The Transcriptional Coactivator PGC-1 Regulates the Expression and Activity of the Orphan Nuclear Receptor Estrogen-Related Receptor (cid:1) (ERR (cid:1) )*

The estrogen-related receptor (cid:1) (ERR (cid:1) ) is one of the first orphan nuclear receptors identified. Still, we know little about the mechanisms that regulate its expression and its activity. In this study, we show that the transcriptional coactivator PGC-1, which is implicated in the control of energy metabolism, regulates ERR (cid:1) at two levels. First, PGC-1 induces the expression of ERR (cid:1) . Consistent with this induction, levels of ERR (cid:1) mRNA in vivo are highest in PGC-1 expressing tissues, such as heart, kidney, and muscle, and up-regulated in response to signals that induce PGC-1, such as exposure to cold. Second, PGC-1 interacts physically with ERR (cid:1) and enables it to activate transcription. Strikingly, we find that PGC-1 converts ERR (cid:1) from a factor with little or no transcriptional activity to a potent regulator of gene expression, suggesting that ERR (cid:1) is not a constitutively active nuclear receptor but rather one that is regulated by protein ligands, such as PGC-1. Our findings suggest that the two proteins act in a common pathway to regulate processes relating to energy metabolism. In support of this hypothesis, adenovirus-mediated delivery of small interfering RNA for ERR (cid:1) , or of PGC-1 mutants that interact selectively with different types of nuclear receptors, shows that PGC-1 can induce the

The nuclear receptor ERR␣ 1 was identified in 1988 as a protein that shares significant sequence similarity to known steroid receptors, such as the estrogen receptor (1). ERR␣ and its relatives ERR␤ and ERR␥ form a small family of orphan nuclear receptors that are evolutionarily related to the estrogen receptors ER␣ and ER␤, and whose in vivo function is still unclear (Refs. 1 and 2 and reviewed in Ref. 3). The three ERRs recognize and bind similar DNA sequences, which include estrogen response elements (EREs) recognized by ERs, as well as extended ERE half-sites that have been termed ERR response elements (4 -7). Despite their high similarity to ligand-dependent receptors, ERRs seem to regulate transcription in the absence of known natural lipophilic agonist ligands. Searches for ligands have so far identified only synthetic antagonists. 4-Hydroxytamoxifen, which binds ERR␤ and ERR␥ but not ERR␣, and diethylstilbestrol, which binds all three ERRs, inhibit the ability of ERRs to activate transcription (8,9). In support of the pharmacological data, elucidation of the crystal structure of the ERR␥ LBD suggests that the ERRs assume the conformation of ligand-activated nuclear receptors in the absence of ligand (10) and that agonist ligands may not be required. These findings raise the question of how the activity of these nuclear receptors is regulated.
One way to control orphan receptor activity is to express the receptors in a temporally and spatially restricted manner. ERR␣ is expressed widely; however, particularly high ERR␣ mRNA levels have been noted at sites of ossification during development, and in heart, kidney, brown fat, and muscle in adults (Refs. 5 and 11-16 and reviewed in Ref. 17). Thus, differential expression of ERR␣ may contribute to the regulation of ERR␣-mediated transcription. The mechanisms and signals that regulate ERR␣ expression are not clear.
The activity of orphan nuclear receptors may also be regulated at the protein level via interactions with specific cofactors. ERR␣ has been reported variably as an activator, a repressor, or a DNA-binding factor with little activity, suggesting that cellular factors determine the ability of the ERR␣ protein to activate transcription (4 -6, 11, 16, 18 -21). Possible candidates for exerting such control are coactivators that interact with ERR␣, such as members of the p160 family of coactivators. Overexpression of p160 coactivators can indeed enhance ERR␣mediated transcription at model reporters (18 -20). However, ERR␣ shows weak transcriptional activity in cells that express endogenous p160 coactivators (5,20), suggesting that additional cofactors must be important.
