Modulation of Estrogen Receptor-α Transcriptional Activity by the Coactivator PGC-1*

A transcriptional coactivator of the peroxisome proliferator-activated receptor-γ (PPARγ), PPARγ-coactivator-1(PGC-1) interacts in a constitutive manner with the hinge domain of PPARγ and enhances its transcriptional activity. In this study we demonstrate that PGC-1 is a coactivator of estrogen receptor-α (ERα)-dependent transcriptional activity. However the mechanism by which PGC-1 interacts with ERα is different from that of PPARγ. Specifically, it was determined that the carboxyl terminus of PGC-1 interacts in a ligand-independent manner with the ERα hinge domain. In addition, an LXXLL motif within the amino terminus of PGC-1 was shown to interact in an agonist-dependent manner with the AF2 domain within the carboxyl terminus of ERα. The ability of PGC-1 to associate with and potentiate the transcriptional activity of an ERα-AF2 mutant that is unable to interact with the p160 class of coactivators suggests that this coactivator may have a unique role in estrogen signaling. It is concluded from these studies that PGC-1 is a bona fideERα coactivator, which may serve as a convergence point between PPARγ and ERα signaling.

The steroid hormone estrogen is a key regulator of cellular processes involved in the development and maintenance of reproductive function. In addition, estrogen exhibits biological actions in bone, the cardiovascular system, and the central nervous system (1,2). In these target organs the biological action(s) of estrogen are manifest through either one of two specific, high affinity, estrogen receptors (ER␣ or ER␤) 1 located within target cells (3,4). Given the similarity in the structure of these two ER subtypes, it is likely that they share a similar mechanism of action; however, most studies thus far have focused on defining the ER␣ signal transduction pathway. These studies have revealed that the apo-receptor resides in the nucleus of target cells in an inactive form associated with a large inhibitory heat-shock protein complex (5). Upon binding ligand, the receptor undergoes a conformational change, an event that leads to the displacement of heat-shock proteins, receptor dimerization, and subsequent interaction of ER␣ with specific estrogen response elements (EREs) located within the regulatory regions of target genes (6 -8). Depending on the cell and promoter context, the DNA-bound receptor can exert either a positive or negative effect on target gene transcription (9).
Although the precise mechanism by which ER␣ modulates target gene transcription remains to be determined, significant insights in this regard have come from studies that probed the molecular pharmacology of different ER␣-ligands (10). Notable was the finding that the biological activity of the same compound can differ between cell types, indicating that ER␣ does not function in an identical manner in all contexts (10 -12). For example, the drug tamoxifen has been shown to function as an antiestrogen in most ER␣-positive breast cancer cells where it opposes the mitogenic activity of estradiol (13). In other tissues, such as the uterus and the cardiovascular system, this compound exhibits estrogenic activity (1,14,15). The molecular basis underlying the distinct activities of estrogen and tamoxifen became apparent from crystallography studies, which demonstrated that ER␣ bound to estradiol and tamoxifen assumes different conformations (16). Specifically, it was determined that the structural changes, which occur in the ligand binding domain of ER␣ when it is bound to estradiol, but not tamoxifen, facilitated the interaction of this receptor with cellular adaptors or coactivators, i.e. proteins that link ER␣ to the general transcriptional machinery (17,18). The best studied of these factors are SRC-1, GRIP-1, and AIB-1, which are members of the p160 superfamily of coactivators (19 -24). The ability of these coactivators, when overexpressed in cells, to potentiate the transcriptional activity of estradiol-activated ER␣ confirmed the functional significance of these interactions (20,22,25). The p160 coactivators exhibit a relatively ubiquitous tissue distribution pattern, making it unlikely that they play a major role in determining the tissue-selective actions of different ER␣-ligand complexes. However, the recent identification of factors, other than those of the p160 class, that interact with ER␣ and that are expressed in a tissue-selective manner adds to the complexity of estrogen action. One of these factors, PGC-1, was originally identified as a transcriptional coactivator of the peroxisome proliferator-activated receptor-␥ (PPAR␥) (26), and it was determined that PGC-1 also interacts with ER␣ in vitro, although its role as a coactivator of ER␣ has not yet been established.
