Requirement of Helix 1 and the AF-2 Domain of the Thyroid Hormone Receptor for Coactivation by PGC-1*

Although PGC-1 (peroxisome proliferator-activated receptor- (cid:1) coactivator-1) has been previously shown to enhance thyroid hormone receptor (TR)/retinoid X re-ceptor-mediated ucp-1 gene expression in a ligand-in-duced manner in rat fibroblast cells, the precise mechanism of PGC-1 modulation of TR function has yet to be determined. In this study, we show that PGC-1 can po-tentiate TR-mediated transactivation of reporter genes driven by natural thyroid hormone response elements both in a ligand-dependent and ligand-independent manner and that the extent of coactivation is a function of the thyroid hormone response element examined. Our data also show that PGC-1 stimulation of TR activity in terms of Gal4 DNA-binding domain fusion is strictly li-gand-dependent. In addition, an E457A AF-2 mutation had no effect on the ligand-induced PGC-1 enhancement of TR activity, indicating that the conserved charged residue in AF-2 is not essential for this PGC-1 function. Furthermore, GST pull-down and mammalian two-hy-brid assays demonstrated that the PGC-1 L XX LL motif is required for ligand-induced PGC-1/TR interaction. This agonist-dependent PGC-1/TR interaction also requires both helix 1 and the AF-2 region of the TR ligand-bind-ing domain. Taken together, these results support the notion that PGC-1 is a bona fide TR coactivator and that PGC-1 modulates TR activity via a mechanism different from that utilized with peroxisome Analysis— GST pull-down assays were performed as previously described (49) with minor modifications. Bacterially expressed GST fusion proteins bound to glutathione-Sepharose 4B beads were incubated with in vitro translated 35 S-labeled receptors in binding buffer containing 20 m M Tris (pH 7.5), 75 m M KCl, 50 m M NaCl, 1 m M EDTA (pH 8.0), 0.05% Nonidet P-40, 10% glycerol, 1 m M dithiothreitol, and one tablet of protease inhibitor mixture (Roche Molecular Bio- chemicals). After incubation for 1–2 h at room temperature in the presence or absence of 10 (cid:2) 6 M T 3 , the beads were extensively washed with binding buffer, and the bound proteins were analyzed by SDS-PAGE and visualized by autoradiography.

domain (DBD) containing two zinc fingers (5), and a C-terminal ligand-binding domain (LBD) containing a ligand-dependent activation function (AF-2) and a major receptor dimerization interface (6 -9). In addition to those three major functional domains, a hinge region separating the TR DBD from the LBD also appears to be important for receptor function (10).
Control of gene expression is a dynamic and complex process. Gene transcription involves assembly of multiple transcription factors at the distal enhancer region and the basal transcriptional machinery, including RNA polymerase II, at the core promoter of the target gene. Like other transcription factors, TRs exert their effects on gene regulation via binding to specific DNA sequences referred to as thyroid hormone response elements (TREs) in the regulatory regions of T 3 target genes (11). Unlike the classical steroid receptors, TRs can positively or negatively regulate T 3 -responsive gene expression, depending on the nature of the TREs and auxiliary proteins (12). In the past few years, there has been enormous progress in elucidating the molecular mechanisms of nuclear receptor-mediated gene expression. In the current model, TR-modulated gene expression involves the sequential assembly of an array of coregulatory proteins, including coactivators and corepressors (13,14). In general, the binding of unliganded TRs to "positive" TREs results in repression of basal transcription, and this silencing is mediated by interaction of TR with a corepressor complex that contains histone deacetylase activity (15,16). Binding of ligand triggers significant conformational changes (17), including the repositioning of the amphipathic helix 12 containing the core of AF-2 in the TR LBD (14,18), that result in release of the corepressor complex and recruitment of a coactivator complex (19). A coactivator complex is usually associated with histone acetyltransferase activity, through which the chromatin structure may be modified, leading to transcriptional activation (20,21). It appears that ligand-dependent recruitment of coactivators is critical for TR and other nuclear receptor-mediated transcriptional activation. To date, a large number of coactivators for TR have been defined and characterized. They include at least several proteins: 1) three members of the structurally and functionally related p160 family, SRC-1/NCoA-1 (19), TIF2/GRIP-1/NCoA-2 (22,23), and p/CIP/ ACTR/AIB1/RAC3/TRAM-1 (24 -28); 2) p300/CBP (cAMP response element-binding protein-binding protein) (29) and p/CAF (p300/CBP-associated factor) (30); and 3) TRAP/DRIP/ ARC (TR-associated proteins) (31). Most of the characterized coactivators interact with the LBDs of ligand-bound nuclear receptors, including TRs, through helical LXXLL motifs (where X is any amino acid) present within p160 family members and other coactivators (32). p160 family proteins and other coactivators contain one or more copies of the LXXLL motif in their central nuclear receptor interaction domains (33). Thus, TRs and other nuclear receptors may differentially interact with various combinations of these multiple LXXLL motifs to assem-ble distinct coactivator complexes (34,35), possibly leading to selective expression of target genes. However, emerging evidence indicates that tissue-specific or inducible coactivators may play a critical role in achieving selective and tissue-specific gene expression (36,37).
