| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
-Cofactor Interactions
Duke University Medical Center, Durham, North Carolina 27710
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(ERR) is an orphan nuclear receptor that does not appear to require a classical small molecule ligand to facilitate its interaction with coactivators and/or hormone response elements within target genes. Instead, the apo-receptor is capable of interacting in a constitutive manner with coactivators that stimulate transcription by acting as protein ligands. We have screened combinatorial phage libraries for peptides that selectively interact with ERR to probe the architecture of the ERR-coactivator pocket. In this manner, we have uncovered a fundamental difference in the mechanism by which this receptor interacts with peroxisome proliferator-activated receptor-
coactivator-1, as compared with members of the steroid receptor coactivator subfamily of coactivators. Our findings suggest that it may be possible to develop ERR ligands that exhibit different pharmacological activities as a consequence of their ability to differentially regulate coactivator recruitment. In addition, these findings have implications beyond ERR because they suggest that subtle alterations in the structure of the activation function-2 pocket within any nuclear receptor may enable differential recruitment of coactivators, an observation of notable pharmaceutical importance. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(ERR), a member of the nuclear receptor (NR) superfamily of transcription factors, was identified because of its high degree of sequence similarity to the canonical estrogen receptors (ER and ERß) (1). Although the transcriptional activity of ERR is not regulated directly by estrogens, it has been demonstrated in a convincing manner that it can influence some aspects of ER-mediated signal transduction. This is not surprising in light of the observation that the P-box regions of the DNA binding domains (which determine DNA specificity) are conserved in both the ERs and ERR, indicating that they likely recognize similar DNA response elements (2). Indeed, it appears that within cells and in the context of natural promoters, most estrogen response elements (EREs) can function as ERREs (ERR-responsive elements) but that only a subset of ERREs function as EREs (3, 4). A global analysis of the ER target genes, coregulated by ERR, has not yet been performed. It has been demonstrated in an empirical manner, however, that the established ER targets, pS2, lactoferrin, and osteopontin, are also regulated by ERR (5, 6, 7). Taken together, these data provide significant evidence of ER/ERR cross talk, although the extent of this interrelationship and its physiological implications remain to be determined.
In recent years, two independent studies have demonstrated that ERR expression is a negative prognostic factor for disease-free survival in breast cancer (8, 9). In the larger of these studies, it was demonstrated that ERR expression in greater than 10% of malignant cells was associated with a 20% decrease in overall disease-free survival at 13 yr (relative risk = 5.1). From a biological perspective, this is consistent with the observation that ERR and HER2 expression were positively associated in advanced, tamoxifen-resistant, and/or ER-negative tumors. The link between ERR and cancer has also been suggested by additional studies in ovarian, colorectal, and prostate cancers (10, 11, 12). Although the role of ERR in these cancers remains to be determined, its expression and association with a negative prognosis suggest that it may be a useful target for chemotherapeutics.
In addition to its actions in the ER signal transduction pathway and in cancer, recent data establish a significant role for ERR in the regulation of metabolic function. This receptor is expressed in nearly all organs at some level although it is most highly expressed in kidney, heart, cerebellum, intestine, and skeletal muscle, tissues that preferentially recognize fatty acids as energy sources (D. Mangelsdorf, personal communication). The function of ERR as a metabolic regulator is supported by the finding of impaired fat metabolism and absorption in ERR –/– mice (13). The elevated expression of this receptor subtype in exercising muscle and in fasting liver also suggests a role for this receptor in ß-oxidation of fatty acids (13). Indeed, the findings of several elegant studies have revealed that ERR is involved in the transcriptional regulation of genes required for mitochondrial biogenesis, oxidative phosphorylation, and fatty acid oxidation (14, 15, 16, 17, 18). Interestingly, in most of these studies, ERR is the downstream effector of its interacting partner PGC-1 (PPAR coactivator 1), a cofactor the expression of which is low in cells until it is induced by fasting or other metabolic stresses (19). PGC-1ß, a related cofactor, may have similar functions under certain circumstances although its expression level is not as acutely regulated by different energy demands (18, 20, 21). Interest in the ERR-PGC-1 regulatory axis was heightened by the observation that there is a decrease in both PGC-1 and -ß in the skeletal muscle of diabetic and obese patients (22, 23, 24, 25). In addition to the regulation of metabolism, ERR has been shown to have a role in the establishment and maintenance of bone through its regulatory actions in osteoclasts (26, 27, 28, 29). Thus, although antagonists of this receptor may have utility in cancer, there appears to be an unmet medical need for ERR agonists for use in the treatment of metabolic diseases.
As yet, ERR has not been shown to interact with any physiologically relevant small molecules leading to the hypothesis that it manifests constitutive activity (30, 31, 32). Crystallographic analysis of apo-ERR and comparison with other agonist-activated NRs indicate that the ERR coactivator binding pocket adopts a transcriptionally active conformation in the absence of ligand (33). Thus, it appears likely that the activity of this receptor is regulated by the relative and absolute abundance of coactivators and their ability to function as protein-ligands (6, 16, 34, 35). This hypothesis is supported by the finding that ERR transcriptional activity, although constitutive, is relatively weak until a coactivator, such as PGC-1, is coexpressed in cells.
