A Polymorphic Autoregulatory Hormone Response Element in the Human Estrogen-related Receptor α (ERRα) Promoter Dictates Peroxisome Proliferator-activated Receptor γ Coactivator-1α Control of ERRα Expression

The orphan nuclear estrogen-related receptor α (ERRα) and transcriptional cofactor peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α) are involved in the regulation of energy metabolism. Recently, extensive cross-talk between PGC-1α and ERRα has been demonstrated. The presence of PGC-1α is associated with an elevated expression of ERRα, and the two proteins can influence the transcriptional activities of one another. Using a candidate gene approach to detect regulatory variants within genes encoding nuclear receptors, we have identified a 23-bp sequence (ESRRA23) containing two nuclear receptor recognition half-site motifs that is present in 1–4 copies within the promoter of the human ESRRA gene encoding ERRα. The ESRRA23 sequence contains a functional ERR response element that is specifically bound by ERRα, and chromatin immunoprecipitation shows that endogenous ERRα occupies its own promoter in vivo. Strikingly, introduction of PGC-1α in HeLa cells by transient transfection induces the activity of the ESRRA promoter in a manner that is dependent on the presence of the ESRRA23 element and on its dosage. Coexpression of ERRα and PGC-1α results in a synergistic activation of the ESRRA promoter. In experiments using ERRα null fibroblasts, the ability of PGC-1α to stimulate the ESRRA promoter is considerably reduced but can be restored by addition of ERRα. Taken together, these results demonstrate that an interdependent ERRα/PGC-1α-based transcriptional pathway targets the ESRRA23 element to dictate the level of ERRα expression. This study further suggests that this regulatory polymorphism may provide differential responses to ERRα/PGC-1α-mediated metabolic cues in the human population.

Nuclear hormone receptors are transcription factors that control essential developmental and physiological pathways (1). Although the transcriptional activity of nuclear receptors is primarily regulated by specific ligands, several members of the superfamily of nuclear receptors have no known natural ligands and are therefore referred to as orphan receptors (2). Estrogen-related receptor ␣ (ERR␣ 1 ; NR3B1) was the first orphan nuclear receptor to be identified on the basis of its similarity with estrogen receptor ␣ (ER␣; NR3A1) (3). Phylogenic tree reconstruction confirmed that ERR␣ belongs to the subgroup of receptors for steroid hormones (4), and ERR␣ was subsequently shown to share both structural and functional attributes with the ERs including binding to synthetic estrogenic ligands (reviewed in Ref. 5). ERR␣ also recognizes estrogen response elements (EREs), but characterization of its DNA binding properties demonstrated a preference for sites composed of a single half-site preceded by three nucleotides with the consensus sequence TNAAGGTCA, referred to as an ERRE (6 -10). The transcriptional activity of ERR␣ is independent of exogenously added ligand, and its relative potency as a transcriptional activator appears to be cell context-and promoterdependent (3,8,(11)(12)(13)(14)(15). ERR␣ has also been described as a potent repressor of the SV40 late promoter (6) and to interfere with the functions of glucocorticoid, retinoic acid, and peroxisome proliferator-activated receptors (8,16,17). Although the exact physiological role of ERR␣ has not been defined precisely, increasing evidence suggest that ERR␣ plays an important role in regulating cellular energy balance. ERR␣ is predominantly expressed in tissues demonstrating a high capacity for fatty acid ␤-oxidation (3,8,18), and has been shown to regulate the medium-chain acyl coenzyme A dehydrogenase gene (8,9). More recently, ERR␣ null mice have been shown to display reduced body weight and peripheral fat deposition and be resistant to high fat diet-induced obesity (19). In agreement with the observed phenotype, gene microarray experiments demonstrated alteration in the expression of genes involved in adipogenesis, mitochondrial biogenesis, and energy metabolism, including cytochrome c, medium-chain acyl-coenzyme A dehydrogenase, acetyl-coenzyme A synthetase 2, and fatty-acid synthase (19).