PGC-1 is a transcriptional coactivator of many nuclear receptors, as well as specific other transcription factors like the nuclear respiratory factor 1 (NRF-1) and members of the MEF2 (myocyte enhancer factor 2) family (22)(23)(24)(25)(26)(27)(28). PGC-1 is expressed in a tissue-selective manner, with the highest mRNA levels found in heart, kidney, brown fat, and muscle (22,25,29,30). Moreover, PGC-1 expression is induced in a tissue-specific manner by signals that relay metabolic needs. Exposure to cold leads to the induction of PGC-1 in brown fat and muscle, starvation induces PGC-1 expression in heart and liver, and physical exercise increases its expression in muscle (22,28,(31)(32)(33). PGC-1 function has been implicated in the control of energy metabolism, as PGC-1 expression stimulates mitochondrial biogenesis and modulates mitochondrial functions and utilization of energy (Refs. 23, 24, and 31 and reviewed in Ref. 34). The nuclear receptors PPAR␥, TR␣, PPAR␣, HNF4, and  GR, and the transcription factors NRF-1, MEF2C, and MEF2D, interact with, and may recruit, PGC-1 to the promoters of target genes that execute the metabolic effects of PGC-1. Additional transcription factors are likely to contribute to PGC-1 function.
In the study presented here, we show that PGC-1 regulates, first, the expression of ERR␣ mRNA and, second, the transcriptional activity of the ERR␣ protein. Our findings indicate that ERR␣ by itself is a poor activator of transcription and that PGC-1 fulfills a specific role as a cofactor required for ERR␣ function. The interactions of PGC-1 and ERR␣ suggest that the two proteins act in a common pathway.

EXPERIMENTAL PROCEDURES
Plasmids and Adenoviral Vectors-Expression plasmids for wildtype and mutant human PGC-1, and luciferase reporters pGK1, p⌬LUC, and p⌬(cERE)x2-Luc (referred in this study as pERE-Luc), have been described previously (35,36). pSG5-mERR␣ for the expression of full-length mouse ERR␣ was a gift of J.-M. Vanacker (11). The human ERR␣ ligand-binding domain (LBD) was amplified by PCR, using HeLa cDNA and primers CGAATTCATATGGGGCCCCTG-GCAGTCGCT and GCTCTAGACTATCAGTCCATCATGGCCTC, and cloned as an NdeI-XbaI fragment into pcDNA3/Gal4DBD (36). The plasmid pSiERR␣ was generated by cloning the annealed primers GAT CCC CGA GCA TCC CAG GCT TCT CAT TCA AGA GAT GAG AAG CCT GGG ATG CTC TTT TTG GAA A (ERR␣907/927-s) and AGC TTT TCC AAA AAG AGC ATC CCA GGC TTC TCA TCT CTT GAA TGA GAA GCC TGG GAT GCT CGG G (ERR␣907/927-a) into pSUPER (37). Yeast expression vectors for Gal4-PGC-1 (aa 91-408, wild-type or mutants) were generated by subcloning the PGC-1 cDNA fragments encoding aa 91-408 in the vector pGBKT7 (Clontech). Plasmid pAS2-ERR␣LBD expresses the ERR␣ LBD fused to the Gal4 DBD and was generated by subcloning the NdeI-XbaI fragment encoding the ERR␣ LBD into pAS2-1 (Clontech). pGBKT7/hER␣.280C expresses the human ER␣ LBD (starting at aa 280) fused to the Gal4 DBD. Human PPAR␥ (full-length), RXR␣ (starting at aa 10) and ERR␣ (starting at aa 221) fused to the Gal4 activation domain (AD) were isolated in a yeast two-hybrid screen and were expressed from the vector pACT2 (Clontech). Adenoviral vectors were generated by CRE-lox-mediated recombination in CRE8 cells (38). Briefly, CRE8 cells were transfected with 3 g of purified ⌿5 adenovirus DNA and 10 g of pAdlox DNA shuttle plasmid (38) carrying the cDNA for human PGC-1 (wild-type or mutant) downstream of the cytomegalovirus promoter. For the expression of siRNA from adenoviruses, the cytomegalovirus promoter and SV40 polyadenylation sequences of pAdlox were replaced by a DNA fragment harboring the expression cassette of pSUPER (37) to generate AdSU-PER. The viral vector AdSiERR␣ expresses the same siRNA as pSi-ERR␣. All viruses were plaque-isolated to obtain single clones, titered by serial dilution in CRE8 cultures that were grown under 0.6% Noble agar overlay, and used as freeze-thaw lysates.