Unlike most characterized coactivators that are often expressed ubiquitously, PGC-1 (PPAR␥ coactivator-1) displays a pattern of tissue-specific expression and is inducible by various stimuli, including exposure to cold temperature and exercise (37,38). Although PGC-1 was initially identified as a coactivator for the nuclear receptor PPAR␥, accumulating data have shown that PGC-1 plays a broader role in mediating transactivation by other nuclear receptors and transcription factors (39). Most importantly, PGC-1 has been implicated in the regulation of many important physiological processes, including adaptive thermogenesis (37,40) and hepatic gluconeogenesis (41). Given that T 3 is a key regulator of energy metabolism and that the major target tissues of T 3 in mammals overlap with the selective distribution of PGC-1, a role of PGC-1 as an important physiological modulator for TR action is suggested. Previous observations that PGC-1 stimulates TR/retinoid X receptor-mediated ucp-1 (uncoupling protein-1) gene transcription in a ligand-induced manner in rat fibroblast cells has strongly supported this notion (37). However, the precise mechanism of PGC-1 coactivation of TR has yet to be determined. Studies of several other nuclear receptors, including PPAR␥, PPAR␣, estrogen receptor-␣, and GR, have suggested that PGC-1 may adopt distinct mechanisms of action, depending on the identity of the nuclear receptor in question (37,(42)(43)(44). For example, PGC-1 interacts with PPAR␥ in a ligand-independent fashion via the hinge region of the receptor (37), whereas the ligand-induced interaction with PPAR␣, estrogen receptor-␣, and GR depends on the AF-2 region of the receptor (42)(43)(44). In this report, utilizing GST pull-down and mammalian two-hybrid assays, we provide evidence that PGC-1 interacts with TR␤1 in a largely ligand-dependent manner. The LXXLL motif of PGC-1 and the intact AF-2 region of the receptor mediate the agonist-dependent PGC-1/TR interaction. Surprisingly, helix 1 of the TR LBD is also essential for this process. Furthermore, functional studies demonstrate that the effect of PGC-1 on TR transcriptional activation can be ligand-dependent and ligandindependent, depending on the structure of the response element with which TR interacts.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The 1XTRE-tk-Luc and 3XTRE-tk-Luc reporters, containing one and three copies, respectively, of a characterized direct repeat TRE derived from the rat ␣-myosin heavy chain (45), were generated by inserting double-stranded oligonucleotide response elements upstream of a minimal thymidine kinase promoter linked to a luciferase gene (pTAL, CLONTECH). The DR4-tk-Luc reporter contains a copy of a direct repeat TRE from human 5Ј-deiodinase type I, TRE2 (46). The IP6-tk-Luc reporter contains a single copy of an inverted palindrome TRE from mouse myelin basic protein (47). PCRamplified full-length human TR␤1 and TR␣1 and rat TR␤2 cDNAs were cloned into the pcDNA3.1(Ϫ) expression vector (Invitrogen). The TR␤1 E457A mutant was generated by PCR-based mutagenesis. pcDNA3-PGC-1 expressing full-length human PGC-1 was a generous gift from Dr. A. Kralli and has been previously described (43). Gal4-TR␤1 and the various deletions and mutations were created by cloning PCR-amplified DNA fragments corresponding to the different TR regions into the EcoRI and SalI sites of the pM vector (CLONTECH). These varieties of DNA fragments were also cloned into pcDNA3. Gal4 DBD-PPAR␥ LBD was described previously (48). VP16-PGC-1 and the LXXLL motif mutant VP16-PGC-1 AXXAL were generated by cloning PCR-amplified DNA encoding PGC-1 amino acids 100 -411 into the EcoRI and XhoI sites of the VP16 vector. The GRIP-1 expression vector was obtained from Dr. R. J. Koenig. To ensure the fidelity of the resulting constructs, all predicted mutations and PCR-based constructs were verified by DNA sequencing.