Given what we know of the biology of ERR and the mechanism(s) by which its transcriptional activity is manifest in cells, we sought to explore the possibility of using combinatorial phage display to 1) identify peptide antagonists that would operate by competitively displacing coactivators and 2) map the architecture of the ERR coactivator pocket to determine whether small molecules could be used to regulate the differential recruitment of coactivators to the receptor and thus stimulate different biological responses. The results of these studies reveal that 1) although the activation function (AF)-2 within ERR is critical for transcriptional activity, its preferred coactivators do not bind to this pocket in an identical manner and 2) peptides that disrupt ERR-coactivator interactions effectively inhibit ERR activity in cells. These findings provide a strong base upon which to embark on a discovery program aimed at the development of ERR modulators.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Activation by Different Coactivators does not appear to require an endogenous small molecule ligand to enable its interaction with either coactivators or DNA (32, 36). However, several studies have noted that the constitutive activity of ERR is lower than that of the other ERR isoforms (ERRß or ERR) and that its activity is strongly enhanced by the coexpression of coactivators such as PGC-1 or steroid receptor coactivator (SRC) family members (6, 37). To evaluate more broadly the significance of the noted constitutive activity, we performed transient transfection assays to assess the relative activity of PGC-1 and SRC2 on ERR-mediated transcriptional activity on several promoters. HeLa cells were cotransfected with an ERR expression plasmid (or a relevant control plasmid) along with an expression plasmid for PGC-1, SRC2, or its cognate control vector. One reporter construct contained three copies of the vitellogenin ERE fused upstream of a TATA promoter and luciferase reporter gene (Fig. 1
; 3xERE-TATA-Luciferase). The others included the ERE/ERRE containing regions of the pS2, short heterodimer partner (SHP), or cytochrome P-450 (CYP)19 promoter sequences fused with the luciferase reporter gene (Fig. 1; pS2-luciferase, SHP-luciferase, and CYP19-luciferase, respectively). In the absence of an exogenous ligand, ERR transcriptional activity can be increased by the addition of coactivators. However, clearly the coactivation potential of each coactivator differed on individual promoters. Whereas PGC-1 can potentiate ERR activity 4- to 6-fold on the pS2, CYP19, and 3xERE-TATA promoters, PGC-1 appears to enhance ERR activity to a greater extent on the SHP promoter. By contrast, SRC2 is a weak coactivator on the SHP, CYP19, and 3xERE-TATA promoters (1.5–2 fold activity); however, its activity is as strong as PGC-1 on the pS2 promoter. This suggests that the promoter context may determine the degree of coactivation by an individual coactivator perhaps by influencing receptor conformation. These data are consistent with previous studies in our laboratory that have shown that the nature of the ERE sequence influences the recruitment of coactivators to the ER at target gene promoters (38). Despite these differences, it is clear that the activity of ERR is strongly stimulated by the coexpression of coactivators without the addition of a small molecule ligand.
|
and SRC-2 Are Necessary and Sufficient for ERR Interaction also interacts with its coactivators through LXXLL-containing regions, the 19-amino acid sequence centered around each LXXLL motif was cloned into a pM vector to create a Gal4DNA binding domain (Gal4DBD)-fusion protein (Fig. 2, A and B). Consistent with previous reports, ERR can efficiently interact with either PGC-1 L2 or L3 (Fig. 2C), whereas there is no significant interaction with PGC-1 L1 (34, 37, 40). Of the three LXXLL regions within SRC2 that have the potential to interact with NRs, ERR interacts best with SRC2 L3 but can also bind to a lesser extent to SRC2 L1 and SRC2 L2 (Fig. 2D). Therefore, we conclude that ERR can interact with PGC-1 L2 and L3 and SRC2 L1-L3. To determine whether the LXXLL motifs were critical for interaction with ERR, we mutated the leucines to alanines within the NRID in each of the coactivators and tested their coactivation potential (Fig. 2, E and F). Consistent with the above data, mutation of the NRIDs disrupted the coactivation potential of each coactivator, suggesting that these regions are the critical sites of interaction with the receptor. We conclude therefore that the LXXLL motifs within the coactivators under study are both necessary and sufficient for ERR interaction.
|
-Binding Peptides Using Phage Display with a view to 1) gaining a better understanding of the nature of the LXXLL-interaction surface(s) of this receptor and 2) developing peptide antagonists that function by disrupting coactivator binding. Using combinatorial phage display technology, we screened seven phage libraries expressing random or semirandom peptides of 13–19 amino acids in length for ERR interacting peptides (Fig. 3A). All of the phage that bound to ERR were amplified, and the amino acid sequences of the expressed peptides were determined by DNA sequencing.
|
in intact cells. To perform this assay, each peptide was expressed as a Gal4DBD-fusion protein. Their interactions with full-length ERR fused to a VP16 activation domain (VP16-ERR) were assessed using this assay (data not shown). Representative examples of peptides that were isolated in the screen against ERR are shown in Fig. 3B
. The majority of peptides were isolated from the LXXLL library. Of significant interest is that all but one of the peptides that were retrieved from the non-LXXLL phage libraries (6Y6, 6N6, SS, etc.; see Fig. 3A) contained an LXXLL motif, suggesting that this is the dominant motif used by coactivators to interact with ERR. A significant number of validated coactivators contain at least one LXXLL motif within the interaction domain. This motif mediates an interaction between the coactivator and helices 3, 5, and 12 of the NR (39, 43). That the peptides all contain an LXXLL motif is consistent with the data in Fig. 2 suggesting that this sequence is critical for coactivator-ERR interactions. Furthermore, this indicates that the peptides likely interact with the coactivator binding pocket of the receptor.
