-terminal AB segment of TR2. In gel mobility shift assays, TR2 competes with P19 nuclear factor binding to the RAR(beta2) promoter, and TR2 and CREMtau bind simultaneously to this DNA fragment. The role of TR2 in the early events of RA signaling process is discussed.

Nuclear receptors comprise a super family of transcription factors that contain a zinc finger-type DNA binding domain (DBD) 1 and a ligand binding (LBD) domain and are able to regulate gene expression in a wide variety of biological processes (1)(2)(3)(4)(5)(6). Some of these nuclear receptors are known hormone receptors, whereas the vast majority of the cloned nuclear receptors remain as orphan members. Despite the lack of identified ligands, the physiological roles of several orphan receptors have begun to be revealed in genetic studies (7)(8)(9)(10).
The mouse orphan receptor TR2 was isolated from an E8.5 embryonic cDNA library (11), and the gene was characterized (12). The human TR2 was cloned from a prostate cDNA library (13). Like many other orphan receptors, the biological activity of TR2 was first demonstrated in several reporter systems. For instance, TR2 repressed reporters driven by a direct repeat (DR)-4 hormone response element derived from the mouse cel-lular retinoic acid-binding protein I gene promoter (14) and a DR1-type RA response element derived from the cellular retinol-binding protein II gene promoter (15). Moreover, TR2 strongly suppressed RA induction of a reporter driven by the DR5 derived from RA receptor ␤2 (RAR ␤2 ) promoter, mediated by competitive binding of TR2 to this DR5 (16 -18). By using this DR5 reporter as a model system, the functional characteristics of TR2 were examined, and its molecular features required for a suppressive activity were revealed. It was demonstrated that the suppressive activity required the DBD, the ability to dimerize, the LBD, and two consecutive glutamate residues at amino acid (aa) positions 553/554 (16). In addition, a novel receptor heterodimeric pathway was identified that involved heterodimers of TR2 and TR4 (19). Recently, the mouse nuclear receptor-interacting protein 140 (RIP140) was cloned and demonstrated as a co-repressor for TR2 by interacting with its LBD (20).
In all these functional studies, TR2 appears to affect RA signaling pathways by regulating the components for RA metabolism (cellular retinoic acid-binding protein-I and cellular retinol-binding protein-II) and modulating RA induction of target gene expression. Of most significance is the potent suppression of RA induction of DR5-type RA response element derived from the RAR ␤2 gene promoter. RAR ␤2 is known as one of the earliest RA-responding genes in several culture systems, most notably the embryonal carcinoma cell cultures such as P19 and F9 (21)(22)(23)(24). This gene serves as one of the master regulators in many RA-induced cellular events, such as proliferation, differentiation, and apoptosis, by regulating a number of downstream effector genes (25,26). The RAR ␤2 gene is weakly expressed in stem cell populations and is rapidly induced by RA (27). A functional promoter of this gene consists of several essential DNA response elements arranged in close proximity. These include, among others, an RA-responding DR5, a cyclic AMP-response element (CRE), and a TATA box (28). The DR5 is responsible for a rapid and potent RA induction, mediated by holo-RAR/RXR binding (21,23,29). The biological effects of cAMP has also been confirmed in the P19 culture model (28).
Although RA induction appears to be the most effective trigger that induces this gene expression in the stem cell cultures, the regulation of this gene in animals appears to be rather complicated. For instance, the expression pattern of this gene in animal tissues does not always correspond to the panel of tissues that are rich in RA (22, 30 -32). Furthermore, this promoter cannot be activated by RA in a number of cell types such as breast tumor and lung cancer cell lines as well as some pituitary cell lines (33)(34)(35) despite the detection of a potent RA induction in these cells using reporters containing the dissected DR5 (34,36). All these observations suggest that the induction of the RAR ␤2 gene may involve factors other than RA.