Cell Culture and Transfections-Hela and CV-1 cells were routinely maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone Laboratories). Prior to transfection, the cells were seeded in 24-well plates at a density of 5 ϫ 10 4 cells/well in Dulbecco's modified Eagle's medium supplemented with 10% serum. After 16 h of growth at 37°C and 5% CO 2 , cells were transfected with LipofectAMINE 2000 reagent (Invitrogen) according to the manufacturer's protocol. The next morning, the transfected cells were washed with phosphate-buffered saline, and fresh media containing 10% charcoal-stripped serum and 10 Ϫ6 M T 3 , if indicated, were added. After 24 h, the cells were washed with phosphate-buffered saline and harvested with a cell culture lysis buffer (Promega). Luciferase activity was measured in a Dynex luminometer. Generally, all transfections included 50 ng of expression vector for receptors, 100 ng of expression vector for coactivators, 250 ng of pG5-Luc reporter, and 100 ng of TRE-Luc reporter. All experiments were done at least twice in triplicate.
GST Pull-down Analysis-GST pull-down assays were performed as previously described (49) with minor modifications. Bacterially expressed GST fusion proteins bound to glutathione-Sepharose 4B beads were incubated with in vitro translated 35 S-labeled receptors in binding buffer containing 20 mM Tris (pH 7.5), 75 mM KCl, 50 mM NaCl, 1 mM EDTA (pH 8.0), 0.05% Nonidet P-40, 10% glycerol, 1 mM dithiothreitol, and one tablet of protease inhibitor mixture (Roche Molecular Biochemicals). After incubation for 1-2 h at room temperature in the presence or absence of 10 Ϫ6 M T 3 , the beads were extensively washed with binding buffer, and the bound proteins were analyzed by SDS-PAGE and visualized by autoradiography.

PGC-1 Potentiates TR-mediated Transcriptional Activation-
To determine the ability of PGC-1 to coactivate TR-dependent transactivation from different TREs in various cell lines, we performed transient transfection experiments in HeLa and CV-1 cells using various luciferase reporter constructs. The 1XTRE-tk-Luc and 3XTRE-tk-Luc reporters contain one and three copies, respectively, of a direct repeat TRE derived from the rat ␣-myosin heavy chain gene upstream of a minimal thymidine kinase promoter linked to a luciferase reporter gene, whereas the DR4-tk-Luc construct contains a single copy of a direct repeat TRE from the human 5Ј-deiodinase type I gene. The IP6-tk-Luc reporter contains a single copy of an inverted palindrome TRE from the mouse myelin basic protein. Fig. 1 shows that, in the presence of T 3 , PGC-1 potently augmented the transcription of 1XTRE-tk-Luc by TR␤1 4.6-fold. This PGC-1-mediated activation was further elevated to 8.7-fold with the 3XTRE-tk-Luc reporter. Similarly, the activities of the TR-dependent DR4-tk-Luc and IP6-tk-Luc re- HeLa or CV-1 cells were cotransfected with plasmids expressing human TR␤1 and human PGC-1 together with the 1XTRE-tk-Luc (direct repeats), 3XTRE-tk-Luc (direct repeats), DR4-tk-Luc, or IP6-tk-Luc (inverted palindrome) reporter. After transfection, the cells were treated with 10 Ϫ6 M T 3 for 24 h as indicated, prior to measurement of luciferase activity. All experiments were done in triplicate, and data are displayed as the means Ϯ S.E. of a single experiment, representative of at least four independent experiments. porters were stimulated by PGC-1 ϳ5.2and 6.9-fold, respectively. Intriguingly, expression of PGC-1 also resulted in a 3.7-fold ligand-independent increase in TR-mediated DR4-tk-Luc reporter activation relative to a vector control containing no TRE, whereas a mild increase (1.5-fold) was also observed with either the 1XTRE-tk-Luc or IP6-tk-Luc reporter. In contrast, no such ligand-independent PGC-1 enhancement was obtained with 3XTRE-tk-Luc. Similar results were obtained using CV-1 cells (data not shown). Thus, the effect of PGC-1 on TR-mediated transactivation has both ligand-dependent and ligand-independent components that are a function of the TRE utilized. Furthermore, these effects do not appear to be dependent on the cell type utilized.
There are at least three functional TR isoforms: TR␣1, TR␤1, and TR␤2. Although they are structurally and functionally related, expression of TR␣1 and TR␤ (TR␤1 and TR␤2) in distinct but overlapping patterns suggests that the TR isoforms may have both distinct and common functional roles (3,50). PGC-1 also displays a tissue-specific pattern of expression and has been shown to be highly expressed in cold-induced brown fat cells and in skeletal muscle (37). Because both PGC-1 and TRs exhibit tissue-selective expression, we investigated whether PGC-1 function on TR exhibits an isoform specificity. To address this possibility, the effect of PGC-1 on the transcriptional activity of the three TR isoforms fused to a Gal4 DBD was examined by mammalian one-hybrid analysis. As shown in Fig. 2, PGC-1 stimulated the transcriptional activation of all three Gal4-TR constructs to a similar level. This indicates that PGC-1 exhibits no preference for TR isoforms, at least within our experimental conditions. Unlike the effect of PGC-1 on TR-mediated reporter gene expression, coactivation of the transcriptional activity from the Gal4-TR constructs by PGC-1 was entirely ligand-dependent. However, in contrast to TR, coactivation of the transcriptional activity of Gal4-PPAR␥ by PGC-1 was ligand-independent, which is in agreement with a previous report (37). Taken together, these observations confirm that PGC-1 is a potent coactivator for TR action and suggest that PGC-1 coactivation of TR is influenced by the structure of the promoter with which TR interacts.