Cell lines, Infections, and Transfections-293, CRE8 (38), HepG2, SAOS2-GR(ϩ) (39), and HtTA-1 (derived from HeLa; Ref. 35) cells were cultured in Dulbecco's modified Eagle's medium supplemented with 9% fetal calf serum. When measuring ERR␣-and GR-mediated transcription, cells were grown in medium with charcoal-stripped serum. SAOS2-GR(ϩ) and CRE8 cultures were supplemented with G418 (400 g/ml). For infection, cells were plated at 2 ϫ 10 5 per well in a six-well dish. The next day, viruses were added at a multiplicity of infection of 40 or 100, as indicated in the figure legends to Figs. 1 and 5, for 2 h. Cells were then washed and replenished with fresh medium. For transfections, cells were incubated with a calcium phosphate/DNA precipitate. Transfections included 0.2 g of p6RlacZ for normalization of transfection efficiency, and 1 g of the luciferase reporters p⌬Luc, pERE-Luc, or pGK1. The amounts of expression plasmids per transfection were as follows: 0.5-1 g of pcDNA3 or pcDNA3/HA-PGC-1, 1 g of pSG5 or pSG5-mERR␣, 1 g of pcDNA3/Gal4 DBD or pcDNA3/ Gal4-ERR␣ LBD, and 1 g of pSUPER or pSiERR␣ for siRNA. Cell lysates were prepared 40 -48 h after transfection and assayed for luciferase activity as described (25). Luciferase values normalized to the ␤-galactosidase activity are referred to as luciferase units.
RNA Analysis-Total RNA was isolated using the TRIzol reagent and checked for its integrity by agarose gel electrophoresis and ethidium bromide staining. RNA was converted to cDNA and specific transcripts were quantitated by real-time PCR using the Light Cycler system (Roche Diagnostics) as described previously (36). A melting curve from 65 to 95°C (0.05°C/s) at the end of the reaction was used to check the purity and nature of the product. In all cases, a single PCR product was detected. The sequences of the primers and the sizes of the PCR products were as follows: AAGACAGCAGCCCCAGTGAA (exon 4) and ACACCCAGCACCAGCACCT (exon 5) for human ERR␣ (product 254 bp), TGTGGAGGTCTTGGACTTGGA (exon 4/5) and TCCTCAGT-CATTCTCCCCAAA (exon 6) for MCAD (product 173 bp), CTGTGC-CAGCCCAGAACACT (exon 4) and TGACCAGCCCAAAGGAGAAG (exon 5) for 36B4/ribosomal protein P0 large (product 201 bp), CGG-GATGAGTTGGGAGGAG (exon 1) and CGGCGTTTGGAGTGGTAGAA (exon 2) for p21 (product 212 bp), GGAGGACGGCAGAAGTACAAA (exon 4) and GCGACACCAGAGCGTTCAC (exon 5) for mouse ERR␣ (product 130 bp); primers for mouse PGC-1 and actin have been described previously (36).
Yeast Two-hybrid Interaction Assays-Yeast carrying Gal4-responsive ␤-galactosidase reporters (CG1945xY187, Clontech) were transformed by the lithium acetate transformation method with expression plasmids for Gal4 DBD and Gal4 AD fusion proteins. Single transformants were grown to stationary phase, diluted 1:20 in selective media, grown for an additional 14 h at 30°C in 96-well plates, and assayed for ␤-galactosidase activity as described previously (36).