It has been shown previously that PGC-1 plays a role in adaptive thermogenesis where, among many activities, it enhances the ability of PPAR␥ and nuclear respiratory factors (NRF-1 and NRF-2) to induce the synthesis of the enzymes required for oxidative metabolism (27). Commensurately, PGC-1 has been shown to be expressed and highly regulated in brown adipose tissue and skeletal muscle in rodents (26). In addition, PGC-1 is expressed in the heart, kidney, and brain, suggesting that it is involved in processes other than thermogenesis (26). Given that ER␣ target cells also exist in the brain, heart, and kidney, we were intrigued by the possibility that PGC-1 may function as an ER␣ coactivator. If PGC-1 could serve as such a coactivator, it might provide a point of convergence between PPAR␥-and ER␣-regulated signaling pathways (28).

EXPERIMENTAL PROCEDURES
Enzymes and Chemicals-Restriction and modification enzymes were obtained from Roche Molecular Biochemicals. Glutathione-Sepharose 4B was purchased from Amersham Pharmacia Biotech. The TNT T7 coupled reticulocyte lysate system was obtained from Promega. QuickChange kits were purchased from Stratagene (La Jolla, CA). 17␤-Estradiol and 4-hydroxytamoxifen were obtained from Sigma.
Plasmids-A vector expressing the ER-LL mutant was generated using oligo-directed mutagenesis as described previously (29). The ER-AF-1 and ER 282-stop mutants were generated as described (12,30), respectively. The ER 253-stop mutant was generated by introducing a stop codon (indicated by underscoring) into the ER open reading frame using pRST7ER as template and the following primers: forward, 5Ј-G-TGGGAATGATGAAATGAGGGATACGAAAAGACC-3Ј; reverse, 5Ј-G-GTCTTTTCGTATCCCTCATTTCATCATTCCCAC-3Ј.
Primers ⌬30H-F and ⌬30H-R lack sequences encoding residues 253-282 of ER␣ but anneal to sequences adjacent to the region at both ends. In a second step of PCR we used oligos pRST7F and 351R (see above) and a mixture of fragments at equimolar amounts from the first PCR step as a template. The final PCR product was digested with NotI and HindIII restriction enzymes and subcloned into the NotI and HindIII sites of pRST7 ER or pRST7-ER-LL. All products from PCR-based cloning were sequenced to ensure the fidelity of the resulting constructs.
The AflII/BglII fragment of mutagenized PGC cDNA was subcloned into the AflII/BglII restriction sites of the original expression plasmid. The presence of point mutations in the LXXLL motif in the resulting construct was confirmed by sequencing.
Cell Culture and Transient Transfection Assays-Human cervical cancer (HeLa) and hepatoma (HepG2) cells were cultured in minimum essential medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Hyclone), 0.1 mM nonessential amino acids, and 1 mM sodium pyruvate (Life Technologies, Inc.) and maintained in a humidified incubator at 37°C with 5% CO 2 . For transient transfections, cells were seeded into 24-well plates 24 h prior to transfection. For HepG2 cells, plates were coated with gelatin by incubating plates with 0.1% gelatin for 30 min at 4°C. Lipofectin (Life Technologies, Inc.)-mediated transfection has been described in detail previously (31). A DNA/Lipofectin mixture, containing a total of 3000 ng of plasmid DNA per triplicate of sample, was incubated with cells for 4 -5 h. Typically, 125 ng of receptor, 250 ng of reporter plasmid, 2500 ng of PGCpSV-Sport or pSV-Sport, as a control, and 125 ng of normalization plasmid were used in each transfection. Receptor ligands were added 14 -16 h before luciferase and ␤-galactosidase activities were measured (31).