The Highly Conserved Glu 457 in AF-2 Is Not Necessary for PGC-1 Coactivation of TR-Biochemical and x-ray crystallographic studies have established that the AF-2 domain within helix 12 of TR and other nuclear receptors plays a critical role in mediating ligand-induced transcriptional activation (17,18). Mutational analysis has revealed that highly conserved residues such as Glu 457 and Leu 454 in this domain are very impor-tant for the recruitment of coactivators such as p160 proteins (51,52). To assess whether the PGC-1 effect on TR also depends on these highly conserved residues, we employed a TR␤1 E457A mutant fused to the Gal4 DBD and performed transient transfection analysis in HeLa cells. As expected, in the absence of T 3 , expression of wild-type Gal4-TR␤1 repressed basal transcription. Addition of T 3 resulted in a 5-fold induction of Gal4-TR␤1 activity (Fig. 3). The ligand-induced transcriptional activation of Gal4-TR␤1 was diminished ϳ3-fold by the E457A mutation when no coactivators were ectopically expressed in the cells (Fig. 3). The decreased Gal4-TR␤1 activity is likely due to impaired interaction with endogenous p160 or other coactivators. As observed earlier, a 20-fold increase in the liganddependent transcriptional activity of Gal4-TR␤1 was achieved when PGC-1 was coexpressed (Fig. 3). Surprisingly, the E457A mutant did not alter the ability of PGC-1 to stimulate the ligand-dependent transcriptional activity of Gal4-TR␤1 (Fig.  3). However, enhancement of Gal4-TR␤1 activity by overexpressing SRC-1 was severely impaired (Ͼ70%) by this E457A mutant (data not shown). Therefore, these data indicate that the requirement of critical amino acid residues in the TR AF-2 region for PGC-1 function is distinct from that seen with p160 coactivators.
Helix 1 and an Intact AF-2 Region Are Required for PGC-1stimulated Transactivation of TR-Initial domain analysis of PPAR␥ revealed that the interaction of PGC-1 with this receptor is ligand-independent and mediated by a region spanning part of the DBD and the hinge region (37). The AF-2 domain was suggested not to be involved in PGC-1 recruitment. However, studies with PPAR␣ and GR revealed that the effect of PGC-1 on these receptors is ligand-and AF-2-dependent (42)(43)(44). To determine the functional domain(s) responsible for PGC-1 coactivation of TR, a series of TR LBD deletion mutants fused to the Gal4 DBD were used to perform mammalian one-hybrid assays. As shown earlier, PGC-1 had no effect on the Gal4 DBD itself either in the presence or absence of T 3 . PGC-1 greatly enhanced the T 3 -dependent transcriptional activity of full-length TR␤1 fused to the Gal4 DBD (Fig. 4B). Surprisingly, the N-terminally truncated mutant containing only the intact TR␤ LBD (Gal4-TR216) not only increased its transcriptional activity without coexpression of PGC-1, but also exhibited a significant stimulation of the transcription by PGC-1 compared with the wild-type receptor (Fig. 4B). This unexpected finding indicates that the deleted region may have an inhibitory effect on the receptor activity modulated by PGC-1 or other coactivators (53). However, as shown in Fig. 4B,   FIG. 2. PGC-1 enhances ligand-induced transcriptional activation by Gal4-TR. Transfection and measurement of luciferase activity were carried out as described in the legend to Fig. 1, but with plasmids expressing Gal4-TR␤1, Gal4-TR␣1, Gal4-TR␤2, Gal4-PPAR␥ LBD, and human PGC-1 together with the luciferase reporter plasmid pG5-Luc containing five copies of the Gal4-binding site. All experiments were done in triplicate, and data are displayed as the means Ϯ S.E. of a single experiment, representative of at least three independent experiments.