Previous efforts at classifying NR-interacting peptides and interaction domains of the common coactivators have identified four primary sequence clusters (41, 44). An analysis of the sequences flanking the core LXXLL motif of peptides identified in our screen revealed that all, except for Class IV, were represented (Fig. 3B; Class IV sequences included for comparison). Interestingly, a majority of the peptides identified (27 of 40 peptides) contain a serine in the –2 position similar to PGC-1 L2 and L3, whereas no peptides were found that were similar to the LXXLL motifs of SRC2. This suggests that ERR has a preference for PGC-1-like LXXLL motifs.
In addition to peptides assignable to previously defined classes, we identified a new class of peptides that contain a glutamic acid in the –1 position (named Class V). This is a novel class of peptides that are similar to the PGC-1 L3 motif, which does not interact with other NRs (34, 37). This suggested that these peptides may also be selective for the ERRs. To test this hypothesis, we performed mammalian two-hybrid assays examining the ability of these peptides to interact with the ERs. A negative interaction was defined as less than 5% of the level of interaction with the positive control [a previously identified ER and ß interacting peptide (41)]. Results are tabulated in Fig. 3B. The data demonstrate in a striking manner that the Class V peptides do not interact with either ER or -ß.
LXXLL-Containing Peptides Inhibit ERR Transcriptional Activity
Previous work in our laboratory has shown that peptides such as these can be used to efficiently inhibit ER and ERß transcriptional activity (41, 42, 45). The similarity of the peptides to the NRIDs of known coactivators suggested that these peptides interact with the coactivator binding pocket of the receptor. To test whether the ERR peptides would act as antagonists of receptor activity, transcriptional interference assays were performed in which mammalian cells were transfected with the luciferase reporter construct, 3xERE-TATA-luciferase, an expression plasmid for ERR and either an empty Gal4DBD vector (pM) or peptide-Gal4DBD fusions. Each of the peptides tested were capable of inhibiting ERR activity to varying degrees (Fig. 4A and data not shown). Peptides L3-09, L3-28, and L3-80 were the most effective inhibitors reducing the transcriptional activity of ERR to basal levels. Coexpression of the peptides similarly resulted in the inhibition of transcriptional activation by ERRß and ERR (data not shown). Under the conditions of this assay, we noted that the transcriptional activity of ER or ERß was unaffected by overexpression of individual ERR peptides (representative peptides shown in Fig. 4B), correlating with the interaction data presented in Fig. 3 and confirming that peptide-receptor interaction is important for inhibition by these peptides.
|
-mediated transcription both on the 3xERE-TATA-luciferase and pS2-luciferase reporter constructs (Fig. 4, C and D, respectively). Surprisingly, in the presence of the peptide, coactivation by PGC-1 remained intact. Similar results were obtained with other peptides (data not shown).
ERR Shows a Preference for the NRID of PGC-1
The ability of ERR-interacting peptides to inhibit SRC2, but not PGC-1, coactivation and the observation that PGC-1 is a more effective ERR coactivator suggested a difference in the receptor’s affinity for individual coactivators. To test this hypothesis, we developed a modified phage ELISA to compare the interaction of ERR or ERß with the coactivator NR boxes. Plates (96-well) were coated with neutravidin, biotinylated oligonucleotides containing an ERE (derived from the vitellogenin promoter), and a saturating concentration of purified ERR or ERß. To each well, increasing concentrations of T7 phage expressing either the NR box region of PGC-1 or SRC2 were incubated with the protein targets. Figure 5 shows the relative binding affinities of SRC2 and PGC-1 to ERR (panel A) and ERß (panel B). This assay clearly shows that the affinity of ERR for the PGC-1-NRID is greater than its affinity for the SRC2-NRID. This binding profile is in striking contrast to that obtained for ERß, where the fragments of PGC-1 and SRC2 have relatively similar binding affinities.
|
peptide antagonists, we have shown previously that placing two copies of the peptide in tandem enhances the ability of the peptide to inhibit receptor activity (41). To determine whether two copies of the peptide could disrupt PGC-1 coactivation of ERR, we created a tandem peptide in which the Gal4DBD is fused to two copies of the peptide separated by the sequence from PGC-1 that lies between NR box 2 and 3 (pM-2xL3-09). In transactivation assays, cotransfection of the pM-2xL3-09 dramatically inhibited potentiation of ERR by PGC-1 (Fig. 5C) indicating that this approach allows the peptides to be used as potent antagonists of coactivator-enhanced ERR activity. These 2x peptides retained their ERR selectivity and did not interact with or inhibit the transcriptional activity when tested on ER (data not shown).
