| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Coactivator Binding Interface: A Structural Basis for Ligand-Independent Transcription
Queensland University of Technology (J.M.H.), Centre for Molecular Biotechnology, Brisbane 4001, Queensland, Australia; and University of Queensland (P.L., S.L.C., G.E.O.M.) Institute for Molecular Bioscience, Australian Research Council Special Research Centre for Functional and Applied Genomics, Ritchie Research Laboratories, St. Lucia 4072, Queensland, Australia
Address all correspondence and requests for reprints to: A/Pr George E. O. Muscat, Institute for Molecular Bioscience, The University of Queensland, Research Road, Ritchie Building B402A, St. Lucia, Queensland 4072, Australia. E-mail: G.Muscat{at}imb.uq.edu.au.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(ROR) is a member of the ROR subfamily of orphan receptors and acts as a constitutive activator of transcription in the absence of exogenous ligands. To understand the basis of this activity, we constructed a homology model of ROR using the closely related TRß as a template. Molecular modeling suggested that bulky hydrophobic side chains occupy the ROR ligand cavity leaving a small but distinct cavity that may be involved in receptor stabilization. This model was subject to docking simulation with a receptor-interacting peptide from the steroid receptor coactivator, GR-interacting protein-1, which delineated a coactivator binding surface consisting of the signature motif spanning helices 3–5 and helix 12 [activation function 2 (AF2)]. Probing this surface with scanning alanine mutagenesis showed structural and functional equivalence between homologous residues of ROR and TRß. This was surprising (given that ROR is a ligand-independent activator, whereas TRß has an absolute requirement for ligand) and prompted us to use molecular modeling to identify differences between ROR and TRß in the way that the AF2 helix interacts with the rest of the receptor. Modeling highlighted a nonconserved amino acid in helix 11 of ROR (Phe491) and a short-length of 3.10 helix at the N terminus of AF2 which we suggest 1) ensures that AF2 is locked permanently in the holoconformation described for other liganded receptors and thus 2) enables ligand-independent recruitment of coactivators. Consistent with this, mutation of ROR Phe491 to either methionine or alanine (methionine is the homologous residue in TRß), reduced and ablated transcriptional activation and recruitment of coactivators, respectively. Furthermore, we were able to reconstitute transcriptional activity for both a deletion mutant of ROR lacking AF2, and Phe491Met, by overexpression of a GAL-AF2 fusion protein, demonstrating ligand-independent recruitment of AF2 and a role for Phe491 in recruiting AF2. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-helix (helix 12) to create a hydrophobic cleft. The helices forming this cleft show a high level of sequence conservation across the nuclear receptor subfamilies and are known as the signature motif - (F/W)AKXXXXFXXLXXXDQXXLL. An elegant series of structural studies by Feng and others (4, 5, 6) revealed that this hydrophobic cleft plays an important role in coactivator recruitment through binding a highly conserved LXXLL motif (7). Consistent with this, LXXLL-containing peptides have been shown to be potent antagonists of retinoid orphan-related receptor- (ROR) activity (8). This motif and the immediate flanking amino acids determine the specificity of coactivators for particular nuclear receptors. Interestingly, a similar sequence -
xx- (9, 10) has been implicated in recruiting corepressor molecules such as silencing mediator of retinoid and thyroid hormone receptor and nuclear receptor corepressor. Previously we investigated corepressor recruitment by the constitutive repressors Reverb and RVR. That study implicated a hydrophobic region formed by helices 3, 4, and 11 in corepressor recruitment. This region overlaps the coactivator binding site. Accordingly, it is likely that coactivators and corepressors bind in a similar position and repression or activation is dictated by the positioning of a dynamic equilibrium between coactivator and corepressor binding (11).
Paradoxically, there exists a class of receptors that contain the nuclear receptor signal sequence, but for which no ligands have been found. These receptors are known as orphan receptors, some of which appear to function in the absence of ligands [e.g. Shp and nur77 (12, 13)], raising the question of how these transcription factors are controlled. Retinoid orphan-related receptor (ROR) is a member of the ROR subfamily of orphan receptors and a constitutive activator of transcription (14). ROR, together with the other members of the subfamily RORß and ROR
, is widely expressed and is integral to a number of physiological and developmental processes including retinal development (15), thymopoiesis (16, 17), inflammation (18, 19), and bone remodeling (20). Accordingly, their modulation is an attractive pharmaceutical goal.