Like the studies of several other orphan receptors COUP-TF, nerve growth factor-1B, Dax1, and hepatocyte nulcear factor-4 (37)(38)(39)(40), previous studies of TR2 in different labs have utilized RA response element-and other hormone response elementcontaining reporters to examine its biological activities (5,24,25,28,29). The finding that apo-TR2 is able to bind to this DR5 with a high affinity (an estimated K d 7.4 nM) in the absence of putative ligands (12,16,17) has prompted us to examine the effect of TR2 expression on the endogenous RAR ␤2 gene activity in the absence of RA. In this study, it is demonstrated that apo-TR2 functions as a constitutive activator for reporters driven by the RAR ␤2 promoter containing the DR5 element, and overexpression of TR2 in P19 stem cells activates the endogenous RAR ␤2 gene expression. Furthermore, cAMP exerts an enhancing effect on the activating function of TR2 on this promoter, which contains a CRE and a DR5. The activating function of TR2 is mapped to its N-terminal segment, corresponding to the A domain, and the AB segment is involved in a direct intermolecular interaction of TR2 with an activator form of cAMP-response element-binding protein (CREB), CREM (41). Finally, in gel mobility shift assays, TR2 competes with P19 nuclear factors binding to the RAR ␤2 promoter, and TR2 and CREM bind simultaneously to this DNA fragment. We now report these studies characterizing TR2 as a constitutive activator for the RAR ␤2 promoter, which can be enhanced by cAMP.

Construction of Reporters and Expression Vectors-
The RAR ␤2 reporters were constructed by fusing genomic sequences generated by polymerase chain reactions (PCRs) (for CRE/DR5/TATA-luc and ϳCRE/ DR5/TATA-Luc), SmaI digestion (for DR5/TATA-luc), or re-annealing of oligonucleotides (for TATA-luc) upstream to a promoterless luciferase cDNA, pGL3 (Promega, Madison, WI). The cDNA for mouse CREM (42) was also obtained in a reverse transcription-coupled PCR (RT-PCR) using mouse testis mRNA as the template and confirmed by DNA sequencing. The cDNA of CREM was cloned into the pSG5 vector at BglII site for expression in mammalian cells and for in vitro transcription/translation reactions (for gel mobility shift assay). The expression vectors for TR2 and deletions and point mutations, cloned in pSG5 vectors, were described previously (16). The dissected TR2 segments to be cloned into mammalian two-hybrid expression vectors (see "Mammalian Two-hybrid and Transactivation Assays" under "Experimental Procedures") were obtained by either restriction digestion of TR2 cDNA or PCR. The fragment containing the AB domain and a small portion of the zinc finger (aa 1-138) was obtained as a 0.4-kilobase fragment by EcoRI/PstI digestion at the N terminus of TR2 cDNA, the A segment (aa 1-50) was obtained as a 0.15-kilobase fragment by EcoRI/BamHI digestion, and the B segment (aa 51138) was obtained by BamHI/PstI digestion in a size of 0.25 kilobases. Further deletions of the A domain (A-1/40, 1/30, 10/30, and 10/25) were obtained in PCRs. Table I summarizes the oligonucleotides used in PCRs to generate specific DNA fragments for this study. All PCR-generated DNA fragments have been confirmed by DNA sequencing.
Cell Culture and Transfection Techniques-The P19 cell line, main-tained as described previously (43), was used to determine the effects of expressing TR2 on the endogenous RAR ␤2 gene activity. COS-1 cells were used in co-transfection experiments to determine the effects of added nuclear receptors on the reporter gene activity as well as mammalian two-hybrid interaction experiments as described (16,20). All the COS-1 cultures were maintained in Dulbecco's modified Eagle's medium containing dextran charcoal-treated serum (DCC medium). Transfection was conducted by using the calcium phosphate precipitation method, and LacZ and luciferase activities were determined as described (16). The luciferase activity was normalized to the LacZ activity of the internal control to obtain specific luciferase unit. To compare the relative activity of different expression vectors on the reporter, the specific luciferase unit of the control vector was assigned an arbitrary value of 1 in order to determine the relative luciferase unit of each expression vector, which represented the relative activity of the expression vector. Triplicate cultures were used in each transfection experiment, and three independent experiments were conducted to obtain the means and S.E. for all the transfection experiments.