Structural Requirements within the Charge Clamp for ERR-Cofactor Interactions Are Distinct
The observed differences in apparent affinity suggest that PGC-1 and SRC2 do not interact with the coactivator binding pocket of ERR in the same manner. To define the factors that influence coactivator recruitment to ERR, we made mutations in the AF-2 of ERR and determined their effect on cofactor-enhanced transcriptional activity (Fig. 6A; protein expression was unchanged by the mutations and is shown in Fig. 6B). HeLa cells were transfected with a 3xERE-TATA-luciferase reporter plasmid, a ß-Gal normalization vector, expression plasmids for a control vector (pcDNA3), pcDNA3-ERR or ERR mutants, and either a vector control plasmid, PGC-1, or SRC2. Truncation of ERR by removal of helix 12 (ERR 
H12) completely abolished the ability of either coactivator to enhance ERR transcriptional activity (compare ERR wt and H12 in Fig. 6, C and D). This result suggests that helix 12 is a region critical for interaction and that there are no autonomous secondary binding sites (i.e. AF-1 domain) for these cofactors.
|
affect its ability to potentiate coactivators, we made the corresponding point mutations in helix 12 of ERR (K412N, E416Q, and E419Q, named ERR-3x; Fig. 6A) that neutralize the charge (40, 41, 46). Unlike the truncation of helix 12, which blocked the ability of either cofactor to bind ERR, neutralization of the charge in helix 12 resulted in the differentiation between PGC-1 and SRC2 (compare ERR-3x; Fig. 6, C and D). These findings are consistent with studies from our laboratory and others showing that the integrity of the charge clamp is not required for PGC-1 to bind to PPAR, ER, or TR (40, 41, 46, 47).
To further define the residues governing the interaction between PGC-1 and ERR, we also mutated the other putative component of the ERR charge clamp (K244, as determined by alignment with ER). The effect of this mutation on ERR transcriptional activity was tested in coactivation assays (Fig. 6, C and D). Once again, mutation of one component of the receptor’s charge clamp did not disturb PGC-1’s ability to coactivate the receptor, whereas SRC2-mediated enhancement was prevented. However, when both components of the charge clamp were mutated (Fig. 6, C and D; ERR K244A/3x), neither coactivator was able to function to increase transcription.
A valine in helix 5 of ER (V380) has been found to contribute to the coactivator binding pocket formed upon ligand binding (48, 49). Alignment with ER shows that a methionine occupies the analogous position in ERR. Mutation of this residue to leucine modestly enhances coactivation by PGC-1 (Fig. 6E) but coactivation by SRC2 is diminished (Fig. 6F). However, combining the M258L and 3x mutations abolishes the ability of SRC2, but not PGC-1, to transactivate the receptor (Fig. 6, E and F). Thus, we conclude that both PGC-1 and SRC2, although interacting with the AF-2 domain of ERR, do so by distinct mechanisms.
A Serine Upstream of the LXXLL Motif Influences the Charge Clamp Requirement for Coactivator Binding
Similar to the ERR peptides selected by phage display, the PGC-1 NR box sequences with which ERR preferentially interacts have a conserved serine in the –2 position relative to the first leucine of the LXXLL motif. However, there are no serines in this position in any of the SRC2 NR boxes. In an analysis of PPAR binding to PGC-1, Wu et al. (46) show that mutation of the S140 (serine in the –2 position of PGC-1 L2) reduces the ability of the coactivator to bind a PPAR mutant in which the charge clamp has been disrupted. This serine may also determine whether the charge clamp is required for interactions with ERR. Testing of this hypothesis is complicated by the fact that PGC-1 can bind ERR through either L2 or L3, and mutation of either motif individually does not diminish the coactivator’s ability to enhance ERR activity (data not shown and Ref. 34). Therefore, it is only necessary to have one of the LXXLL motifs for PGC-1 to interact with ERR. To study the effect of mutating the serine of an individual motif, we first mutated the leucines of L3 to alanines (PGC-1 L3M in Fig. 7A). The PGC-1 L3M mutant is still able to coactivate ERR 3x through an interaction with L2. However, mutation of the serine (S140 upstream of L2) to alanine abolished this ability (PGC-1 S140A/L3M in Fig. 7B), indicating that the serine is important for maintaining the interaction between the L2 of PGC-1 and ERR in the absence of the critical residues in the charge clamp. The expression levels of PGC-1 L3M and PGC-1 S140A/L3M were comparable (data not shown). Interestingly, we have shown that this serine is not required for coactivation of wild-type ERR (data not shown), suggesting that although it may contribute to the ERR-PGC-1 binding surface, it is not necessary.