FIG. 3. E457A point mutation in the AF-2 domain of TR␤1 has no effect on its transcriptional activity stimulated by PGC-1. Transfection and measurement of luciferase activity were carried out as described in the legend to Fig. 1, but with plasmids expressing Gal4-TR␤1 E457A and human PGC-1 together with the luciferase reporter plasmid pG5-Luc containing five copies of the Gal4-binding site. All experiments were done in triplicate, and data are displayed as the means Ϯ S.E. of a single experiment, representative of at least three independent experiments. the T 3 -activated PGC-1 enhancement was completely abolished by the deletion mutant Gal4-TR233, which lacks helix 1, but contains the remainder of the LBD (Fig. 4B). Similar results were obtained with the AF-2 deletion mutant in which fulllength TR␤1 was fused to the Gal4 DBD with the entire AF-2 deleted (Gal4-TR⌬AF-2) (Fig. 4B). These findings suggest that the PGC-1 effects depend on both helix 1 and the AF-2 domain. In addition, Gal4-TR224, a mutant construct similar to Gal4-TR233 except that it contains a partial helix 1, still retained a marginally ligand-dependent activity when PGC-1 was present (Fig. 4B). This further confirms that the intact helix 1 is necessary for PGC-1 function on TR. To ensure that the different effects of PGC-1 on TR transcriptional activation seen with these deletion constructs were not due to the altered expression level of the receptors, Western blot analysis using an anti-Gal4 DBD antibody was performed, which showed no significant protein expression (data not shown).
To determine whether the requirement of helix 1 for PGC-1 function on TR is specific to PGC-1, we cotransfected GRIP-1 along with these deletion mutants into HeLa cells. Similar to PGC-1, GRIP-1-mediated enhancement of TR transcriptional activation was dramatically increased with Gal4-TR216, but was entirely abolished with Gal4-TR233 or Gal4-TR⌬AF-2 (data not shown). The ability of GRIP-1 to enhance the activity of the receptor was also impaired by the deletion to amino acid 224 of TR␤1. Thus, the requirement of helix 1 for receptor coactivator function is unlikely to be PGC-1-specific.
We reasoned that the loss of the transcriptional activity of TR233 could be due to a general disruption of the conformation of the protein, resulting in the inability to bind ligand. Recently, using a newly developed assembly assay, Pissios et al. (54) provided evidence that helix 1 of TR plays an important role in stabilizing the receptor LBD. These investigators divided the TR LBD into two parts, one including helix 1 and the other corresponding to the remainder of the LBD. The presence of a ligand enables reconstitution of a functional LBD from these two fragments. To rule out the possibility of a general disruption of the protein structures, we performed this assembly assay to investigate whether coactivators can still function on TR reconstituted from two fragments. As shown in Fig. 5B, the T 3 -dependent transcriptional activation was restored when TR␤233 in pcDNA3 was coexpressed with the helix 1-containing Gal4-TR/C255 chimera (TR␤1 amino acids 92-255) (Fig.  5B, bar 4), whereas neither construct alone was functional in this assay system (data not shown). Notably, both PGC-1 and GRIP-1 significantly augmented (6 -7-fold increase) the activity of the reconstituted TR LBD in the presence of T 3 (bars 8 and 12). In contrast, the ligand-induced restoration and PGC-1and GRIP-1-mediated enhancement were completely abolished when the helix 1-truncated Gal4-TR/C222 chimera (TR␤1 amino acids 92-222) was used to carry out the above assembly assays (Fig. 5C) instead of Gal4-TR/C255. Thus, these results clearly show that the introduction of the helix 1 domain can restore the ligand-dependent transcriptional activity of TR233 in trans. Importantly, ligand-dependent transactivation and coactivation by PGC-1 and GRIP-1 occur on this reconstituted TR LBD, indicating that the inability of TR233 to transactivate is not due to a deficiency in ligand-binding activity.
Effect of PGC-1 on TR Correlates with the Functional Binding of PGC-1 to TR-Functional analysis has established that the effect of PGC-1 on TR relies on the TR LBD, particularly helix 1 and the AF-2 domain. An N-terminal region of PGC-1 possessing a single copy of an LXXLL motif was previously demonstrated to be required for ligand-dependent interaction with PPAR␣, GR, and estrogen receptor-␣ (42)(43)(44). We investigated whether these domains are also responsible for the interaction between PGC-1 and TR␤1. GST pull-down assays were initially carried out using a series of TR␤1 deletion mutants expressed as in vitro transcribed/translated 35 S-labeled proteins and a fragment of PGC-1 (amino acids 100 -411) containing the LXXLL motif expressed as a GST fusion protein. As shown previously, wild-type TR␤1 weakly interacted with GST-PGC-1 even without T 3 , and this interaction was greatly enhanced in the presence of T 3 (Fig. 6A). Similarly, this ligandinduced interaction with GST-PGC-1 was also observed when the truncated TR␤1 mutant TR216 (spanning the intact helix 1 and the remainder of the LBD) was used for GST pull-down experiments (Fig. 6B). In addition, as expected, the E457A mutant also showed the same pattern of ligand-dependent interaction with PGC-1 as did wild-type TR␤1 (Fig. 6C), consistent with our in vivo functional analysis with this mutant. In contrast, deletion of helix 1 in TR233 eliminated the ligand-dependent binding to GST-PGC-1 (Fig. 6B), as did the TR⌬AF-2 mutant (Fig. 6D). Because the earlier functional studies using the mammalian one-hybrid system had revealed that TR233 and TR⌬AF-2 were unable to be coactivated by PGC-1, it is reasonable that these constructs lose their ability to interact with PGC-1. Interestingly, the weak ligand-independent interactions between GST-PGC-1 and TR were retained with these FIG. 4. Mapping the region of TR␤1 required for PGC-1 coactivation. A, shown is a schematic representation of full-length human TR␤1 and its various N-terminal deletions used in this functional analysis. B, transfection and measurement of luciferase activity were carried out as described in the legend to Fig. 1, but with plasmids expressing Gal4-TR␤1, Gal4-TR216, Gal4-TR224, Gal4-TR233, Gal4-TR⌬AF-2, and human PGC-1 together with the luciferase reporter plasmid pG5-Luc containing five copies of the Gal4-binding site. All experiments were done in triplicate, and data are displayed as the means Ϯ S.E. of a single experiment, representative of at least three independent experiments. two mutants. These results suggest that helix 1 and the AF-2 domain of TR␤1 play a critical role in mediating ligand-induced interaction with PGC-1 and confirm that this ligand-dependent interaction accounts for the ligand-dependent PGC-1 coactivation of TR.