The transcriptional activity of nuclear receptors is dependent on specific interactions with coregulatory proteins (20). The recent identification and functional characterization of peroxisome proliferator-activated receptor ␥ coactivator-1␣ (PGC-1␣), PGC-1␤, and PGC-1-related protein revealed the existence of a family of coactivators that possess the unique characteristic of relaying diverse physiological signals to transcription factors that regulate gene networks controlling energy balance (reviewed in Refs. 21 and 22). In particular, PGC-1␣ has been shown to regulate thermogenesis in brown fat (23), mitochondrial biogenesis and respiration in skeletal muscle (24), and glucogenesis in the liver (25)(26)(27). PGC-1␤ may also contribute to the control of energy metabolism as overexpression of this gene in transgenic mice induces a high energy expenditure and antagonizes obesity (28). Although PGC-1␣ was originally identified as a transcriptional coactivator specific for peroxisome proliferator-activated receptor ␥ (23), subsequent studies have demonstrated that PGC-1␣ influences the activity of numerous transcription factors, including a wide array of nuclear receptors (22). Recently, two groups, using similar yeast two-hybrid approaches, independently identified ERR␣ as a novel PGC-1␣-binding protein (18,29). Huss et al. (29) demonstrated that PGC-1␣ enhances ERR␣ transcriptional activity on the medium-chain acyl-coenzyme A dehydrogenase promoter. In contrast, Ichida et al. (18) described ERR␣ as a repressor of PGC-1␣ activity on the PEPCK promoter. ERR␣ transcriptional activity has also been shown to be stimulated by PGC-1␣ and -1␤ in transient transfection assays using synthetic promoters (30). Interestingly, ERR␣ and PGC-1␣ show similar expression profiles in adult tissues, including induction of expression of both genes by exposure to cold (18,30). Consistent with this observation, PGC-1␣ has been shown to induce the expression of ERR␣ (30). However, the molecular mechanisms underlying this phenomenon remain to be elucidated.
In this study, we first used a candidate gene approach to detect variants within genes encoding nuclear hormone receptors likely to play a role in physiology and be associated with disease. We first performed a family-wide screen of the genes encoding nuclear receptors in French Canadian women for novel frequent variants. This screen included exons, splice consensus sites, and ϳ1 kb of sequence located upstream of the first exon presumably containing promoter regulatory sequences. This led us to identify a 23-bp sequence referred to as ESRRA23, located at position Ϫ682 in the ESRRA gene promoter that can be found in 1-4 copies in human chromosomes. Remarkably, this sequence includes a functional ERRE that is also responsive to the presence of PGC-1␣. In this report, we describe the properties of this novel polymorphic regulatory element and functional consequences of its dosage in the ESRRA promoter responses to ERR␣ and PGC-1␣. Our results demonstrate the existence of an autoregulatory mechanism by which ERR␣ can control its own expression and further suggest the existence of an interdependent PGC-1␣/ERR␣ pathway involved in the control of energy balance.

EXPERIMENTAL PROCEDURES
Identification of a Polymorphic Repeat in the ESRRA Promoter-The primers 5Ј-CCTTGGTGTGGCCTCGACTG-3Ј and 5Ј-GCACTCGCGAG-CCAAGAGA-3Ј were used to produce a 1054-bp fragment upstream of ESRRA exon 1. PCR was performed according to standard protocols with Taq polymerase from Qiagen. PCR products were purified with a QIAquick PCR purification kit (Qiagen) and quantified by gel electrophoresis with standardized amounts of DNA. Automated sequencing of PCR products was performed with fluorescently labeled dideoxy terminators using the BigDye terminator cycle-sequencing kit on a ABI 377 DNA-Sequencer (Applied Biosystems).