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-selective peptide antagonists. One approach to inactivate ERR is to formulate ligands that destabilize the active conformation by shifting helix 12 away from its natural position. This appears to be the mechanism of action of diethylstilbesterol on ERR and (33, 50). The recently described antagonist XCT790 likely functions in an analogous manner, although its utility as a modulator of ERR is limited by its cross-reactivity with PPAR and by its cellular toxicity at the concentrations required to observe an inhibitory response. Thus, there is an unmet need for reagents with which to regulate ERR function. In light of this observation, we have taken a different approach and identified peptide inhibitors of ERR activity that will likely be useful in the study of ERR biology. Importantly, one class of peptides (Class V) selectively interacted with ERR but not with either ER. Thus, although the leucines of the LXXLL motifs are the primary determinants of the interaction with the coactivation binding pockets of NRs (39), the specificity and selectivity of individual NR boxes are modulated by sequences surrounding the LXXLL motif (41, 42, 51, 52, 53). The distinguishing feature of this novel class is the conservation of a glutamic acid residue in the –1 position and an arginine in the +2 position (second X of the LXXLL). A similar pair of residues is present in the PGC-1 L3 motif and is observed in the crystal structure with ERR-ligand-binding domain (LBD) to form a salt bridge (33). To examine these residues in the context of our phage-derived peptides, the side chains of L3-09 were placed onto the peptide of the PGC-1-ERR-LBD complex using computational modeling (Fig. 8). We found that a salt bridge between the glutamic acid and the arginine may also form in the peptide. This salt bridge could act to stabilize the helical structure of L3-09. However, the position of the glutamic acid does not explain the peptide’s specificity. ERR and ER were superimposed over all backbone atoms [root mean square (rms) = 1.39Å] to compare structural differences that could impact peptide binding and alter specificity. Helices 3, 5, and 12 of both receptors form the interaction surface with the ERR-LBD peptide and generally overlay closely in the two structures, suggesting that the peptides are likely binding in a similar location. However, at the helix 5 interface, L350 of ERR makes potential van der Waals contacts with the tryptophan at the +1 position (first x of LxxLL) of the peptide. In the ER, L372 may cause a steric clash with this tryptophan. Analysis of peptides with substitutions at the +1 position and at the putative salt bridge should further clarify the determinants of binding specificity and will be the goals of future studies.
|
interacts with individual coactivators. Through analysis of the regions within ERR and its coactivators that allow for their interaction, we identified the AF2 domain of ERR as an essential component of the binding surface for all currently known ERR coactivators. Whereas the N terminus of ERR contains an AF-1 domain that can bind to two cofactors, proline-rich nuclear receptor coregulatory protein and transducine-like enhancer of split/Groucho-related gene product 1 (54), no AF1 has been identified in ERR. Our data suggest that the AF-2 domain is the primary site of interaction for the coactivators tested, although it does not rule out the possibility that other coactivators may be found that will bind other regions of the receptor.
Mutations of the hydrophobic cleft of ERR differentially affect coactivator recruitment. The interaction of ERR with SRC family members, but not PGC-1, is disrupted by mutations of key charge clamp residues K244, E416, K412, and E419 on the top and bottom of the clamp, respectively. Instead, it is necessary to mutate both ends of the clamp to prevent the interaction between ERR and PGC-1. These data are consistent with several studies that determined that the integrity of the charge clamp is not required for PGC-1 recruitment to other receptors (40, 41, 46, 47), indicating that our findings likely extend beyond the ERR subfamily of receptors. The pattern of interaction (dependence vs. independence on the charge clamp) was consistent among members of families of coactivators. For example, as with SRC2, the interaction between ERR and SRC1 and -3 was disrupted by perturbations of the charge clamp, whereas interactions with PGC-1ß (like PGC-1) were not (data not shown).
In this study, we also identified M258 of ERR as a significant contributor to the hydrophobic surface important for coactivator binding. Importantly, SRC-mediated coactivation was prevented by this mutation, whereas there was no effect on transactivation by PGC-1. The recently solved crystal structure of ERR with a peptide of PGC-1 L3 docked in the hydrophobic cleft suggests one explanation for why the double mutation is necessary to interfere with PGC-1 binding (33). This study identified a number of ERR residues that participate in van der Waals contacts with the leucines of PGC-1 L3; among them are K244, M258, and E416 [referred to as K340, M354, and E512, respectively, by Kallen et al. (33); we have adjusted the amino acid numbering to correspond with NCBI accession number NP_004442 and to be consistent with the numbering throughout this manuscript]. M258 can form contacts with both L210 and L211, whereas E416 and K244 form contacts with L210 and L214 of PGC-1, respectively (33). The positions of K244, M258, and E416 suggest that they stabilize the interaction with the LXXLL motif on three sides. Although we cannot rule out that other contacts are also important for the formation of the coactivator-receptor complex, our data show that mutation of residues K244 or E416, K412, and E419 is sufficient to perturb the interaction between ERR and PGC-1. Whether a double mutation consisting of K244 and M258 would also affect PGC-1 binding has not yet been tested.
The position of M258 seems particularly suited to form contacts with the LXXLL of PGC-1. The side chains of the two leucines (L210 and L211 of PGC-1 L3) pack around the side chain of the methionine (M258 of ERR). Kallen et al. (33) also suggest that the replacement of L211 with a histidine (as in the LXXLL of SRC1; ILHRLLQE) would result in less favorable interactions. This may be one explanation for the increased sensitivity of the SRC-ERR complex to the mutations evaluated in this study.