These in vitro GST pull-down data were then further verified by in vivo mammalian two-hybrid assays. As shown in Fig. 7, wild-type or E457A mutant TR␤1 fused in frame with the Gal4 DBD interacted with VP16-PGC-1-(100 -411) in a ligand-dependent fashion, which was reflected as the additive activity from the TR and VP16 activation domains. No such interactions were observed when the expression vectors expressing Gal4-TR233, Gal4-TRN (corresponding to amino acids 1-222), or Gal4-TR⌬AF-2 were cotransfected with VP16-PGC-1-(100 -411). These results confirm the importance of helix 1 and the AF-2 region of TR in mediating the interaction with PGC-1.
The LXXLL Motif Is Necessary for Interaction of PGC-1 with TR-The helical LXXLL motif present in p160 proteins and other coactivators has been defined as a critical element for efficient interaction with ligand-bound nuclear receptors (32). The role of the LXXLL motif in the PGC-1/TR interaction was initially examined with in vitro GST pull-down assays using similar GST-PGC-1 constructs with substitutions of three leucines for alanines in the LXXLL motif, thus destroying the amphipathic nature of this short helical domain. As shown in Fig. 6, the ligand-dependent interaction between PGC-1 and full-length TR␤1 as well as truncated TR␤216 was abolished by the PGC-1 AXXAL mutant. This PGC-1 mutant had no impact on the ligand-independent binding with these TR constructs. Furthermore, the importance of the PGC-1 LXXLL motif in medi-ating the ligand-induced interaction between PGC-1 and TR was confirmed by in vivo mammalian two-hybrid analysis. As demonstrated in Fig. 7, expression of wild-type VP16-PGC-1-(100 -411) elevated the T 3 -induced luciferase reporter activity by Gal4-TR␤1 another 2-fold. However, this enhancement of the reporter gene transcription by VP16-PGC-1 was abolished when the VP16-PGC-1 AXXAL mutant was introduced. Similar results were observed with the Gal4-TR E457A mutant. This clearly indicates that the LXXLL motif mutation impairs the ligand-dependent binding of TR to PGC-1, consistent with our findings with the in vitro GST pull-down assays. Thus, we conclude that the intact LXXLL motif present in PGC-1 is required for the ligand-dependent component of PGC-1 coactivation of TR.

DISCUSSION
Because nuclear receptors control a wide array of physiological processes (including growth, development, and metabolism) in response to a variety of stimuli, selective regulation of target gene expression appears to be critical for diverse nuclear receptor actions. The assembly of distinct coactivator complexes on the promoter-interacting nuclear receptor in response to cognate ligands and various physiological stimuli appears, at least in part, to be one of the powerful tools utilized by organisms to achieve this selectivity. PGC-1 exhibits a regulated expression pattern both in terms of a tissue-selective pattern of expression and in various physiological states and modulates a variety of nuclear receptor activities that are important for energy metabolism and adaptive thermogenesis. Therefore, understanding the mechanisms that regulate the FIG . 5. Effects of PGC-1 and GRIP-1 on the activity of the reconstituted TR␤ LBD. A, shown is a schematic representation of the various Gal4-TR␤ LBD constructs used in this assembly analysis. B and C, pcDNA3-TR233 was cotransfected along with either Gal4-TR/C255 or Gal4-TR/C222 into HeLa cells in the presence of PGC-1 or GRIP-1 as indicated. After transfection, the cells were treated with 10 Ϫ6 M T 3 for 24 h as indicated, prior to measurement of luciferase activity. All experiments were done in triplicate, and data are displayed as the means Ϯ S.E. of a single experiment, representative of at least three independent experiments. recruitment of PGC-1 by nuclear receptors may provide insight into the physiological roles of PGC-1.