RT-PCR to Detect Endogenous RAR ␤2 Expression-RT-PCR was conducted to detect the expression of endogenous gene expression in P19 cultures. Primers for RAR ␤2 are 5Ј-TGGACTTTTCTGTGCGGC-3Ј (nucleotide position 391-408 in Fig. 2 of Ref. 24), where the 5Ј-untranslated region of RAR ␤2 is located, and 5Ј-GGGAATGTCTGCAACAGCT-GGA-3Ј from the 3Ј-untranslated region (24). A fragment of approximately 1.6 kilobases of RAR ␤2 was amplified by using this pair of primers. P19 cells seeded at a density of 5 ϫ 10 5 /10-cm plate were used for each treatment. Cells from one 10-cm plate were harvested for RNA isolation and resuspended in a final volume of 20 l. Two l total RNA prepared from each P19 culture was reverse-transcribed using oligo(dT) as the primer in a total volume of 20 l. Two l of each RT reaction was used in PCR in a total volume of 25 l and primed with two sets of oligos, one specific to actin (11) and the second specific to RAR ␤2 . The PCR reaction cycle was 94°C for 45 s, 55°C for 45 s, and 72°C for 1 min, for a total of 30 cycles. Five l of each PCR product was analyzed on Southern blots and probed with actin-and RAR ␤2 -specific probes.
Mammalian Two-hybrid Interaction and Transactivation Assays-For the mammalian two-hybrid interaction assay, pM (containing the DBD of GAL4) and pVP (containing the activation domain of VP16) (CLONTECH, Palo Alto, CA) were used to construct the expression vectors. Pairs of pM and pVP fusions were tested in parallel. To examine intermolecular interactions, the cDNA for CREM was cloned into pM and tested against TR2 fragments cloned in pVP. The reporter construct for the mammalian two-hybrid interaction as well as transfection procedures were as described previously (20).
To detect the intrinsic transactivator function of TR2, various TR2 segments were fused to pM vector. A GAL4 binding site-driven luciferase reporter and the internal control lacZ were as described (20). Activation of reporter by GAL4 fusions was determined by comparing their specific luciferase units to that of the control.
Co-immunoprecipitation Assay-The anti-TR2 antibody generated previously (12,17) was not effective for immunoprecipitation; therefore, we utilized a hemagglutinin (HA)-tagged TR2 expression vector (20) in immunoprecipitation experiments. The biological activity (suppression of RA induction on the DR5-containing reporter) and biochemical property (binding to DR5 element) of HA-TR2 expressed from this vector a Sequences for the initiating codon are underlined.

TR2 as a Constitutive Activator for RAR␤2
were demonstrated previously (20). The activation function of HA-TR2 on RAR ␤2 promoter was confirmed in this study (see "Results"). This vector was used to transfect COS-1 cells in co-immunoprecipitation experiments involving TR2. COS-1 cells were transfected with the HA-TR2 in the presence or absence of the CREM expression vector. Cells from one 10-cm plate for each treatment were harvested at 36 -48 h and resuspended in 200 l of lysis buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 1 mM dithiothreitol, 2 M phenylmethylsulfonyl fluoride, 10% glycerol, protease inhibitor mixture). For immunoprecipitation, 100 l of the cell lysate was incubated with a mouse anti-HA monoclonal antibody (Roche Molecular Biochemicals) at 4°C for 2 h, followed by the addition of 15 l of protein G-Sepharose CL4B resin (Sigma). The precipitation was conducted overnight at 4°C, and the resin was vigorously washed three times with the lysis buffer and resuspended in a SDS-polyacrylamide gel electrophoresis loading buffer for Western blot analysis.
A 10% polyacrylamide gel was used for protein separation, and the gel was transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA). The blot was incubated with a rabbit anti-CREM antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C overnight, followed by washing and reaction with a mouse anti-rabbit secondary antibody and detection with ECL (Amersham Pharmacia Biotech). To monitor TR2 precipitated in the reactions, the signals on the blot were stripped off by washing in a solution of 2% SDS, 100 mM 2␤-mercaptoethanol, 62.5 mM Tris, pH 6.7 at 50°C for 30 min. The blot was then reacted with the rabbit anti-TR2 antibody (12,17) and detected with ECL.