In Vitro Binding Assays-[ 35 S]Methionine-labeled ER␣ or ER␣ mutants were synthesized using a TNT-coupled in vitro transcription/ translation system. The resultant radiolabeled proteins were incubated with GST-PGC1, or its mutants, immobilized on glutathione-Sepharose beads in binding buffer (50 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 20 mM Tris, pH 7.5) for 14 -16 h at 4°C, washed three times with a buffer composed of 100 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 20 mM Tris, pH 7.5. Proteins were eluted by boiling the beads in Laemmli loading buffer and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) followed by autoradiography.

Definition of the ER-interacting Sites within PGC-1-We
examined the ability of PGC-1, and a series of PGC-1 mutants, to interact with ER␣ using in vitro pulldown assays. The results of these analyses are shown in Fig. 1. Full length PGC-1 binds to ER␣ in the absence of ligand (Fig. 1B). This interaction is enhanced significantly by addition of the agonist 17␤-estradiol (E 2 ), whereas addition of the antagonist tamoxifen has no impact on binding (Fig. 1B). By assessing the interaction of ER␣ with fragments of PGC-1 we were able to define one domain whose interaction with the receptor was ligand-dependent and a second that bound in a ligand-independent manner. Specifically, we observed that polypeptides encoding either the amino-terminal 400 or 170 amino acids bound efficiently to E 2 -activated ER␣, but not to apo-ER␣ or that complexed with tamoxifen. The most carboxyl-terminal peptide studied (residues 604 -797) interacted with ER␣ in a ligandindependent manner. Fragments of PGC-1 encoding amino acids 1-588 or 1-670 bound ER␣ in the absence of ligand; however, their interaction was enhanced by the addition of E 2 (Fig. 1B). Cumulatively, these data suggest that there are two major contact sites for ER␣ within PGC-1: a hormone-dependent binding site located between residues 1 and 170 and a hormone-independent binding site located within the carboxylterminal half of PGC-1.
An LXXLL Motif within PGC-1 Is Required for the Liganddependent Binding to ER␣-The PGC-1 fragment 1-170 interacts with ER␣ in an agonist-dependent manner (Fig. 1B). This fragment contains an LXXLL sequence (residues 142-146), a motif that has been shown to mediate ligand-dependent recruitment of the p160-type of coactivators to nuclear receptors (32). Mutation of the conserved leucines within the p160 LXXLL motifs has been shown to disrupt these interactions (23,32,33). We tested whether the LXXLL motif present within PGC-1 is required for the hormone-dependent binding of the amino-terminal fragment of PGC-1 to ER␣. Using an in vitro GST pulldown assay we were able to demonstrate that a PGC-1 fragment (PGCLA), harboring an L142A mutation, was unable to bind to ER␣ in the presence or absence of E 2 (Fig. 2). These data suggest that the LXXLL motif is required for the ligand-dependent component of the ER␣/PGC-1 interaction.
Definition of the Regions within ER␣ That Interact with PGC-1-To define the region(s) within ER␣ that bind PGC-1, we performed a series of in vitro binding assays using either full length ER␣, a mutant lacking the AF2 domain (ER-351), or mutants lacking either part or all of the hinge region of ER␣ (ER-282 and ER-253) (Fig. 3A). These studies revealed that the ER-351 mutant, which contains an intact hinge domain, but lacks the AF2 domain, binds PGC-1 (Fig. 3B). The ER-282 fragment also binds to PGC-1 in a ligand-independent manner; however, it does not bind as well as ER-351 or ER␣-wild type. The ER␣ fragment 1-253, lacking the entire hinge region, does not interact with PGC-1. These data indicate that the hinge region of ER␣ between amino acids 253 and 283 mediates the ligand-independent component of the ER␣/PGC-1 interaction. We confirmed the importance of this domain by deleting residues 253-282 in the full-length receptor (ER-⌬30) and showing that this mutant protein did not exhibit constitutive ER␣ binding. However, as expected, ER-⌬30 was able to bind PGC-1 when tested in the presence of E 2 (Fig. 3C).