Sequence analysis of the NR boxes that interact with ERR suggested that a serine in the –2 position relative to the first leucine of an LXXLL may determine whether the coactivator can bypass the requirement of the charge clamp for binding to ERR. Mutation of S140 (in the background of PGC-1 L3M) abolished its ability to coactivate ERR 3x, suggesting that this residue is involved in making other contacts to stabilize the interaction when the glutamic acid portion of the charge clamp of ERR is absent. Consistent with our data, peptides with a conserved serine in the –2 position have been shown to maintain interactions with a mutant ER in which the charge clamp has been disrupted (41). In addition, mutation of PGC-1 S140A attenuated its ability to coactivate a charge clamp mutant of PPAR (46). However, although the serine appears necessary for maintaining interactions in the absence of the charge clamp, by itself it is not sufficient. Short peptides containing both the LXXLL motif and the –2 serine were unable to interact to the same degree with ERR-3x as with the wild-type receptor (data not shown). Overall, these data support the hypothesis that sequences flanking the LXXLL motif influence the NR binding capabilities of the coactivator and may provide an approach to develop small ligands to modulate receptor-cofactor interactions.
The cellular response to individual pharmaceuticals is determined by ligand-specific conformational changes that regulate cofactor recruitment along with tissue-type differences in cofactor expression (55). As an example, different PPAR ligands were shown to affect whether or not the charge clamp is required for its receptor to bind PGC-1 (46), suggesting that the coactivator binding capabilities of the receptor may be modulated pharmacologically. PGC-1 and ERR together induce oxidative phosphorylation and fatty acid ß-oxidation in tissues where they are coexpressed. Patients with mitochondrial dystrophies and diabetes may benefit from a drug that enhances the interaction between PGC-1 and ERR in selected tissues. Because of the tissue-restricted expression pattern of PGC-1 (compared with SRC2), it may be possible to develop pharmacological drugs that preferentially allow PGC-1/ERR interactions in targeted tissues. ERR activity can be selectively modulated by small compounds (56, 57), and our data suggest that it may be possible to develop selective ERR modulators, that facilitate the recruitment of individual cofactors through subtle conformational changes.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(58), pRST7-ERß (59), pRST7-ER and the 3xERE-TATA-luciferase reporter construct (60), pCXX-SRC2 L1L2L3M (also named GRIP1 L1L2L3M) and pcDNA3-PGC-1 L2L3M (61), pM-SRC2 NRID and 5xGal4-luciferase plasmids (41). The pCXX-SRC2 plasmid was generated by excising the cDNA for the full-length SRC2 from the HA-SRC2 plasmid (gift of M. R. Stallcup, University of Southern California, Los Angeles, CA) and subcloning into the pCXX vector, a derivative of the pCMX vector (Stratagene, La Jolla, CA) in which the two XbaI sites and intervening base pairs were removed. The following plasmids were gifts: flag-PGC-1 (B. M. Spiegelman, Dana Farber Cancer Institute, Boston, MA), pS2-luciferase (V. Giguere, McGill University, Montreal, Quebec, Canada), SHP-luciferase (S. A. Kliewer, University of Texas Southwestern Medical Center, Dallas, TX), and CYP19-luciferase (E. Simpson, Prince Henry’s Institute of Medical Research, Victoria, Australia). The cytomegalovirus (CMV)ß-Gal plasmid was purchased from Clontech Laboratories, Inc. (Palo Alto, CA) and the pBSII vector was purchased from Stratagene. The coactivator Gal4DBD-LXXLL motifs were generated by ligating annealed oligonucleotides corresponding to the sequence surrounding the LXXLL motif (X7LXXLLX7) into the pM vector (Clontech, Mountain View, CA). The following mutants were generated using the QuikChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA): (templates listed in parentheses) pcDNA3-ERR H12, pcDNA3-ERR 3x, pcDNA3-ERR K244A, and pcDNA3-ERR M258R (wild-type pcDNA3-ERR as template), pcDNA3-ERR KA/3x and pcDNA3-ERR MR/3x (pcDNA3-ERR 3x), flag-PGC-1 L3M (wild-ype flag-PGC-1), and pcDNA3-PGC-1 S140A L3M (pcDNA3-PGC-1 L3M). The corresponding pM or VP16 vectors were generated by excising the full-length cDNA and subcloning into the appropriate sites within the vector. All PCR products were sequenced to ensure the fidelity of the resulting constructs.