Previously Lau et al. (21) demonstrated that ROR appears to be a true orphan receptor in that it efficiently recruits the coactivator P300 in vitro in the absence of added ligand in glutathione-S-transferase (GST) pull-down assays. Additional studies by Atkins et al. (22) showed similar results for the coactivators GR-interacting protein 1 (GRIP-1) and VDR-interacting protein 205 (DRIP205) (TR-associated protein/PPAR binding protein). Coactivator recruitment was also demonstrated in yeast, which is unlikely to harbor a ligand for a higher eukaryotic nuclear receptor. Additionally, it was shown that ROR could activate transcription and interact with coactivators in vivo in mammalian twin hybrid analysis (21). Two mutations performed in the Atkins study indicated that integrity of the signal sequence in addition to helix 12 was required for transactivation and coactivator recruitment. Significantly, both mutations (Val335Arg and LeuPhe510AlaAla) cause disruption of the hydrophobic coactivator interface. More recent work by Kane and Means (8) demonstrated that ROR activity was stimulated by coexpression of calcium calmodulin-dependent kinase IV (CaMKIV). Although ROR itself was not directly phosphorylated by CaMKIV, treatment of ROR expressing cells with an artificial calcium ionophore (ionomycin), resulted in a 6-fold increase in transcriptional activity. Recently, McKinsey et al. (23, 24) have shown that CaMKIV-dependent phosphorylation stimulates export of histone deacetylases from the cell nucleus regulating muscle differentiation. Hence, the regulatory effect of CaMKIV may well be one of derepression rather than activation. Involvement of CaMKIV and the enhancement of ROR activity by calcium flux provide a conceptual link between the ROR transcriptional activity and the extracellular environment. However, it does not explain how ROR can activate transcription in the absence of ligand.
To date, the ROR LBD has proved refractory to crystallization. Here we have used the strong sequence homology between ROR and TRß (for which the tertiary structure has been determined) to construct a model of ROR with a bound LXXLL motif peptide. This enabled dissection of the critical determinants of ROR interaction with coactivators by site-directed mutagenesis using structural cues. Additionally, the model suggested a structural basis for ROR’s ability to dispense with ligand binding through trapping activation function 2 (AF2) in the active holoposition.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
solved by Renaud et al. (25) as a template for modeling the Rev-erbA and RVR LBDs. Initially, we also used this receptor as a template for analysis of the ROR LBD on the basis of the high homology between ROR and RAR across the entire receptor and the supposition that ROR would most closely resemble a retinoid-type receptor. However, as modeling progressed it became apparent that critical areas for coactivator interaction showed much higher conservation with TRß than RAR; in particular, the AF2 region was almost identical between the two receptors. Accordingly a ClustalW alignment was carried out and formed the basis for threading the ROR amino acid sequence onto a TRß template (see Fig. 1
) and homology modeling. The template structure had a number of side chains missing, and the C backbone was incomplete between helix 2 and 3. These features were rebuilt for the ROR model by scanning an archive of loop structures for suitable minithreading templates, which were then used to rebuild the missing amino acids. This procedure produced a robust solution with fewer than 3% of residues in disallowed regions of a Ramachandran plot (see Fig. 2).
|
|
model were defined by a spherical probe with a radius of 1.4 Å. Analysis of TRß revealed a ligand binding cavity with a volume of 568 Å3 in which T3 (305 Å3) makes a tight complementary fit (see Fig. 3). The cavity is lined by aromatic and bulky hydrophobic residues forming van der Waals contacts with bound ligand. A single strong hydrogen bond is formed between His435 and T3. ROR, on the other hand, shows a much smaller cavity with a volume of 236 Å3. This reduction is the result of numerous substitutions of bulky hydrophobic side chains in ROR compared with homologous residues in TRß within the ligand binding cavity as shown in the enlarged portion of Fig. 3. Substitutions include replacement of TRß Ser328 with phenylalanine in ROR, Ala331 with methionine, Met348 with tyrosine, and Gly358 with leucine. The ROR cavity proved to be robust to Monte Carlo minimization and simulated annealing, maintaining its small volume. T3 approaches the lower limit for the size of ligand bound by nuclear receptors or any of the steroid receptor superfamily (26). However, we consider the existence and size of the modeled ligand pocket significant and do not rule out a role for a novel nonretinoid agonist that modulates LBD stability or activity.