Many nuclear receptor coactivators have been shown to use LXXLL to mediate the interaction with the AF2 domain of ER␣.
The demonstration that a fragment of PGC-1, which contains an LXXLL motif, was also able to interact with ER␣ in a ligand-dependent manner suggests that the AF2 domain of ER␣ may be responsible for this protein-protein interaction. This hypothesis was tested by assessing the ability of PGC-1 to interact with an ER␣ mutant in which the AF2 domain was disrupted (Fig. 4A). Mutation of the hydrophobic residues in the ER␣ AF2 domain (ER-LL) abolishes the ligand-induced activation function and prevents the interaction of the LXXLLcontaining coactivators with ER␣ (34,35). We evaluated the ability of PGC-1 to interact with ER-LL and observed that the ligand-dependent, but not ligand-independent, component of the ER␣/PGC-1 interaction was abrogated (Fig. 4B).
Mutation of the three charged amino acid residues (D538, E542, D545 3 N538, Q542, N545) in helix 12 of the AF2 domain (ER-3X) leads to the loss of ER␣ transcriptional activity in some, but not all, cell contexts (12,36). This mutation abrogates recruitment of p160-type coactivators such as GRIP-1 and SRC-1 but not RIP-140 (19,25,37). Although these coactivators all interact with ER␣ through their LXXLL motifs, it is clear that not all LXXLL motifs interact with ER␣ in the same manner (35). Because the LXXLL motif and the surrounding amino acids in PGC-1 resemble those in RIP-140, we determined whether or not PGC-1 is capable of interacting with the ER␣ mutant (ER-3X). The results of these experiments are shown in Fig. 4B and demonstrate that full length PGC-1 and the ER-3X mutant interact and that their association is enhanced by the addition of E 2 . Thus, PGC-1, through the LXXLL motif, interacts with the AF2 domain of ER␣ in a manner that is different from that of the p160-type of coactivators. A double mutant in which the hinge region and AF2 domains have been disrupted (ER⌬30/LL) is unable to bind PGC-1 in either the presence or absence of E 2 (Fig. 4B). This result confirms the importance of both the AF2 domain and the hinge region in mediating the interaction of ER␣ with PGC-1 and indicates that, if other binding sites on ER␣ are involved in this interaction, they do not contribute in a significant manner. We conclude that the ER␣/PGC-1 interaction differs from that of PPAR␥/PGC-1 and, unlike the latter, is enhanced by the addition of the hormone.
PGC-1 as Coactivator of ER␣-dependent Transcriptional Activity-The functional significance of the ER␣/PGC-1 interactions were next evaluated using estrogen-responsive transcription systems reconstituted in heterologous cells. Initially, the ability of PGC-1 to potentiate the transcriptional activity of wtER␣ was assessed in HeLa cells (Fig. 5). In this cell line, ER␣ functioned as an estradiol-dependent transactivator when assayed on a TATA promoter containing either one (1XERE) or three (3XERE) copies of a canonical ERE. However, when PGC-1 was coexpressed in these cells, ER-dependent transcriptional activity was significantly elevated. Thus, for the first time we demonstrate that PGC-1 is capable of activating ERdependent transcriptional activity and therefore is a bona fide coactivator of ER␣. Interestingly, the partial agonist activity of tamoxifen was not affected by PGC-1 overexpression. To determine the functional significance of ER␣/PGC-1 contact sites we have identified, we next evaluated the impact of PGC-1 on the transcriptional activity of ER␣ mutants. When compared with wtER␣, the transcriptional activity of ER-LL, ER⌬30, ER-3X, and the double mutant ER⌬30/LL were greatly reduced, reaffirming the importance of both the AF2 and the hinge regions in ER␣ action. The transcriptional activity of the mutant ER-LL, which contains an intact hinge domain and which interacts with PGC-1 in vitro, is not potentiated by PGC-1 overexpression. Thus, although PGC-1 interacts well with the hinge domain of ER␣ in vitro, this interaction alone is not sufficient for coactivation. In contrast, deletion of the PGC-1 binding site within the hinge region of ER␣ (ER⌬30) reduces, but does not eliminate, the ability of this coactivator to potentiate ER␣mediated transcriptional activity. Indeed, in PGC-1-expressing cells, wtER␣ and ER⌬30 displayed similar transcriptional activity when assayed using the 3XERE-reporter vector. Thus, although PGC-1 can bind to both the hinge and AF2 domains, it appears that, with respect to function, the contact mediated by the AF2 domain is the most important for the transcriptional activation. This conclusion is consistent with the observation that an intact AF2 domain is important for ER␣/coactivator interactions. The recent solution of the co-crystal structure of the ER␣-ligand binding domain with a fragment of the coactivator GRIP-1 provides a universal model with which to explain ER␣/coactivator interactions (16). The interaction

FIG. 4.The domains that permit ligand-independent and liganddependent interactions between ER␣ and PGC-1 are distinct.