PGC-1 Induces ERR␣ Expression-
To identify genes that are induced by PGC-1 and that could execute the cellular processes activated by PGC-1, we have compared the RNA profiles of SAOS2-GR(ϩ) cells infected with adenoviral vectors expressing PGC-1 with those of cells infected with control vectors expressing ␤-galactosidase or GFP. Analysis of the RNA profiles after hybridization to high density oligonucleotide arrays (data not shown) identified the orphan nuclear receptor ERR␣ as a gene that is induced strongly by PGC-1. Expression of PGC-1 led to the induction of ERR␣ at the RNA and protein level in SAOS2-GR(ϩ) cells, as well as in HtTA-1, HepG2, and 293 cells ( Fig.  1A and data not shown). Evaluation of protein levels by immunoblotting showed that the increase in the levels of ERR␣ protein followed closely the appearance of PGC-1 protein at different times after infection, suggesting that ERR␣ induction is an early event upon PGC-1 expression (Fig. 1B).
ERR␣ mRNA levels have been reported to be high in PGC-1 expressing tissues, such as kidney, heart, muscle, and brown adipose tissue (5, 13-16, 22, 25, 29, 30). Analysis of mRNA expression levels in tissues of adult mice shows that indeed ERR␣ levels correlate with PGC-1 mRNA levels (Fig. 1C). PGC-1 expression in some of these tissues is known to be induced in response to physiological signals, such as exposure to cold (22). Thus, to test the ability of PGC-1 to induce ERR␣ in vivo, we determined PGC-1 and ERR␣ mRNA levels in the brown fat and muscle of mice that were exposed to cold for 6 h. As seen in Fig. 1D, the increase in PGC-1 expression was accompanied by an increase in ERR␣ mRNA levels, suggesting that PGC-1 can also induce ERR␣ expression in vivo.
PGC-1 Strongly Induces ERR␣-mediated Transcription-The finding that PGC-1 induces the expression of ERR␣ suggests that PGC-1 enhances also the activity of ERR␣-regulated promoters. To test this, we transfected 293 cells with a PGC-1 expression vector and a reporter that carries the luciferase gene under the control of the minimal ADH promoter with or without binding sites for ERR␣ (pERE-Luc and p⌬Luc, respectively). PGC-1 strongly enhanced expression from the pERE-Luc reporter, in a manner dependent on the presence of the binding sites for ERR␣ ( Fig. 2A). Estradiol, tamoxifen, or hydroxytamoxifen did not affect the enhancement by PGC-1 (data not shown), suggesting that it was not mediated by receptors that are regulated by these ligands and can recognize the same DNA binding site (e.g. ER␣, ER␤, ERR␤, or ERR␥). To confirm that endogenous, PGC-1-induced ERR␣ was mediating the effect of PGC-1 on the pERE-Luc reporter, we determined the effect of inhibiting the expression of ERR␣. For this, cells were transfected with a vector expressing a small interfering (si) RNA specific for ERR␣ (pSiERR␣) (37). Expression of the ERR␣-specific siRNA led to a decrease in ERR␣ mRNA levels (Fig. 2B) and a decrease in the PGC-1-mediated induction of the luciferase reporter, demonstrating that endogenous ERR␣ was required for the PGC-1 effect (Fig. 2C). In the absence of PGC-1, pSiERR␣ decreased ERR␣ expression (Fig. 2B) but had no effect on the pERE-Luc reporter (Fig. 2C), suggesting that in this context ERR␣ was not transcriptionally active.