In this study, we investigated the molecular mechanism by which PGC-1 modulates TR-mediated transcriptional activation. Our results show that ectopic expression of PGC-1 in HeLa or CV-1 cells results in enhancement of TR-mediated reporter gene activation in a ligand-dependent and ligandindependent manner, depending on the TRE context. Thus, these data first confirm the role of PGC-1 as a potent coactivator for TR and further imply that the promoter context with which TR interacts has an impact on the effect of PGC-1 on the activity of the receptor. In addition, we have also demonstrated that PGC-1 stimulation of TR activity in the context of Gal4-TR is exclusively ligand-dependent, whereas enhancement of the transcriptional activity of Gal4-PPAR␥ by PGC-1 is ligandindependent. These observations suggest that the conformation of different nuclear receptors may influence the effect of PGC-1 on the activity of the nuclear receptor.
Direct interaction with the AF-2 region of agonist-bound nuclear receptors is an initial committed step for coactivator function. The LXXLL motifs present within coactivators are usually required for this process (32). Previous reports have demonstrated that the ligand-induced interaction of PGC-1 with nuclear receptors, including PPAR␣, estrogen receptor-␣, and GR, is dependent on AF-2 and the LXXLL motif within PGC-1 (42)(43)(44). In contrast, the binding of PGC-1 to PPAR␥ is not influenced by the presence of ligand (37). Because the hinge region of the receptor mediates the PGC-1/PPAR␥ interaction, instead of AF-2, it is reasonable that this interaction is independent of ligand. However, the physiological relevance for this ligand-independent interaction still remains unclear. Using mammalian one-hybrid analysis, the TR LBD was initially defined to be necessary and sufficient for PGC-1 coactivation. We have further shown that the AF-2 domain of TR is indispensable for PGC-1 action. Using GST pull-down and mammalian two-hybrid assays, we further demonstrated that a large component of the PGC-1/TR interaction is ligand-dependent and that this ligand-dependent interaction requires both the intact AF-2 domain of the receptor and the LXXLL motif within PGC-1. These in vitro data are in agreement with the results from our in vivo functional analysis. Interestingly, the ligandindependent component of the PGC-1/TR interaction requires neither helix 1 nor the AF-2 domain. The previous observation that the stimulation of TR/retinoid X receptor-mediated ucp-1 gene expression in rat fibroblast cells by PGC-1 is exclusively dependent on thyroid hormone (37) is consistent with our observations. Thus, we conclude that, unlike PPAR␥, the effect of PGC-1 on TR is mediated through the receptor LBD, largely via the AF-2 domain.
Structural studies have revealed that the binding of ligand triggers a conformational change in the receptor that results in the repositioning of several ␣-helices, including helixes 3, 4, 5, and 12, within the TR LBD, thereby generating a hydrophobic groove for coactivator binding (55). Among other residues, a highly conserved lysine at the C terminus of helix 3 and a glutamate in helix 12 near this groove form a "charge clamp" with the LXXLL motif, which is critical for coactivator/TR interaction (18,55). Mutational analysis has revealed that mutation of these two residues severely impairs or abolishes the transcriptional activation mediated by TR, presumably because of the inability of these mutants to interact with coactivators such as the p160 proteins (52,55). However, our results from mammalian one-hybrid assays show that the point mutation E457A has no effect on coactivation of TR by PGC-1. In vitro protein/protein interaction assays confirmed that the ligand-dependent interaction between PGC-1 and TR is not influenced by the E457A mutation. These observations indicate that the interface mediating PGC-1/TR interaction appears to be different from that for p160 coactivators, at least in terms of the requirement for the charge clamp. It has been previously demonstrated that LXXLL motifs from various coactivators are not functionally equivalent for the binding of nuclear receptors (56). The sequences adjacent to the LXXLL core motif appear to play a critical role in determining receptor preference. Using a phage display approach, different types of LXXLL motifs have been grouped into at least three classes (56). Each class displays different preferences for various nuclear receptors. In contrast to the LXXLL motifs present in p160 coactivators, the FIG. 6. PGC-1 interaction with wild-type and mutant TR␤1 in vitro. A, the LXXLL motif in PGC-1 is required for ligand-induced interaction of PGC-1 with full-length TR␤1. Glutathione beads bound with Escherichia coli cell-expressed GST-PGC-1-(100 -411) or the GST-PGC-1 AXXAL mutant were incubated with in vitro translated 35 S-labeled full-length human TR␤1 for 1 h at room temperature in the presence (ϩ) or absence (Ϫ) of 10 Ϫ6 M T 3 . After washing extensively, the proteins bound on the beads were analyzed by SDS-PAGE and visualized by autoradiography. B, helix 1 of TR␤ is critical for the ligand-dependent interaction of PGC-1 with TR␤1. The GST pull-down assay was carried out as described for A, except that in vitro translated 35 S-labeled truncated TR216 and TR233 were used in this experiment. C, the E457A mutation has no effect on the ligand-dependent interaction of PGC-1 with TR␤1. The GST pull-down assay was carried out as described for A, except that the in vitro translated 35 S-labeled TR E457A mutant was used for this experiment. D, the AF-2 region is required for the liganddependent interaction of PGC-1 with TR␤1. The GST pull-down assay was carried out as described for A, except that in vitro translated 35 S-labeled TR⌬AF-2 was used for this experiment.