|
Ligand Binding Cavity Silence Transcription shows weak activity with its native promoter, we fused full-length receptor to the yeast GAL4 DNA binding domain and gauged the effects of the alanine mutants using a reporter carrying five copies of the GAL response element upstream of a luciferase reporter enzyme (Luc). The two cavity mutations (Phe365Ala and Phe391Ala) dramatically compromised the ability of ROR to activate transcription (see Fig. 4A). In contrast, the neutral mutation of the residue (Glu329) outside the interface had little effect on ROR activity. Western blot analysis of the N-terminal FLAG-tagged native and mutant RORs (Fig. 4B) demonstrated all the mutant proteins were efficiently expressed. Moreover, it indicates that diminished activity was not due to ineffective transfection or compromised expression. We note that the cavity mutants were expressed at slightly lower levels than the wild-type ROR or other mutants investigated.
|
, together with their almost complete identity in signature and AF2 motifs, it is highly likely that they interact with similar NR boxes (5). Accordingly, NR box II from GRIP-1 was selected for modeling in the coactivator binding groove. Docking was achieved by superimposing TRß onto ROR and merging the TRß coactivating peptide into the ROR coordinate file. Interaction of the peptide with ROR was then refined using Monte Carlo minimization, which resulted in a translation of the GRIP interacting peptide by 1.68 Å along the axis of the GRIP-1 peptide -helix in the direction of the C terminus (see Fig. 5, A and B). Peptide docking caused little rearrangement of the receptor interface with the exception of Lys339, which showed considerable movement, rotating 98o about the side chain C and causing it to make extensive van der Waals contacts with leucines 693 and 694 of the coactivator peptide LXXLL motif. This contrasts with the equivalent residue in TRß where contact is restricted to Leu693. Additionally, in the TRß structure, coactivator peptide forms a single hydrogen bond to the Glu509 carboxylate through main chain amides at Ile689 and Leu690.
|
via the three leucines forming the LXXLL motif and, to a lesser extent, via Ile689 and Asp696 (see Fig. 5A
). Side chains from these amino acids are buried in a hydrophobic groove surrounded by a charged periphery. Coactivator peptide-bound structures for other receptors (ER, TR, and PPAR) show similar features. This surface is formed by contributions from 16 amino acids from helices 3, 4, 5, and 12. Surprisingly, AF2 (helix 12) contributes relatively little to this feature with the majority of the hydrophobic groove being comprised of Lys339 and Val335 with the charged periphery being formed by the amino group of Lys339 and carboxylate of Glu509 constituting the charge clamp previously described by Nolte et al. (27), which has been suggested to have a role in stabilization of the coactivator helix. Additional contributions are made by Gln352, Glu336, Gln352, and Ile353. The TRß interface is composed of identical amino acids except that Gln352 is replaced by threonine. Figure 5B shows the van der Waals contact surface between coactivator and receptor for the ROR model. We used scanning alanine mutagenesis to probe the subset of residues forming this contact surface and assayed the resulting mutants for coactivator binding and transcriptional activity.
Transcriptional and in Vitro Binding Analysis of ROR Coactivator Interface Mutants: Lys339 and Ile353 Are Essential for Transcriptional Activity
Initially, six signature sequence mutants were constructed: Gln332, Val335, Glu336, and Lys339 in helix 3 and Gln352 and Ile353 in helix 4, with wild-type residues being replaced by alanine. These residues were selected on the basis of the van der Waals contact surface defined in Fig. 4B and were deemed unlikely to cause gross structural changes to the receptor as a whole. As a first step in gauging the effect of these mutations in vivo, we set out to assess transcriptional activity of the mutants. Repeated assays demonstrated that the Lys339 mutant was transcriptionally inactive and that Ile353 had significantly reduced activity (see Fig. 6A). Subsequently, mutants were modified by addition of an N-terminal FLAG tag. Western blotting of cell extracts from FLAG-fusion protein transfections and immunodetection using a FLAG-specific antibody showed that FLAG-ROR mutants (in particular Lys339 and Ile353) were expressed at essentially identical levels to wild-type receptor. Furthermore, there was no evidence of proteolytic degradation of fusion proteins, indicating that the effects of mutations on reporter activity were unlikely to be compromised by protein stability or expression levels (see Fig. 4B).
|
when compared with GST alone presented bound to glutathione agarose at a similar concentration (see Fig. 6B). The steroid receptor coactivator GRIP-1 showed the highest level of interaction followed by P300 and DRIP205. These interactions occurred in the absence of added ligands in contrast to similar assays performed for TRß (4) or ER (28) and confirms the findings of Atkins et al. (22) and Lau et al. (21). Increased coactivator interaction was apparent for Glu336 with all three coactivators. This increase probably reflects the enhanced hydrophobicity of the coactivator binding interface caused by deletion of the glutamate carboxylate and its replacement with uncharged alanine, which in turn would facilitate interaction with the hydrophobic LXXLL motifs common to each of the coactivators.