A, a schematic representation of the wild type ER␣ and ER␣ mutants used in this analysis. Residues that were altered in each of the specific mutants are indicated. B, the interaction between PGC-1 and ER or ER␣ mutants was assessed using an in vitro GST pulldown assay. Full length PGC-1 fused to GST was immobilized on glutathione-Sepharose beads and incubated with [ 35 S]methionine-labeled ER or a specific mutant (ER-LL, ER-3X, and ER-⌬30/LL) in the absence of hormone (NH) or with 10 Ϫ6 M 17␤-estradiol (E2). A control lane on each panel (C) represents the amount of ER␣ or ER␣ mutant that bound to GST alone.

FIG. 5. PGC-1 functions as an ER␣ coactivator in HeLa cells.
HeLa cells were transfected with vectors expressing either wtER␣ or its mutants (ER-LL, ER-⌬30, ER-⌬30/LL, or ER-3X) along with 1XERE-TATA-Luciferase reporter (A) or 3XERE-TATA-Luciferase reporter (B). To assess the ability of PGC-1 to enhance ERE-dependent transcription, we also introduced either empty vector or vector expressing PGC-1. Luciferase activity was normalized to the activity of the cotransfected pCMV-␤gal plasmid. Cells were grown in the absence of hormone (gray bar) or were treated either with 10 Ϫ7 M of 17␤-estradiol (black bars) or with 10 Ϫ7 M 4-hydroxytamoxifen (white bars). Transfections were performed in triplicate (n ϭ 3). The error is presented as standard error of the mean. between GRIP-1 and ER␣ could be abrogated by mutating the charged amino acid residues in the AF2 domain (ER-3X). Therefore, it was surprising that PGC-1 can function as a very efficient coactivator of the ER-3X mutant (Fig. 5). However, this observation is consistent with our in vitro data, which indicate that PGC-1 can in fact bind directly to ER-3X (Fig. 4), an ER␣ mutant that is unable to interact with the p160 class of coactivators (25). Thus, although AF2 mediates the functional interaction between ER␣ and PGC-1, it appears that the interaction of PGC-1 with the AF2 domain of ER is different from that of most known p160-type coactivators. The potential significance of this finding, which distinguishes PGC-1 from other known coactivators, prompted us to examine the ability of PGC-1 to potentiate ER␣ action in other cell and promoter contexts.