PGC-1 Activates ERR␣ at the Protein Level-PGC-1 interacts physically with many nuclear receptors and enhances their transcriptional activity (reviewed in Ref. 34). Thus, PGC-1 could also interact with ERR␣. In this case, the increased ERR␣-mediated transcription could be the combined result of PGC-1 inducing ERR␣ levels and enhancing ERR␣ activity. To address this, we first asked whether overexpression of ERR␣ would lead to the same phenotype as PGC-1 expression. If the only function of PGC-1 were to increase ERR␣ levels, we would expect that exogenous ERR␣ expression would mimic the PGC-1 effect. Surprisingly, overexpression of ERR␣ had very little effect on pERE-Luc (Ͻ2-fold), suggesting that ERR␣ alone was not sufficient for the transcriptional activation of this reporter (Fig. 3A). Coexpression of PGC-1 with ERR␣ led to an increase in luciferase expression that was stronger than that seen with just endogenous ERR␣, indicating that PGC-1 activated the exogenously introduced ERR␣ (Fig. 3A).
To determine the effect of PGC-1 on the activity of ERR␣ directly, we evaluated the consequence of PGC-1 expression on the activity of a Gal4 DBD-ERR␣ LBD chimera, using a Gal4responsive luciferase reporter. In this context, endogenous ERR␣ does not interfere with the luciferase readout. As seen in Fig. 3B, Gal4-ERR␣ LBD by itself activated transcription modestly, ϳ2-fold, suggesting that the LBD of ERR␣ carries only a weak transcriptional activation function. Addition of PGC-1 converted the Gal4-ERR␣ LBD fusion to a strong activator of transcription, indicating that PGC-1 enables the transcriptional function of ERR␣ (Fig. 3B).
ERR␣ Interacts with PGC-1 via an Atypical L-rich Box-PGC-1 harbors three Leu-rich motifs (L1, L2, and L3), one of which (L2) bears the consensus LXXLL sequence present in many proteins that interact with the LBD of nuclear receptors. The L2 motif serves as the major binding site for many nuclear receptors, and mutations in L2 disrupt the interactions of PGC-1 with nuclear receptors tested so far (24,26,35). Surprisingly, PGC-1 harboring a mutant L2 (L2A) was still capable of interacting with ERR␣ in a yeast two-hybrid assay; in the same context, the L2A mutant was severely compromised for interaction with PPAR␥, RXR␣, and ER␣ (Fig. 4, A and B). In previous studies, we had noted that the L3 site can mediate a weak interaction with the glucocorticoid receptor (35). We thus tested the contribution of the L3 site to the PGC-1/ERR␣ interaction. As seen in Fig. 4, A and B, PGC-1 bearing a disruption of just L3 (L3A) was also capable of interacting with ERR␣, while the double L2A/L3A mutation abolished the interaction. Mutations in motif L1, alone or in combination with L2, had no effect on the physical interaction of PGC-1 with ERR␣ (data not shown). Thus, we concluded that motifs L2 and L3 can be used equivalently for physical interactions between PGC-1 and ERR␣, while L2 is the preferred site for most other receptors (Fig. 4, A and B).
Next, we determined the requirement of the physical interaction between PGC-1 and ERR␣ for the activation of the ERR␣ LBD in mammalian cells, using the context of the Gal4-ERR␣ LBD chimera. Single mutations in either L2 or L3 did not compromise the PGC-1 effect (Fig. 4C), suggesting that interaction via either site is sufficient for activation of ERR␣ by PGC-1. The double L2A/L3A mutation abolished the activation, indicating that the physical interaction between the two proteins is necessary for the effect of PGC-1 on the ERR␣ LBD (Fig. 4C).
PGC-1 Can Induce the Expression of the Endogenous Gene MCAD in an ERR␣-dependent Manner-The ability of PGC-1 to induce ERR␣ expression and activity predicts that PGC-1 should also induce the expression of ERR␣ target genes. To test this, we determined the effect of PGC-1 on the RNA levels of a proposed ERR␣ target, the MCAD, an enzyme in fatty acid oxidation (5,13). As seen in Fig. 5, PGC-1 expression led to the induction of MCAD in HtTA-1 and SAOS2-GR(ϩ) cells. To address whether the induction was mediated by ERR␣, we asked whether suppression of ERR␣ expression affected the response of MCAD to PGC-1. Infection of HtTA-1 cells with adenoviruses that express ERR␣-specific siRNA led to a decrease in ERR␣ mRNA levels (Fig. 5A), and a reduced induction of MCAD (Fig. 5B), consistent with ERR␣ mediating the PGC-1 effect at the MCAD promoter.