Electrophoretic Mobility Shift Assay-The mobility shift assay was conducted as described previously (20). Nuclear extract of P19 was isolated as described (44). The extract from a 10-cm dish was resuspended in a volume of 60 l, and 4 l (about 10 g of nuclear protein) was used in each gel shift reaction. Protein was also prepared using in vitro transcription and translation reactions (TNT, Promega). For protein-DNA interactions, the in vitro translated protein or nuclear extract isolated from P19 cells was incubated with 1 ng of probe in a 20 l of binding buffer. The probes were prepared by labeling the doublestranded DNA fragments isolated from the genomic segments described in a previous section (CRE/DR5/TATA and DR5/TATA) with [ 32 P]dCTP using Klenow enzyme. A fragment containing only the DR5 (16) was also prepared by annealing oligonucleotides spanning DR5 site and used as a control for TR2 binding.

RESULTS
Constitutive Activation of RAR ␤2 by TR2 Expression-Previously, the biological activity of orphan receptor TR2 was demonstrated to be primarily suppressive on both RA induction of DR5-containing reporters (16,17) and reporters regulated by some hormone response elements such as DR4 (14) and DR1 (15). It was demonstrated that the suppressive activity required the intact LBD and DBD but not the N-terminal A/B domains of the molecule (16). In addition, the suppressive effects of TR2 on RA induction of DR5-driven reporter was attributed to its competitive binding to DR5 (16,18). In an attempt to determine the constitutive activity of TR2 on DR5 reporters, i.e. its biological activity in the absence of RA, we set up experiments to examine the effects of TR2 expression on RAR ␤2 gene activities using both reporter and endogenous gene expression systems.
To examine the effects of TR2 expression on the endogenous RAR ␤2 gene expression, we used P19 stem cells, which expressed a negligible level of TR2 and RAR ␤2 (18), as an experimental system. P19 cells maintained in medium containing charcoal-treated serum were induced with RA or transfected with the wild type TR2 expression vector. RNA was isolated 12 or 24 h after the addition of RA or transfection. To detect RAR ␤2 mRNA specifically, RT-PCR was conducted to examine its mRNA level. As shown in Fig. 1 To further demonstrate the biological activities of TR2 on RAR ␤2 promoter, we constructed a luciferase reporter driven by a contiguous regulatory region of this promoter, a 91-base pair genomic segment containing the CRE followed by the DR5 element and TATA box (Ϫ100 to Ϫ10, Ref. 24). This reporter was designated as CRE/DR5/TATA-luc. The effects of TR2 expression on this reporter was assessed in transfected COS-1 cells supplemented with charcoal-depleted serum. As shown in Fig. 2, TR2 expression activates this reporter activity in a dose-dependent manner (filled columns). For a control, a mutant deleted in the most N-terminal A segment (TR2-⌬A) has been included and shown to be inactive in this assay (open columns). To confirm the activation function of HA-TR2, which was to be used in co-immunoprecipitation experiments, this expression vector was also examined in parallel experiments (striped columns). As shown in the same figure, TR2 tagged with an HA epitope remains as effective as the wild type protein in terms of the activation function in this reporter system.
Collectively, these data indicate that TR2 encodes a constitutive activation function for the endogenous RAR ␤2 gene as well as the reporter driven by its promoter in the absence of putative ligands. Furthermore, the activation of RAR ␤2 by TR2 is enhanced by cAMP.
Domains of TR2 Required for Its Activation Function-To determine the molecular domains required for this novel activation function of TR2, a panel of TR2 deletion mutants generated previously (16) that have been shown to express well and localize properly were first used to determine the required molecular features of TR2 as an activator for the CRE/DR5/ TATA reporter. As shown in Fig. 3, deletions from either the N-(TR2-⌬A) or the C terminus (Ϫ20, Ϫ50, Ϫ100, and Ϫ200) completely abolishes this activation function (columns 3-7), indicating that the activation function of TR2 requires an intact molecule. In addition, point mutations that abolish dimerization

TR2 as a Constitutive Activator for RAR␤2
and DNA binding (LLL mutant, column 9) or affect the putative AF-2 domain conformation (EE mutant, column 8) are also defective in this activation function. Therefore, the activation function of TR2 on RAR ␤2 promoter requires an intact receptor including the N-terminal domain, the DBD, and the LBD.