Previously, we and others have demonstrated that the two functional domains of ER␣ (AF1 and AF2), function in a cooperative manner in most cell and promoter contexts (12,30,38). However, in some environments it appeared that AF1 alone was sufficient for ER␣-mediated transcriptional activity (39). The most compelling data in support of this hypothesis came from studies that showed that the transcriptional activity of ER␣ mutated in AF2 was affected differentially by cell and promoter context. For instance, we demonstrated previously that the ER␣ mutant, ER-3X, was inactive in most cells, attesting to the importance of AF2 (12). In some contexts, most notably in HepG2 cells, it was observed that the transcriptional activity of this mutant was indistinguishable from wtER␣ (12,25). These data appeared to indicate that AF1 was the dominant activator in this context. Given the observation that PGC-1 can interact with and potentiate the activity of ER-3X, it now appears that the amino acid changes within this mutant do not destroy AF2 but merely block its ability to interact with certain classes of coactivators, but not with PGC-1. To explore this possibility further, we performed an analysis of PGC-1 function in HepG2 cells on the complement 3 promoter (C3 Luc). The results of this analysis are shown in Fig. 6. Both ER-3X and wtER␣ display equivalent responses to E 2 . Interestingly, neither ER-LL, ER⌬30, nor ER⌬30/LL, all of which can bind E 2 and contain an intact AF1, are capable of activating the complement 3 promoter. Thus, AF1 alone is not suffi-cient for ER␣-mediated transcriptional activity in this context. When the transcriptional activity of these ER␣ proteins was assayed in cells overexpressing PGC-1, it was observed that the activity of wtER␣ and ER-3X were greatly enhanced. A minor enhancement of ER⌬30 activity was also observed, but the transcriptional activity of the AF1-containing ER-LL mutant was unaffected. To exclude the possibility that the activation of ER␣-dependent transcription is a result of increased ER␣ protein expression in cells expressing PGC-1, we performed a Western immunoblot analysis using anti-ER␣-specific antibodies. This analysis revealed that ER␣, and mutants thereof, were expressed at the same level in the presence and absence of PGC-1 (data not shown). Thus, the mutations considered previously to disrupt AF2 activity (ER-3X) were sufficient to block the interaction of ER␣ with most, but not all, coactivators. The ability of ER-3X to function in a cell-selective manner in some contexts, therefore, may reflect the cell-specific expression and regulation of cofactors like PGC-1, which can interact with and potentiate the transcriptional activity of ER-3X.
An Intact LXXLL Motif in PGC-1 Is Necessary for Full Coactivation of ER␣ Transcriptional Activity-The interaction of PGC-1 with PPAR␥ does not require the LXXLL motif found in the amino terminus of PGC-1. Therefore, it was surprising that an intact AF2 domain was required to enable PGC-1-mediated enhancement of ER␣ transcriptional activity. This observation, however, is in agreement with the in vitro binding studies that revealed that an intact LXXLL domain within PGC-1 is required for the ligand-dependent component of the ER␣/PGC-1 interaction. Therefore, the AF2 domain of ER␣ and the LXXLL motif of PGC-1 are required to permit these two proteins to functionally interact. To confirm this we mutated the LXXLL motif within PGC-1 to AXXAL and assessed the ability of the mutant protein (PGC LALA) to potentiate ER␣ transcriptional activity in different cell and promoter contexts. The results of this analysis are shown in Fig. 7. ER␣-mediated transcriptional activity was measured in HeLa cells using the 1XERE TATA luciferase reporter construct. As observed above, ER␣ activity was enhanced by overexpression of PGC-1. The activity of ER-3X and ER⌬30 was potentiated as well. When these experiments were performed using the PGC LALA mutant, we observed that, although it was able to function as an ER␣ coactivator, its activity was only 25% that of the wild type PGC-1. The ER⌬30 mutant, lacking the hinge region, was potentiated equally well by PGC-1 and PGC LALA, whereas ER␣ mutated at the AF2 domain (ER-3X) was minimally affected by PGC LALA overexpression. Similar results were obtained when the transcriptional activity of ER␣ and the selected mutants were measured using the C3 luciferase reporter construct in HepG2 cells (Fig. 7B). Cumulatively, these data confirm the importance of the LXXLL motif in mediating the functional interaction between ER␣ and PGC-1. It also suggests that, although PGC-1 potentiation requires an intact ER␣-AF2 domain, sequences in addition to the LXXLL domain may be involved in forming the ligand-dependent interaction between ER␣ and PGC-1.