Genotyping-Genomic DNA was purified from 200 l of peripheral blood leukocytes with whole blood DNA purification kits for 96-well plates from Qiagen. The collected DNA was further diluted five times with a solution of 10 mM Tris-HCl, pH 7.5, containing a 56.3 M concentration of an inert fluorescent dye, ROX (Molecular Probes, C-1309). The final concentration of ROX was 45 M in each DNA sample, and the mean DNA concentration was 5 ng/l. For each PCR, 5 l of reaction mixture was used regardless of the DNA concentration. Genotyping for the ESRRA23 minisatellite in the ESRRA promoter was performed using the following primers; 5Ј-CGTGGCCCCGCCCTTCC-3Ј and 5Ј-GTAGACCCAGTAGCCCCACAG-3Ј. PCR was performed in a 96-well microplate (Axygen) with 5 l (25 ng) of genomic DNA and 20 l of PCR premix containing 2.5 l of 10ϫ buffer (Qiagen), 200 M concentration of each dNTP, 7.5 pmol of each primer, 1ϫ Q-solution (Qiagen), 2% Me 2 SO, and 1 unit of HotStart Taq DNA polymerase. PCR setup in the microplates was performed with a Qiagen Biorobot 3000 or manually with a multichannel (8-or 12-channel) pipette. The plates were then covered with a silicone mat (Axymat from Axygen) and properly sealed using a roller (MJ Research). PCR was performed on a MJ PTC-200 (MJ Research), 95°C for 15 min, 30 cycles of 45 s at 95°C, 45 s at 58°C, 45 s at 72°C, and a final extension at 72°C for 7 min. PCR products were run on a 2% agarose gel in TBE 1ϫ buffer (0.089 M Tris, 0.089 M boric acid, 2 mM EDTA) for 2 h at 10 V/cm. The reaction produces amplicons of 198 bp with two repeats, 221 bp with three repeats, 244 bp with four repeats, and 175 bp with only one repeat. The gels were photographed, and the genotype was assigned by two independent readers. 180 samples were run in duplicate with a concordance rate of 99%.
Cloning of the ESRRA Promoter-Genomic DNA from blood samples or lymphoblastoid cell lines containing two or three repeats was prepared with the QIAamp kit (Qiagen) and used as the template in a PCR to amplify the promoter region of the human ESRRA gene. The primers used had the sequences 5Ј-GCGGTACCTGAGTGCCCTGCGCTAC-3Ј (forward) and 5Ј-CCCAAGATTCCTACTCCGCTTCCTC-3Ј (reverse) and produced a product of 1.2 kb. This fragment was digested with KpnI and HindIII and subcloned into the luciferase reporter plasmid pGL3 (Promega, Madison, WI). All of the selected clones were sequenced with fluorescently labeled dideoxy terminators using the BigDye terminator cycle-sequencing kit on a ABI 377 DNA-Sequencer.
Plasmids and Cell Transfections-The ERR␣ cDNA was cloned into the expression vector pCMX. Plasmids expressing the ERR␣-VP16 fusion protein were constructed by subcloning PCR-amplified ERR␣ cDNA into pCMX-VP16 downstream of the VP16 activation domain. The DNA binding null mutant ERR␣ DBDm was generated by substituting the glutamic acid and alanine residues of the ERR␣ P box for glycine residues. ERR␣ DBDm does not bind DNA as examined by EMSA in vitro but locates to the nucleus when transfected in mammalian cells. 2 Expression vector for human ER␣ has been described (31). The pCDNA3.1 HA-hPGC-1␣ vector was described previously (32) and obtained from A. Kralli (La Jolla, CA). The luciferase reporter plasmid ESRRA23-TKLuc and ESRRA23 (3)-TKLuc contained 1 and 3 copies, respectively, of the ESRRA23 response element (see Fig. 1A) cloned into pTKLuc. A fragment containing ESRRA promoter sequence 1.2 kb upstream of the transcriptional start site was subcloned into the luciferase reporter plasmid pGL3 to give pGL3ESRRA. To construct the ⌬ESRRA promoter luciferase reporter gene, sequences 5Ј and 3Ј adjacent to the ESRRA23 elements and putative ERR␣ binding site were amplified by PCR and subcloned sequentially into pGL3. HeLa cells were obtained form American Type Culture Collection and maintained in Dulbecco's modified Eagle medium (Invitrogen) with 10% fetal bovine serum. Mouse embryonic fibroblasts (MEFs) were isolated from 13.5-day-old wild-type and ERR␣ null embryos (19). The embryos were minced with a razor blade, and the cells were dissociated by trypsin. The cells were cultured in Dulbecco's modified Eagle medium and supplemented with 10% heatinactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Twenty-four hours prior to transfection, cells were seeded in 12-well plates and grown in phenol red-free Dulbecco's modified Eagle medium containing charcoal-treated fetal bovine serum. Onehundred ng of expression vector for nuclear receptors, 0.4 g of expression vector for the PGC-1␣ expression vector, 0.4 g of luciferase reporter, and 0.3 g of CMX ␤-galactosidase plasmids were introduced into cells using LipofectAMINE (Invitrogen) or FuGENE 6 transfection reagent (Roche Applied Science). When using LipofectAMINE, cells were maintained in the presence of liposomes for 16 h and cultured for an additional 24 h. Cells were harvested in potassium phosphate buffer containing 1% Triton X-100. Luciferase activity was determined using Steady-Glo (Promega), and values were read with the Victor 2 in the luminescence mode. The transfection was normalized to the ␤-galacto-sidase activity of each sample. All results represent experiments conducted in duplicate at least three times.