Defining the AF-1 Transactivation Domain-To define the activation domain of TR2 and to examine whether TR2 encoded an intrinsic, transactivation function, the entire TR2 as well as its dissected portions was each fused to pM for the test of a transactivation activity as described under "Experimental Procedures." The intact TR2 (TR2-f), its LBD (aa 166 -590), the putative AF-2 (aa 570 -590), the AB domain with a small portion of the zinc finger (aa 1-138), and the dissected A domain (aa 1-50) or B domain (aa 51-138) were first individually fused to the same pM vector and assessed in transactivation tests. As shown in Fig. 4, neither the intact TR2, the LBD, the putative AF-2, the AB, or the B domain is able to transactivate the GAL4 DBD (columns 1-5 and 11). Interestingly, the fusion of A domain alone to the GAL4 DBD dramatically induces GAL4 reporter (column 6), indicating that the dissected A domain encodes an activation function that can be transferred to a heterologous molecule such as the DBD of GAL4. However, this activation function may be masked in the context of the AB segment, since AB segment does not transactivate the reporter (column 4). To further dissect the minimal sequence required for such a transactivation function, more deletions were made in the A domain. As shown in the same Fig. 4 (columns 7-10), the A domain that retains aa 1/40, 1/30, or 10/30 (columns 7-9) remains active in this assay, whereas a further 5-aa deletion from the smallest pM-10/30 fragment (leaving only aa 10 to 25, pM-10/25, column 10) abolishes this activity completely. Therefore, the transferable activation function (AF-1) of TR2, as demonstrated in a GAL4 fusion, requires only a small segment (aa 10 to 30) of the N-terminal A domain.
The Effects of cAMP-Studies described in Fig. 1 show that TR2 constitutively activates the RAR ␤2 promoter activity, which can be enhanced by cAMP. In the RAR ␤2 promoter, a CRE is located approximately 40 base pairs upstream of the DR5 element, and cAMP is known to enhance RA induction of this promoter (28). To examine the effect of cAMP on the activation function of TR2, we generated three promoter deletions that retained only the DR5/TATA or TATA sequences or was deleted only at the CRE element, designated as DR5/ TATA-Luc, TATA-Luc, and ϳCRE/DR5/TATA-Luc, respectively (Fig. 5A). The effects of TR2 and cAMP on these promoter deletions were then compared with the wild type promoter (CRE/DR5/TATA-Luc). As shown in Fig. 5B, the expression of TR2 activates reporters CRE/DR5/TATA (columns 2 and 4), ϳCRE/DR5/TATA-Luc (columns 6 and 8), and DR5/TATA (columns 10 and 12) but not TATA-Luc (columns 14 and 16). The addition of cAMP enhances the activation by TR2 only for the CRE/DR5/TATA promoter (column 4). Interestingly, without TR2 expression, cAMP has no effect on any of these promoter activities (columns 3, 7, 11, and 15). Therefore, the RAR ␤2 promoter is activated by TR2 (through DR5 sequence) in the absence of RA, which can be enhanced by cAMP (through the CRE sequence). However, cAMP alone has no effect, consistent with the result shown in Fig. 1 that cAMP alone does not affect the endogenous RAR ␤2 expression.