The distinct utilization of the L3 site of PGC-1 for interaction with ERR␣ and not other receptors like GR suggests that mutations in the L2 and L3 sites could be used to diagnose the type of nuclear receptors that mediate specific functions of PGC-1. Functions that are mediated by receptors utilizing the L2 site should be abrogated by the single PGC-1 mutation L2A, while functions that rely on ERR␣ should be disrupted only by the double L2A/L3A and not the single L2A mutation. To test this, we infected SAOS2-GR(ϩ) cells that express GR from a stably integrated locus with adenoviruses expressing PGC-1, wild-type or mutant variants. As predicted, the glucocorticoiddependent induction of the endogenous GR target p21 (39) was enhanced by both wild-type PGC-1 and the L3A mutant, but not by the L2A mutant (Fig. 5C). In contrast, induction of MCAD in the same cells was not affected by the L2A mutation and was only abolished by the double L2A/L3A mutation (Fig.  5D). These findings indicate that the L2 and L3 sites of PGC-1 are indeed used selectively by different nuclear receptors to recruit PGC-1 at their respective endogenous target genes. DISCUSSION Many members of the nuclear receptor superfamily are still orphan receptors, with no known physiological ligands. The mechanisms that regulate the activity of these receptors are not fully understood. The results presented here provide evidence that the transcriptional coactivator PGC-1 is a key regulator of the orphan nuclear receptor ERR␣. PGC-1 acts at two levels. First, it induces ERR␣ expression, and second, it associates with ERR␣ and enables the transcriptional activation of ERR␣ target genes. PGC-1 expression is known to be regulated in a tissue-selective manner by physiological signals that relay metabolic needs (22,28,(31)(32)(33). Accordingly, PGC-1 function has been implicated in the regulation of energy metabolism (Refs. 23, 24, and 31 and reviewed in Ref. 34). Our findings suggest that ERR␣ functions in PGC-1-regulated pathways, where it may contribute to the transcriptional activation of genes important for energy homeostasis.
The activity of several orphan nuclear receptors is restricted by expression of the receptors in specific tissues or at particular times (reviewed in Ref. 17). The mechanisms that control the selective expression of these receptors are often not clear. The observation that PGC-1 induces ERR␣ mRNA levels provides a molecular explanation for the high ERR␣ expression in heart, kidney, muscle, and brown fat, i.e. tissues that express PGC-1. Moreover, it suggests physiological signals that are likely to control ERR␣ expression, as shown here for exposure to cold in brown fat and muscle. In support of these findings, Ichida et al. have recently shown that fasting, which is known to induce PGC-1 expression in the liver (28,33), also increases ERR␣ mRNA levels (40). The spatial and temporal correlation of PGC-1 and ERR␣ expression implies that ERR␣ induction is an early, and possibly direct, outcome of PGC-1 action. Future studies must address whether PGC-1 acts directly at the ERR␣ promoter. Additional regulatory mechanisms may restrict or enhance ERR␣ induction by PGC-1, in a tissue-or physiological state-dependent manner.
Interestingly, we find that in the absence of PGC-1, ERR␣ is a very weak activator of transcription. Coexpression of PGC-1 enables potent transcriptional activation by ERR␣. These findings suggest that ERR␣ is not a constitutively active receptor and that transformation into an active form is favored by binding to protein ligands, such as PGC-1, rather than to small lipophilic ligands. Expression levels of PGC-1 may explain why ERR␣ has been reported as an efficient transcriptional activator in some cells (e.g. ROS 17.2/8) and a poor activator in others (4 -6, 11, 16, 18 -21). Many established cell lines express very low, if any, levels of PGC-1. Importantly, the ability of PGC-1 to activate ERR␣ at the protein level predicts that physiological signals that induce PGC-1 are likely to activate ERR␣-mediated transcription, even in the absence of increased ERR␣ expression.