Electromobility Shift Assay-ERR␣ and ER␣ proteins were synthesized by in vitro transcription/translation using rabbit reticulocyte lysates (Promega). DNA binding reactions were conducted as described previously (33) using 5 l of programmed lysates in each binding reaction. The entire reaction was loaded onto a 5% polyacrylamide gel and electrophoresed at 150 V at room temperature. The gel was dried and exposed overnight at Ϫ85°C. The oligonucleotides and their complements that were used as probes and competitors are shown in Fig. 2A.

Identification of a Polymorphic Hormone Response Element
in the ESRRA Promoter-Our search for functional coding and regulatory polymorphisms in genes encoding members of the nuclear receptor superfamily led us to identify a 23-bp element (Fig. 1A) located at position Ϫ682 in the ESRRA promoter that is present in 1-4 copies in human chromosomes. Sequence analysis of the 23-bp element, herein referred to as ESRRA23, revealed the presence of two nuclear receptor half-site recognition motifs (Fig. 1A). The upstream half-site is preceded by the three nucleotides TGA thus generating a consensus ERR␣ binding site, also referred to as an ERRE (8). An additional putative ERRE, TCAAGGTCA, can also be found in the promoter region 1 bp downstream of the ESRRA23 sequence (Fig.  1B). The ESRRA23 element and the few base pairs flanking it, including the putative ERRE downstream of the repeated element, are absolutely conserved between human and mouse genomic sequences (Fig. 1C). However, the ESRRA23 element is present in a single copy in the mouse genome. The observed ESRRA allelic frequencies among 5490 human chromosomes (2745 individuals) were "1" ϭ 0.06, "2" ϭ 93.15, "3" ϭ 6.3 and "4" ϭ 0.36 (Table I).
Functional Characterization of the Polymorphic ESRRA23 Element-Given the observation that the polymorphic sequence contained a putative ERRE, we first tested whether the ESRRA23 motif could serve as an autoregulatory element for ERR␣. Electromobility shift assays using in vitro translated proteins and a set of oligonucleotide probes derived from the ESRRA23 element ( Fig. 2A) showed that ERR␣ binds ES-RRA23 with high specificity (Fig. 2B). Nucleotide changes within the upstream AGGTCA motif (ESRRA23m1) abolished ERR␣ binding, whereas similar mutations in the downstream CGGTCA half-site (ESRRA23m2) had little effect on recognition of the element by ERR␣ (Fig. 2B). The ERR␤ and -␥ isoforms also bound the ESRRA23 element with similar affinity (data not shown). It has been shown recently (35) that treatment with the synthetic estrogen diethylstilbestrol can enhance the expression of ERR␣. However, the related ER␣ did not significantly bind the ESRRA23 element or the downstream half-site in this assay, but it did recognize a control ERE (Fig. 2B and data not shown). Previous studies (8,10,15,36) have shown that the ERRs can bind to their response element as either monomers or homodimers. The presence of an intermediate retarded complex in a binding experiment using a mixture of wild-type and amino-terminal truncated receptors clearly demonstrates that ERR␣ preferentially binds ESRRA23 as a homodimer (Fig. 2C, arrow). We next examined whether the ESRRA23 element could act as a functional ERRE in vivo when linked to a heterologous promoter. As expected, the generally transcriptionally silent ERR␣ failed to generate a significant response when assayed in HeLa cells (Fig. 2D). To test whether ERR␣ recognizes the ESRRA23 element in vivo, we used a mammalian one-hybrid system in which ERR␣ is linked to the potent transcriptional activation domain of the viral VP16 protein. Indeed, the constitutively active ERR␣-VP16 chimera induced strong transcriptional responses (10 -57-fold) in an element-dosage manner (Fig. 2E) demonstrating that ERR␣ can recognize the ESRRA23 element in vivo. We next studied the functional consequence of the ESRRA23 regulatory variant on the ESRRA promoter itself. Human ESRRA promoters containing either 2 or 3 copies of the ESRRA23 element representing the most commonly observed genotypes were cloned upstream of the luciferase reporter gene (Fig. 3A). We