Interaction of TR2 and CREM-Studies described in previous sections have demonstrated that cAMP enhances TR2 activation of RAR ␤2 promoter activity, which is mediated by their cognate DNA elements, CRE and DR5. The CREB protein family is expressed in a wide variety of cell types, but CREM is more specific to the testis (45). In particular, the activator CREM (named CREM) is restricted to post-meiotic germ cells (41,46) where TR2 is found most abundantly (12,17,18). It is interesting to test whether TR2 directly interacts with CREM or CREM-like protein, which may explain the enhancing effect of cAMP on the activation function of TR2. The potential interaction of TR2 with CREM, or CREM-like molecule was first tested in immunoprecipitation assays. As COS-1 cells did not express TR2, we then introduced TR2 into COS-1 cells by transfection. The HA-TR2 construct, which retained the biological activity (Fig. 2) in terms of activating RAR ␤2 promoter and DNA binding, was used in this study. A mouse monoclonal antibody against HA-epitope was used to immunoprecipitate HA-TR2 as described under "Experimental Procedures." The immunoprecipitate was resolved on a polyacrylamide electrophoresis gel and detected with a rabbit anti-CREM antibody that also reacted with CREM. As shown in Fig. 6A, CREM-or CREM-like protein was co-precipitated with TR2 (lane 2), and a significant increase in the precipitated CREM was detected in cells transfected with a combination of CREM and TR2 expression vectors (lane 1). In the control, where no TR2 expression vector was introduced, no CREM-like signal was detected. To examine TR2 expression in these cells, the signal A, four RAR ␤2 reporters were generated as described under "Experimental Procedures", and the relative positions of these regulatory elements in the RAR ␤2 promoter are indicated. B, the effects of TR2 and cAMP on the expression of four RAR ␤2 reporters. COS-1 cells were transfected with the vectors indicated (C for control, TR2 for the wild-type TR2) and treated with vehicle (Ϫ) or cAMP (ϩ). Specific luciferase units were determined at 24 h. By using the control activity (C) as 1 in each group, the relative activity (RUL) of each treatment in the same group was determined. The open bars show the group using CRE/DR5/TATA-luc as the reporter, the shaded bars show the group using ϳCRE/DR5/TATA-Luc as the reporter, the striped bars show the group using DR5/TATA-luc as the reporter, and the black bars show the group using TATA-luc as the reporter.
was stripped off the blot, which was subsequently detected with a rabbit anti-TR2 antibody as shown in the lower panel.
To determine which portion of TR2 molecule interacted with CREM, mammalian two-hybrid interaction tests were performed. The CREM-coding region was fused to the pM vector and tested against various portions of TR2 fused to the pVP vector. COS-1 cells were co-transfected with pairs of pM and pVP fusions together with the GAL4 reporter and an internal control lacZ vector. As shown in Fig. 6B, two pairs of control vectors induce only a background level of reporter activities (columns 1 and 2), and the cloned CREM interacts strongly with the AB segment (column 4) but not the LBD (column 5). Interestingly, deleting the B domain from the AB segment dramatically reduces this interaction (comparing columns 3 and 4), and the B domain alone fails to interact with CREM (column 6). Therefore, CREM is able to interact with the AB domain of TR2, and the B domain affects the ability of TR2 to interact efficiently with CREM.
DNA Binding Properties of TR2 and CREM on RAR ␤2 Promoter-Our previous studies demonstrated a specific binding of TR2 to the dissected DR5 of RAR ␤2 promoter, and the binding affinity was estimated to be approximately 7.4 nM to the dissected DR5 DNA fragment (16). Since the expression of TR2 activated the endogenous RAR ␤2 expression in P19 cells (Fig.  2), it was of interest to compare the binding of TR2 and P19 nuclear factors to this sequence in its genomic context. Gel shift experiments were conducted to examine P19 nuclear factors binding and TR2 binding patterns on the DR5-TATA fragment of RAR ␤2 promoter. As shown in Fig. 7A, consistent with our previous studies in which an isolated DR5 probe was used (16) . This result indicates that TR2 strongly competes with P19 endogenous nuclear factors in binding to this promoter. The finding that TR2 was able to interact with the cloned CREM and cAMP enhanced the activation of RAR ␤2 by TR2 prompted us to examine the DNA binding patterns of TR2 and CREM. A gel shift experiment was then conducted by using the contiguous CRE/DR5/TATA segment as the probes. As shown in Fig. 7C, TR2 alone binds to this fragment, shown as a major retarded band (lane 5, single arrow). Similarly, CREM alone also binds to this sequence, shown as a slightly faster migrating band (lane 3, small arrow head). In the presence of both TR2 and CREM, a super-shifted band appears (lane 2, double arrow), indicating that TR2 and CREM together are able to bind to this sequence at the same time. Interestingly, the addition of an anti-CREM antibody abolishes the super-shifted band but not the TR2 band (lane 4), indicating that this antibody alters the conformation of CREM, thereby affecting its interaction with DNA or TR2. Therefore, TR2 and CREM not only are able to interact directly with each other as demonstrated in two-hybrid interaction and immunoprecipitation assays but also can si- multaneously bind to the RAR ␤2 promoter DNA elements as demonstrated in the gel retardation assay. DISCUSSION This study demonstrates for the first time a constitutive activation function of orphan nuclear receptor TR2 on the endogenous RAR ␤2 gene expression in P19 cells as well as reporters driven by the promoter of this gene. The activation is mediated by the DR5 element, which can be enhanced by cAMP through the upstream CRE element. The activation domain of TR2 was mapped to aa 10 -30 in its N-terminal A segment. Intermolecular interaction occurs between TR2 and CREM, as demonstrated in two-hybrid interaction and co-immunoprecipitation assays. The molecular interaction of TR2 with CREM is mediated by the N-terminal AB segment. On the RAR ␤2 promoter, TR2 and CREM bind simultaneously to the DNA, and TR2 competes efficiently with P19 nuclear factor binding to this promoter.