Phage ELISA
The phage binding assays were carried out in 96-well plates (Costar, Acton, MA) as previously described (41). Briefly, the wells were first coated with 40 µg/well neutravidin in 100 µl 100 mM NaHCO3, pH 8.5, and then blocked with a 2% milk-100 mM NaHCO3 solution. Synthetic biotinylated double-stranded DNA (4 pmol), containing an ERE sequence, was added to the well. Purified recombinant biotinylated-ERR (1 µg) was added to the DNA. In addition, there were two negative control wells, one incubated with 0.25 µg powdered milk in 100 µl 100 mM NaHCO3 and the other with 2.5 µg BSA in 100 µl 100 mM NaHCO3. Wells were blocked with 150 µl 2% milk in PBS for 1 h at room temperature (RT). Wells were washed five times with PBST (PBS + 0.1% Tween 20) followed by 2 h incubation with a range of phage concentrations as delineated in the figure legend followed by a PBST wash. Diluted anti T7-horseradish peroxidase (HRP) antibody (Novagen, Madison, WI) was incubated in the wells for 1 h at RT followed by a PBST wash. Bound antibody-enzyme conjugate was detected by 29, 29-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid in the presence of 0.05% H2O2, and the color change was measured at 405 nm on a plate reader (Multiskan MS, Labsystems, Inc., Marlboro, MA).
Mammalian Cell Culture and Transfections
HeLa (human cervical carcinoma) and HepG2 (hepatoma) cells were obtained from American Type Culture Collection (Manassas, VA) and maintained in MEM (Invitrogen, Carlsbad, CA) supplemented with 8% fetal bovine serum (Hyclone Laboratories, Inc., Logan, UT), 0.1 mM nonessential amino acids, and 1 mM sodium pyruvate (Invitrogen) in a 37 C incubator with 5% CO2. Before use, culture flasks and 24-well plates for HepG2 cells were coated with 0.1% gelatin for 10 min at 25 C. HeLa cells were used for transactivation assays. HepG2 and HeLa cells were used for mammalian two-hybrid assays with similar results obtained from both cell types. For transient transfections, cells were split into 24-well plates 24 h before transfection. Lipofectin (Invitrogen)-mediated transfection has been described in detail previously (62). Briefly, before transfection, the media were replaced with phenol-free MEM (Invitrogen) containing 8% charcoal-stripped serum (Hyclone Laboratories) and 0.1 mM nonessential amino acids and 1 mM sodium pyruvate (Invitrogen). A DNA-Lipofectin mixture containing a total of 3000 ng of plasmid for each triplicate sample was added to the cells. Receptor ligands were added to the cells 14–16 h before the assay. The plasmid amounts used in each experiment are listed in the figure legends. Luciferase and ß-galactosidase activities were measured as described elsewhere (62). Luminescence was measured on a PerkinElmer (Wellesley, MA) Fusion luminometer. Results are expressed as normalized luciferase activity (normalized with ß-Gal for transfection efficiency) ± SEM per triplicate sample of cells.
Generation of an ERR Antibody
The ERR LBD, consisting of amino acids 193–423 of ERR, was cloned into the EcoRI and NotI sites of pGEX6P1, a glutathione-S-transferase (GST)-fusion vector (Amersham Pharmacia Biotech, Piscataway, NJ). The pGEX-ERR LBD fusion vector was transformed into Escherichia coli BL21 cells and prepared according to the manufacturer’s instructions. In brief, pGEX-ERR LBD-transformed BL21 cells were grown to OD600 = 0.6 at 37 C, 100 mM isopropyl-ß-D-thiogalactopyranoside was added to cells to a final concentration of 0.1 mM, and the cells were allowed to grow for another 18 h in a 37 C shaking incubator. The cells were collected by centrifuging at 7700 x g for 10 min at 4 C. Cells were resuspended in 50 ml ice-cold PBS per ml of culture and lysed by sonication (five pulses x 20 sec each). Triton X-100 was added to a final concentration of 1% and mixed by gentle rocking for 30 min at 4 C. The lysate was centrifuged at 12,000 x g for 10 min at 4 C, and the soluble fraction was collected. The protein was purified with Glutathione Sepharose 4B beads (Amersham Pharmacia Biotech) using affinity chromatography in a batch purification protocol. Beads were prepared according to the manufacturer’s instructions and incubated with the lysate for 2 h at 4 C with gentle rocking. Beads were then washed three times with ice-cold 1x PBS. The GST-ERR LBD protein was eluted from the beads using an elution buffer containing 50 mM Tris-HCl and 10 mM reduced glutathione, pH 8.0. The GST tag was removed by using Prescission protease (Amersham Pharmacia Biotech) according to the manufacturer’s directions, followed by an incubation with Glutathione Sepharose 4B beads to remove the GST. The ERR-LBD protein was sent to D. Edwards (University of Colorado Health Sciences Center, Denver, CO) for generation of the mouse monoclonal antibody. Several antibody clones were tested for their ability to recognize recombinant ERR. Clone no. 1148 was selected for its ability to recognize both endogenous and recombinant ERR and low background.
Production and Purification of Recombinant Estrogen Receptor-Related Receptor
The full-length ERR cDNA was subcloned into the EcoRI and XbaI sites (5'- and 3'-ends, respectively) of the pFASTbac-HTa vector (Invitrogen) to generate an in-frame fusion of ERR with a 6x histidine (His) tag separated by a recombinant tobacco etch virus protease cleavage site. Baculovirus and recombinant protein were generated in Spodoptera frugiperda (Sf9) cells (Invitrogen) and purified using Ni-nitrilotriacetic acid beads (Invitrogen) according to the manufacturer’s protocol. Sf9 cells were maintained in serum-free SFX media (Invitrogen or Hyclone Laboratories) in a 27 C shaking incubator. To assess purity, protein samples were run on an SDS-PAGE gel followed by staining with Coomassie dye. Western blots were performed by transferring the proteins to a nitrocellulose membrane followed by detection with a 6xHis-HRP-conjugated antibody (Clontech).