also engineered a mutant ESRRA promoter construct in which all copies of the ESRRA23 element as well as the nonpolymorphic putative downstream ERRE were removed (Fig.  3A, ⌬ESRRA). ERR␣ induced a small but significant transcriptional response of 1.5-or 2-fold on the ESRRA promoter containing 2 or 3 copies of ESRRA23, respectively (Fig. 3B). The effect is specific as removal of the elements abolishes the ERR␣-induced transcriptional response. As observed with the synthetic ESRRA23-thymidine kinase (TK) promoters, the ERR␣-VP16 chimera elicited strong responses from the reporter gene driven by the ESRRA promoters containing either 2 or 3 copies of ESRRA23 being 7-and 14-fold, respectively (Fig. 3C). The mammalian one-hybrid assay thus confirms that ERR␣ can directly interact with the ESRRA promoter. The specificity of the transcriptional effect was demonstrated in that the strength of the response was directly related to the copy number of ESRRA23 and that ablation of the ESRRA23 completely abolished ERR␣-VP16-induced luciferase activity. Lastly, we used a chromatin immunoprecipitation assay to test whether endogenous ERR␣ interacts with the ESRRA promoter in the context of the native chromatin. The human breast cancer cell line MCF-7 was shown previously to express endogenous ERR␣ and therefore was used for this assay (15). As shown in Fig. 3D, an antibody raised against human ERR␣ immunoprecipitates a DNA fragment that includes the ES-RRA23 elements. Quantitative PCR showed a 25-fold enrichment of the promoter fragment over the control fragment located 4 kb upstream of the ESRRA23 element. Taken together, these experiments clearly show that ERR␣ recognizes its own promoter via the polymorphic ESRRA23 element.
The ESRRA23 Element Dictates PGC-1␣ Control of ERR␣ Expression-As introduced above, recent studies (18,29,30) have shown that the coactivator PGC-1␣ can regulate both the expression and transcriptional activity of ERR␣. However, the molecular mechanism underlying the action of PGC-1␣ on ERR␣ expression has not yet been elucidated. As shown in Fig.  4A, the introduction of PGC-1␣ alone by transient transfection in HeLa cells has a significant effect on ESRRA promoter activity, leading to a 4-and 6.5-fold induction in luciferase activity generated by the ESRRA promoter reporter constructs containing 2 and 3 copies of the ESRRA23 element, respectively. The increased ESRRA promoter activity induced by PGC-1␣ is not only ESRRA23 dosage-dependent but is mediated through ESRRA23 and possibly with a contribution by the flanking ERRE, given that deletion of the region encoding these elements from the ESRRA promoter resulted in a complete loss of the stimulatory activity (Fig. 4A). The direct involvement of the ESRRA23 element in the PGC-1␣ response was further demonstrated by the observation that PGC-1␣ can activate TK-luciferase reporter genes containing 3 copies of the FIG. 2. Functional characterization of the ESRRA23 element. A, sequences of the DNA probes used in the EMSA experiments and schematic representation of the ESRRA23-TK reporter constructs. Mutated nucleotides are shown in lowercase letters. B, ERR␣ binds to the 5Ј-half-site of the ESRRA23 element. Mutation of the 5Ј-half-site (ESRRA23m1) abolishes binding, whereas similar changes in the 3Ј-half-site (ESRRA23m2) have no effect on binding. ER␣, a closely related nuclear receptor, does not recognize the ESRRA23 element. An EMSA performed with an ERE probe is shown as a positive control. C, ERR␣ binds as a homodimer to the ESRRA23 element. An EMSA was performed with wild-type and an N-terminal-truncated ERR␣, individually (lanes 2 and 3) or in combination (lane 4). The intermediate band (arrow) indicates the formation of a dimeric complex. D and E, the ESRRA23 element confers ERR responsiveness to a heterologous promoter. The ESRRA23 element was cloned in 1 (1X) or 3 (3X) copies upstream of the herpes simplex TK promoter and cotransfected in HeLa cells together with wild-type ERR␣ or the ERR␣-VP16 chimera to assess responsiveness. Results are expressed as the -fold induction over control vector in the absence of a receptor.