DISCUSSION
The recent discovery of coactivators, proteins that can interact with and enhance the transcriptional activity of agonistactivated ER␣, and the demonstration that their overexpression in target cells influences ER␣ pharmacology indicate that differential co-factor expression is a primary determinant of a cell's ability to respond to different agonists and antagonists (18,37,40,41). The first bona fide steroid hormone receptor coactivator, SRC-1, was identified by virtue of its ability to interact with the hormone binding domain of agonist-activated progesterone receptor (20). Subsequently, it was demonstrated FIG. 6. PGC-1 potentiates estrogen receptor transcriptional activity in HepG2 cells. HepG2 cells were transfected with vectors expressing either wtER␣ or its mutants (ER-LL, ER-⌬30, ER-⌬30/LL, or ER-3X) along with C3 Luc reporter. To assess the ability of PGC-1 to enhance ERE-dependent transcription, we also introduced either empty vector or vector expressing PGC-1. Luciferase activity was normalized to the activity of the cotransfected pCMV-␤gal plasmid. Cells were grown in the absence of hormone (gray bars) or were treated either with 10 Ϫ7 M of 17␤-estradiol (black bars), or with 10 Ϫ7 M 4-hydroxytamoxifen (white bars). Transfections were performed in triplicate (n ϭ 3). The error is presented as standard error of the mean. that SRC-1 was able to interact efficiently with most of the nuclear receptors. The physiological relevance of this interaction was confirmed by demonstrating that mice bearing a genetic disruption of the SRC-1 gene display only a mild form of resistance to estrogens and progestins (42). The subtle phenotypes exhibited by the SRC-1 knockout mice are probably due to the fact that other coactivators can functionally substitute for SRC-1. Indeed, two proteins closely related in structure to SRC-1, GRIP-1 and ACTR have been identified, and each has been shown in cell transfection studies to exhibit the properties of a coactivator (21,43). Because of the similarity in their structure and function, the latter proteins have collectively been described as the p160 coactivators. In addition to the p160 class of coactivators, there are several additional proteins such as TRAP220, RIP140, DAX-1, and SHP-1 that also have been shown to interact with the hormone-activated ligand binding domain of ER␣ (19, 44 -46). However, their role in ER␣ action has not yet been established. It is interesting that none of the coactivators identified thus far are specific for one or another type of receptor. This is somewhat surprising, because nuclear receptors regulate very different physiological processes. One consequence of this is that coactivators identified by virtue of their ability to interact with one receptor are generally found to interact with several different receptors. One example of this is PGC-1, identified initially as a PPAR␥ coactivator but which was subsequently shown to bind to ER␣, TR␤, retinoid X receptor ␣, and retinoic acid receptor ␣ (26). In this study we demonstrated that PGC-1 is an ER␣ coactivator that significantly enhances the efficacy of estrogenic ligands when overexpressed in target cells.
The PGC-1 interaction site on PPAR␥ has previously been mapped to a single region within the hinge domain of the receptor (26). Interestingly, the interaction does not involve the LXXLL motif found at the amino terminus of PGC-1 nor the AF2 domain of PPAR␥ receptor. In our studies we have shown that the hinge domain in ER␣ (residues 253-282) interacts with PGC-1. In addition, we have shown that the AF2 domain of ER␣ also contacts PGC-1. The interaction of the AF2 domain of the receptor with the amino terminus of PGC-1 requires an LXXLL motif within the coactivator.
Crystallographic analysis of the ER␣ revealed that, upon binding of an agonist, the receptor undergoes a conformational change that results in the repositioning of four helices (H3, H4, H5, and H12) within the ligand binding domain and the subsequent formation of a hydrophobic cleft (16). This hydrophobic cleft constitutes the functional AF2 domain and forms a pocket for the LXXLL motif contained within the p160 coactivators. The functional importance of the hydrophobic cleft in binding PGC-1 was confirmed by showing that mutations, which alter the hydrophobicity (ER-LL) of the AF2 domain, prevent the interaction of ER␣ with the LXXLL domain and significantly reduce the ability of PGC-1 to coactivate ER␣. However, we also demonstrated that an intact hinge region in ER␣ was required for maximal coactivation by PGC-1. Our unexpected results have led us to conclude that PGC-1 is a versatile coactivator, which interacts with ER␣ and PPAR␥ by different mechanisms.