The nature of TR2 as an activator exhibits two features. First of all, the activation is specific to RAR ␤2 promoter, since TR2 represses other promoters that also contain a response element for TR2 such as the cellular retinoic acid-binding protein-I or SV40 promoters (data not shown). Although cAMP is able to enhance the activation of RAR ␤2 by TR2, the intrinsic activity of TR2 cannot be attributed solely to the cAMP pathway, since the RAR ␤2 promoter deleted in the CRE can still be activated by TR2 but at a lower level. We have failed to detect any interaction of TR2 with other potential co-activators such as TBP or SRC-1 in either immunoprecipitation or two-hybrid interaction tests (data not shown). Therefore, the biochemical basis of this activation function of TR2 remains to be determined. Second, although the activation domain of TR2 is transferable as demonstrated in transactivation assay (Fig. 4), this activity appears silent in the context of its intact molecule (pM-TR2-f fusion) and becomes apparent only when the domain is dissected out (the pM-A fusion). Therefore, this activity is mostly masked in the context of intact receptor molecules and can be revealed by molecular interaction with specific promoter sequence (RAR ␤2 ) or conformational changes (separation from other domains of TR2 and fusion to the DBD of GAL4). It is possible that certain promoter-or cell type-specific cofactors induce a molecular change of TR2 and contribute to such a novel biological activity. We are currently investigating this possibility.
The signal of cAMP is important in many biological systems, mediated by a number of nuclear proteins that belong to the CREB family. Among this family, CREM is an activator and specific to post-meiotic germ cells in the testis (41), where TR2 is also most highly expressed (18). The effects of cAMP on germ cell development have long been documented (41,45,47). RA is also an essential component for testis development, particularly during germ cell maturation (48 -50), and RAR ␤2 is known as one of the earliest RA-responding genes. These observations suggest that the enhancement of TR2 activation function by cAMP and the interaction of TR2 with CREM are of physiological significance. TR2 may be involved in the initial cross-talk between cAMP and retinoid pathways.
Like many other orphan nuclear receptors, the activities of TR2 have been demonstrated to be primarily repressive in many systems without putative ligands (14,15,19). Many genes that are demonstrated to contain a DNA response element for TR2 are involved in RA metabolism, such as cellular retinoic acid-binding protein I and cellular retinol-binding protein II. In these cases, TR2 is shown to play a negative role. Furthermore, TR2 strongly suppresses RA induction of reporters driven by the DR5 derived from RAR ␤2 promoter, indicating that TR2 may be directly involved in fine-tuning RA signaling pathways, primarily in a negative fashion. The activation func-tion of TR2 in the absence of RA, as demonstrated for the RAR ␤2 promoter in this study, suggests a potentially positive role of TR2 in the early events of the RA-signaling process. In fact, the mouse TR2 cDNA was originally isolated from an E8.5 mouse embryo library (45). During embryonic development, the expression of TR2 starts even before the implantation occurs (45); this would support such a hypothesis. Furthermore, TR2 is able to activate RAR ␤2 in the absence of RA; it is likely that TR2 is one of the earliest triggers that activates the response machinery for RA signaling. However, once RA is generated and RAR ␤2 is highly induced, TR2 then plays a suppressive role. According to these studies and observations, it is tempting to speculate a tight regulatory loop for RA signaling processes that may be integrated with the orphan receptor TR2 system. Nevertheless, a physiological connection between these two pathways remains to be further established by genetic tests.