ESRRA23 element but not the parent vector (Fig. 4B). The presence of a single copy of the ESRRA23 element was not sufficient to confer PGC-1␣ responsiveness to the TK promoter in HeLa cells. However, PGC-1␣ activity is much more potent in COS-1 cells on both the ESRRA and ESRRA23-TK promoters suggesting that cell context may be important for the PGC-1␣ response (data not shown).
ERR␣-dependent PGC-1␣ Activity-PGC-1␣ has been described recently (29, 30) as a potent coactivator of ERR␣ and the related ERR␥ isoform. We therefore investigated the interaction between ERR␣ and PGC-l␣ on the polymorphic ESRRA promoter in HeLa cells. As observed in Fig. 5, coexpression of ERR␣ and PGC-l␣ in HeLa cells results in a synergistic activation of the ESRRA promoter. This set of experiments also demonstrates that both the independent and combined transcriptional activities of ERR␣ and PGC-1␣ are not observed in the absence of either the region containing the ESRRA23 elements or a functional ERR␣ DNA binding domain. The response to ERR␣ and PGC-l␣ is also element dosage-dependent, as the promoter containing 3 copies of the ESRRA23 element displays higher activity than the promoter containing 2 copies of the element in the presence of these regulatory factors (Fig. 5).
We next tested whether the presence of ERR␣ was absolutely essential to the activity of PGC-1␣ on the ESRRA23 element. Mouse embryonic fibroblasts (MEFs) were derived from both wild-type and ERR␣ null mice (19) and transfected with PGC-1␣ and ERR␣, alone or in combination with the ESRRA promoter reporter construct containing 3 copies of the ES-RRA23 element. ERR␣ is not transcriptionally active in MEFs, but the constitutively active ERR␣-VP16 chimera displays identical activity in MEFs derived from both strains, indicating that ERR␣ recognizes the ESRRA promoter in a similar manner in both cell types (Fig. 6). As observed previously in HeLa cells, the introduction of PGC-1␣ in wild-type MEFs leads to a significant (3-fold) induction of ESRRA promoter activity. However, PGC-1␣ transcriptional activity is considerably reduced in ERR␣ null MEFs, but this activity can be completely restored by the introduction of exogenous ERR␣ (Fig. 6). The response to PGC-1␣ was not observed when the ⌬ESRRA construct was used as a reporter in this assay (data not shown). Taken together, these results demonstrate a central role for ERR␣ in PGC-1␣-induced activation of the ESRRA promoter via the ESRRA23 element.

DISCUSSION
In this report, we described the identification and functional characterization of a new polymorphic hormone response element (ESRRA23) present in the ESRRA gene promoter, the gene encoding the orphan nuclear receptor ERR␣. Functional analysis of the polymorphic ESRRA23 sequence showed it to act as an autoregulatory element for ERR␣ in which activity is dependent on the presence of PGC-1␣. Conversely, our results also demonstrate that PGC-1␣ activity on the ESRRA promoter is dependent on the presence of ERR␣. Our study thus delineates the molecular mechanisms by which PGC-1␣ can up-regulate ERR␣ expression (30) and by which ERR␣ can control its own expression in a positive fashion. Our results also clearly establish a direct correlation between the number of ESRRA23 repeat elements present and the response of the ESRRA promoter to ERR␣ and PGC-l␣ proteins alone or in combination. To our knowledge, this is the first example of a natural regulatory polymorphism consisting of a sequence, present in one copy or in tandem-repeated elements two, three, or four times, containing a functional hormone response element for a nuclear receptor-coactivator complex.