The activation of ERR␣ at the protein level requires the physical interaction of PGC-1 with ERR␣. Surprisingly, this interaction differs from that of PGC-1 with other nuclear receptors. While PGC-1 recognizes most receptors tested until now (GR, ER␣, TR␣, RXR␣, RAR␣, PPAR␣, PPAR␥, HNF4) via the canonical LXXLL motif L2, it can interact with ERR␣ equally well via the L2 or the L3 site. Similar to our findings, Huss et al. have recently shown that ERR␣, as well as the related receptor ERR␥, bind the L3 site of PGC-1 (41), suggesting that the L3-mediated interaction is characteristic of the ERR subfamily of receptors. Interestingly, the differential utilization of the Leu-rich motifs can be used to dissect the receptors that mediate specific PGC-1 functions, as shown by the FIG . 5. PGC-1 induces the endogenous MCAD gene in an ERR␣dependent manner. A and B, HtTA-1 cells were infected with either control (AdSUPER) or siERR␣ expressing (AdSiERR␣) adenoviruses on day 1 and either GFP-or PGC-1-expressing adenoviruses on day 2. RNA was harvested on day 3, and mRNA levels for ERR␣ and MCAD were analyzed by quantitative RT-PCR, normalized to 36B4 levels, and expressed relative to levels in cells infected with AdSUPER/GFP viruses. Data represent the mean Ϯ S.D. of three experiments performed in duplicates. C and D, SAOS2-GR(ϩ) cells were infected with adenoviruses expressing either GFP or PGC-1 (wild type (wt) or mutants L2A, L3A, or double mutant L2A/L3A). C, 24 h after infection cells were treated with either 50 nM corticosterone (ϩH) or just vehicle ethanol (ϪH). RNA was harvested 8 h after hormone addition, and p21 mRNA levels were determined by quantitative RT-PCR, normalized to 36B4 levels, and expressed relative to levels in cells infected with GFP virus and treated with just ethanol. Data represent the mean Ϯ range of duplicates of one experiment. D, RNA was harvested 48 h after infection, and MCAD mRNA levels were determined by quantitative RT-PCR, normalized to 36B4 levels, and expressed relative to levels in cells infected with GFP virus. Data represent the mean Ϯ S.D. of two experiments performed in duplicates. Wild-type, L2A, L3A, and L2A/L3A mutants were expressed at similar levels, as determined by Western blot analysis. fact that L2A mutations disrupt GR-dependent, but not ERR␣dependent, effects of PGC-1. Thus, the L2 and L3 mutants of PGC-1 may provide useful tools for elucidating the types of receptors that recruit PGC-1 at distinct promoters.
The in vivo functions of ERR␣ are not yet defined. Based on its ability to bind EREs and modulate some estrogen-responsive genes, ERR␣ has been proposed to modulate ER signaling and possibly play a role in ER-dependent tumors (6,20,21). A function of ERR␣ in bone development is supported by the high levels of ERR␣ at sites of ossification during embryogenesis, and the ability of ERR␣ to promote osteoblast differentiation in vitro and to activate the promoter of the bone matrix protein osteopontin (11,42). Finally, the strong expression of ERR␣ in tissues with high capacity for fatty acid oxidation, and its ability to bind the promoter of the MCAD gene, suggest a role in the mitochondrial ␤-oxidation of fatty acids (5,13). Our findings support a function of ERR␣ in PGC-1-stimulated cellular processes, such as fatty acid oxidation (24), and possibly other aspects of energy homeostasis. Interestingly, the close relationship of PGC-1 and ERR␣ activity may reflect not only an involvement of ERR␣ in known PGC-1-regulated functions, but also of PGC-1 in processes where ERR␣ roles have been suggested, such as bone development and homeostasis or breast cancer.