Affinity Selection of ERR Binding Sequences
M13 phage particles displaying peptides selective for ERR were obtained as previously described (41). In brief, approximately 1 µg baculovirus-expressed, purified ERR was diluted in 100 µl of NaHCO3 (pH 8.5) and applied to a single well of a 96-well plate (Costar) and incubated at 4 C overnight. The wells were blocked with 150 µl of 2% milk in PBST for 1 h at RT followed by five washes with PBST. The wells were treated with 1010 phage particles of a phage peptide library diluted in 100 µl PBST and incubated for 3 h at RT. Nonbinding phages were removed by washing the wells five times with PBST. Bound phages were eluted with 100 µl of 0.1M HCl by incubating at RT for 10 min, followed by neutralization with 50 µl of 1 M Tris-HCl (pH 7.4). Phages eluted from the targets were amplified in E. coli DH5F9 cells for 3 h, and the supernatant containing amplified phage was collected for use in subsequent rounds of panning. A total of three rounds of panning were performed. Enrichment of ERR binding phage was confirmed by a phage ELISA as described elsewhere (41). PCRs were used to recover peptide inserts from bacterial supernatants that showed significant enrichment of target binding phage. The PCR products were digested with XhoI and XbaI before ligation into the expression vector pMsx (described in Ref. 41) for mammalian two-hybrid analysis. The fusion construct expressing two copies of the LXXLL motifs, Gal4DBD-2xL3-09, was derived from the corresponding single-copy peptide-Gal4DBD fusion plasmids by adding a linker sequence (adapted from the sequences found between the PGC-1 NR box 2 and box 3). A second copy of the LXXLL peptide was added, resulting in the two copies of LXXLL motifs separated by 61 amino acids, the same spacing found between the PGC-1 NR box 2 and box 3.
The X7LXXLLX7 (LXXLL) and X7LXXH/I/N/LXXXL/IX7 (CoRNR) peptide M13 phage libraries were described previously (41, 63). The 6G6, 6Y6, 6N6, CC, and SS peptide M13 phage libraries were gifts from D. J. Kenan (Duke University, Durham, NC).
Modeling Studies
Structural models of ER and ERR in complex with the L3-09 peptide were generated using Pymol molecular graphics and modeling package (64). Specifically, the crystal structure coordinates of the LBDs of the ER (PDB: 3ERD) (65) and the ERR in complex with the PGC-1 L3 peptide (PDB: 1XB7) (33) were aligned using all the backbone atoms of each structure. The overall rms deviation in the aligned structures was 1.39E. The residues of the PGC-1 L3 peptide were mutated in silico to the residues of the L3-09 peptide and the most frequently occurring amino acid rotamer was selected.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
This work was supported by a Nuclear Receptor Signaling Atlas Grant DK62434 (to D.P.M.) and by DAMD17-03-1-0174 (to S.G.). M.A.D. was the recipient of National Research Service Award DK074307.
Disclosure Statement: The authors have nothing to disclose.
First Published Online October 19, 2006
Abbreviations: AF, Activation function; CMV, cytomegalovirus; CYP, cytochrome P-450; ER, estrogen receptor; ERE, estrogen response element; ERR, estrogen receptor-related receptor; ERRE, ERR-responsive element; Gal4DBD, Gal4DNA-binding domain; GST, glutathione-S-transferase; HRP, horseradish peroxidase; LBD, ligand-binding domain; NR, nuclear receptor; NRID, NR-interaction domain; PBST, PBS + Tween 20; PGC, PPAR coactivator; PPAR, peroxisome proliferator-activated receptor-; rms, root mean square; RT, room temperature; SHP, short heterodimer partner; SRC, steroid receptor coactivator.
Received for publication April 26, 2006. Accepted for publication October 13, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
, but not by ERß. EMBO J 18:4270–4279[CrossRef][Medline]) promoter dictates peroxisome proliferator-activated receptor coactivator-1 control of ERR expression. J Biol Chem 279:18504–18510. Cell Growth Differ 9:1007–1014[Abstract] in human breast carcinoma as a potent prognostic factor. Cancer Res 64:4670–4676 and estrogen-related receptor associate with unfavorable and favorable biomarkers, respectively, in human breast cancer. Cancer Res 62:6510–6518 (ERR) and oestrogen receptors (ER and ERß) exhibit different gene expression in human colorectal tumour progression. Eur J Cancer 41:1487–1494[CrossRef][Medline]. Mol Cell Biol 23:7947–7956 and Gabpa/b specify PGC-1-dependent oxidative phosphorylation gene expression that is altered in diabetic muscle. Proc Natl Acad Sci USA 101:6570–6575