FIG. 7. PGC-1 interacts with TR␤1 in vivo.
Transfection and measurement of luciferase activity were carried out as described in the legend to Fig. 1, but with plasmids expressing VP16-PGC-1-(100 -411) and its mutant VP16-PGC-1 AXXAL along with Gal4 DBD-TR␤1 and its various mutants. All experiments were done in triplicate, and data are displayed as the means Ϯ S.E. of a single experiment, representative of at least three independent experiments. PGC-1 LXXLL motif belongs to the class III type (56). This may explain why the highly conserved charged residue Glu 457 in the TR␤1 AF-2 domain is not necessary for the PGC-1/receptor interaction.
Surprisingly, in the course of mapping the functional domain(s) within TR responsible for PGC-1 function, we found that the intact helix 1 of the TR LBD was also required for coactivation of TR transcriptional activity by PGC-1. Deletion of helix 1 within the receptor LBD completely abolished PGC-1-mediated coactivating activity. This unexpected observation initially raised the possibility of a specific role for helix 1 in PGC-1 action. However, it is known from crystallographic studies that this helix is folded tightly into the LBD regardless of the presence of agonist, and it is unlikely that this helix provides the surface for coactivator binding. Recently, Pissios et al. (54) reported an important role of helix 1 in stabilizing the TR LBD. They observed a ligand-dependent interaction between helix 1 and the remainder of the TR LBD and alteration of susceptibility to protease digestion with the receptor LBD lacking helix 1. In addition, they also showed that the triple point mutation (AHT 3 GGA) within helix 1 that was previously shown to disrupt the binding of corepressor to the receptor (15) impairs the interaction of helix 1 with the LBD. Interestingly, our functional studies also demonstrated that the partially truncated helix 1 mutant TR224 containing these AHT residues still retains the partial coactivating effect of PGC-1. These data suggest that the intact helix 1 has an impact on both coactivator and corepressor activities. Based on this information, we hypothesize that the observed requirement of helix 1 for the effect of PGC-1 would also apply to other coactivators. As expected, p160 protein-mediated coactivation of TR also requires the intact helix 1. Therefore, in agreement with the previous reports, helix 1 appears to provide a structural basis for TR LBD integrity, which is essential for coactivator and corepressor function. However, it is still possible that the deletion of helix 1 within the TR LBD may result in the inability of the mutated LBD to bind ligand, even though it is known that helix 1 is not directly involved in the binding of ligand. The data from the receptor assembly assay demonstrate that the coactivating activity of PGC-1 on the helix 1-deleted mutant LBD (TR233) can be restored by introducing in trans the intact helix 1 of the receptor in the presence of ligand. This suggests that the helix 1-deleted LBD protein retains ligand-binding activity because it is able to recruit helix 1 and subsequently PGC-1 in a liganddependent manner. Similar results were also obtained with another class of coactivator represented by GRIP-1. Thus, based on these results, the possibility that helix 1 deletion eliminates ligand-binding activity is unlikely.
In summary, our data imply PGC-1 as a bona fide TR coactivator. The results presented here demonstrate that, unlike PPAR␥, PGC-1 modulation of TR transcriptional activity depends on the AF-2 domain of the receptor. PGC-1 displays both ligand-dependent and ligand-independent components of interaction with TR that are dependent on the context of the DNAbinding activity of TR. The physical interaction of PGC-1 with TR in the ligand-dependent context requires the PGC-1 LXXLL motif and the intact AF-2 domain. The ligand-independent component of the PGC-1/TR interaction requires neither of these domains. Importantly, our evidence also shows that TR LBD integrity is critical for coactivator action. Thus, although both TR and PPAR␥ are involved in many physiological processes such as adaptive thermogenesis and hepatic gluconeogenesis, tissue-specific PGC-1 functions on TR via a mechanism distinct from that of PPAR␥. This difference may reflect the pivotal role that PGC-1 plays in the differential regulation of TR and PPAR␥ activities to achieve sequential and selective target gene expression.