Further dissection of the interaction between the LXXLL motif of PGC-1 and ER␣ mutated at the AF2 domain revealed that PGC-1 may interact with the AF2 domain differently from p160-type coactivators. Specifically, we and others had determined that mutation of the charged residues (D538, E542, D545 3 N538, Q542, N545) within the AF2 domain (ER-3X) prevented ER␣ from interacting with the p160 class of coactivators (35). This loss of interaction was thought to occur as a consequence of a perturbation in the formation of a "charge clamp," which was required to position the LXXLL motif in the coactivator groove. In contrast, these charged residues in ER␣ were not required for PGC-1 binding. This indicates that the mechanism by which p160 proteins and PGC-1 interacted with the AF2 domain of ER␣ is different and that, although the LXXLL motif within PGC-1 or the p160 coactivators is required for this interaction, these motifs are not equivalent. Our recent studies using phage display revealed that, based on their primary amino acid sequence, there are at least three different types of LXXLL motifs (35). Although all three classes of LXXLL motifs interacted with the AF2 domain of ER␣, we demonstrated that one class alone (class III) was able to interact with the ER-3X mutant. Interestingly, the LXXLL motif within PGC-1 is a class III member, whereas LXXLL motifs present in p160-type coactivators are mostly class I or II members. Thus, the differences in binding characteristics of PGC-1 and p160-type coactivators to ER␣ can be attributed to the differences in flanking amino acid sequences in LXXLL motifs.
The discovery of the class III LXXLL motif and the demonstration that this motif is found in PGC-1 could explain transcriptional activity of the ER-3X mutant in some cell environments. Taken together the results of our studies and those of others indicate that the ER-3X mutation may not totally dis- FIG. 7. Mutations in the LXXLL motif within PGC-1 decrease its ability to potentiate ER␣ transcriptional activity. A, HeLa cells were transfected with expression vector for either wild type ER or its mutants (ER-⌬30 or ER-3X) along with either empty vector (pSV-Sport), vector expressing PGC-1, or vector expressing the PGCLALA mutant. The ability of PGC-1 or its mutants to enhance ER-dependent transcription was assayed using cotransfected reporter plasmid 1XERE-TATA-Luc. B, similar experiments performed using the C3 Luc reporter in HepG2 cells. Luciferase activity was normalized to the activity of the cotransfected pCMV-␤gal plasmid. Cells were grown in the absence of hormone (gray bars), treated with 10 Ϫ7 M of 17-␤-estradiol (black bars). Transfections were performed in triplicate (n ϭ 3) for each cell line. The error is presented as standard error of the mean. rupt AF2 function but rather block the interaction of ER␣ with coactivators that contain specific types of LXXLL motifs (class I or II) (35). Thus coactivators, like PGC-1, which contain a class III LXXLL motif and which display a tissue-restricted expression pattern, are likely to be important for the ER␣ action in some cell and promoter contexts.
We demonstrated in these studies that PGC-1 is an ER␣ coactivator, which is functionally and mechanistically distinct from the p160 family of coactivators. Although the physiological relevance of this interaction requires further investigation, we believe that PGC-1 may serve as a point of convergence between the ER␣ and PPAR␥ signaling pathways. One intriguing possibility is that all three signaling systems may be involved in mitochondrial biogenesis and oxidative metabolism (27). Indeed, we have recently shown in MCF-7 breast cancer cells that estradiol induces the transcription of genes required for mitochondrial function and in an indirect manner enhances the transcription of genes encoded by the mitochondrial genome. 2 Our findings also demonstrate that coactivators can interact with ER␣ in different ways. This result, combined with the finding that the structure of ER␣ is influenced by the nature of the bound ligand, provides a molecular model with which to explain how different ligands acting through the same receptor can manifest different biological activities in different cells (34).