The expression patterns of ERR␣ and PGC-l␣ and their response to specific physiological stimuli such as cold and starvation are nearly identical (18,30). Furthermore, in agreement with these observations, PGC-l␣ has been shown to induce ERR␣ expression (30). Our functional characterization of the ESRRA23 element clearly demonstrates it to be the direct target of PGC-l␣ action. However, PGC-l␣ is a coactivator protein that does not bind ESRRA23 3 and thus requires interaction with a transcription factor that has the requisite docking site on the target promoter. One clear candidate is ERR␣ itself. ERR␣ binds to the ESRRA23 element and activates transcription from it in the presence of PGC-l␣ and other coactivators such as GRIP-1 (data not shown). The role of ERR␣ as a PGC-1␣ DNA binding partner is further corroborated by the transient transfection experiments performed in HeLa cells and MEFs. A strong activation of the ESRRA promoter in HeLa cells is only observed in the presence of both factors, and transcriptional activation is not detected with an ERR␣ mutant unable to bind DNA. Using MEFs obtained from ERR␣ null mice, we have also shown that PGC-1␣ activity is considerably reduced in those cells, indicating an important and direct role for ERR␣ in PGC-1␣ action at the ESRRA promoter. We have also observed that PGC-1␣ retains some transcriptional activity in the ERR␣ null MEFs suggesting that factors other than ERR␣ can transduce PGC-1␣ activity in these cells. Indeed, PGC-1␣ has been shown to interact with and stimulate the activity of a large number of transcription factors, including many nuclear receptors (21). The likely candidates are nuclear receptors, because the region encoding the ESRRA23 elements contains several nuclear receptor binding sites but no recognizable binding sites for other transcription factors. It has been shown recently that treatment with the estrogen agonist diethylstilbestrol stimulates the expression of ERR␣ (35). However, our studies do not support a role for ER␣ in the ERR␣-independent PGC-1␣ activity observed in ERR␣ null MEFs, because this activity can be detected with the use of steroiddeprived serum and that ER␣ does not significantly bind to the ESRRA23 element or the adjacent half-site ( Fig. 2B and data not shown). We are currently investigating the role of the ERR␥ isoform, a known partner of PGC-1␣ that recognizes the ES-RRA23 element, in the control of ERR␣ expression.
Increasing evidence points to a role for ERR␣ in regulating energy homeostasis. ERR␣ is expressed in tissues that demonstrate a high capacity for fatty acid ␤-oxidation such as the kidneys, heart, and brown fat (8,37). ERR␣ expression has been shown to be up-regulated by physiological stimuli such as cold and starvation (18,30), and ablation of the gene in mice results in reduced fat mass and a resistance to high fat dietinduced obesity (19). Similarly, PGC-1␣ is a transcriptional coactivator for many transcription factors that control biological programs linked to energy needs (21). The results presented in this study reaffirm the existence of strong physiological and functional links between ERR␣ and PGC-1␣ action and also demonstrate the convergence of PGC-l␣-based transcriptional 3 J. Barry and V. Giguère, unpublished results. MEFs obtained from wild-type or ERR␣ null mice (ERR␣KO) were cotransfected with an ESRRA promoter luciferase reporter construct containing 3 copies of the ESRRA23 element and expression vectors for ERR␣, ERR␣-VP16, and PGC-1␣, alone or in combination. Results are expressed as the -fold induction over control vector in the absence of a receptor or PGC-1␣. pathways on a polymorphic hormone response element controlling ERR␣ expression.
In conclusion, this study clearly demonstrates that the identification of regulatory variants in the human genome can reveal physiologically relevant interactions between the distinct components of complex transcriptional pathways. It would be of interest to pursue further genetic studies to investigate whether the ESRRA23 polymorphism is linked to a particular phenotype or susceptibility to metabolic diseases in the human population, including obesity and diabetes.