ROR Interacts with Coactivators in a Cellular Context: Lys339Ala and Ile353Ala Mutations Compromise Coactivator Recruitment
To gain further insight into the ROR-cofactor binding surface, we examined the interaction between wild-type and mutant RORs with coactivators (GRIP-1, DRIP205, and P300) in a cellular context. These results are summarized in Fig. 7. Previously Lau et al. (21) and Atkins et al. (22) had demonstrated ROR interaction with DRIP205, GRIP-1, and P300 using this technique. All three coactivators showed a failure to interact with both the Lys339 and Ile353 mutants in a cellular context, indicating that they play a pivotal role in the coactivator interface.
|
Receptor Core in the Absence of Ligands and TRß, given that TRß requires ligand to activate transcription, whereas ROR is constitutively active and able to bind coactivators in the absence of ligand. The similarity between ROR and TRß rules out the coactivator interface as the source of ROR’s constitutive activity. This prompted us to focus on intra-LBD interactions, the most significant of which is the ligand-dependent repositioning of helix 12, which constitutes the AF2 function of nuclear receptors.
We investigated intra-LBD interactions involving helix 12 by coexpressing a deletion mutant of ROR lacking helix 12 fused to the GAL DBD (GAL-ROR
H12) and helix 12 only fused to GAL (GAL-AF2). ROR lacking AF2 is transcriptionally inactive. Therefore, if cotransfected GAL-AF2 were to interact with the GAL-ROR deletion mutant, an active receptor would be formed that would be able to activate transcription from a GAL-based reporter construct. It should be noted that the AF2/LBD interaction in this system would be extremely inefficient because it normally occurs in the context of the two partners being physically joined. However, having both AF2 and ROR fused to GAL effectively tethers the AF2 to the reporter construct, maximizing the chances of AF2 complementing the deleted ROR. We found that coexpressed GAL-AF2 synergistically complemented GAL-RORH12, activating transcription 4- to 5-fold above levels recorded for either plasmid on its own or GAL0 with either plasmid (see Fig. 8). In contrast, wild-type ROR was actually inhibited by coexpression of GAL-AF2 (see Fig. 10A). Helix 12 acts as a lid to the ligand binding cavity of all the liganded nuclear receptors. This suggests that the ROR LBD is intrinsically able to recruit the AF2 helix in the absence of ligands and so activate transcription in a ligand-independent manner.
|
|
as shown by our in vitro and in vivo studies. We reasoned that coexpression of GRIP-1 with Gal-AF2 and GAL-RORH12 would further stabilize the helix12/receptor interaction described above. Accordingly, we conducted cotransfection experiments with pSG5-GRIP-1 and the Gal-AF2 and GAL-RORH12 constructs (see Fig. 8). Coexpression of GRIP-1 resulted in a 3-fold increase in the synergized transactivation as a consequence of coexpressed Gal-AF2 and GAL-RORH12 (relative to the vector, pSG5, alone). Transcription of individual constructs was also enhanced but to a much lesser degree. This is consistent with formation of a ternary complex.
Structural Basis of Helix 12/Receptor Interaction: Identification of Key Residues in Helices 11 and 12
These results prompted us to investigate the interface between the receptor and helix 12. Surveying the receptor within 4 Å of helix 12 retrieved 11 side chains from helices 3, 4, 5, and 11. These residues were highly conserved between TRß and ROR with the exception of Phe491 in ROR, which was replaced by the much more polar Met442 in TRß. This methionine makes van der Waals contact with bound T3 in the TRß and is oriented such that its side chain projects into the receptor core. In ROR, modeling suggests that Phe491 is oriented away from the receptor core and makes extensive van der Waals contacts with Tyr504 of the ROR AF2, as shown in Fig. 9. Furthermore, the aromatic side chains overlap each other to such an extent that they may well be able to form the stable ring interactions characterized by base stacking in DNA. This overlap could serve to keep helix 12/AF2 folded across helix 11 in the active holoconformation. Given the similarity between ROR and TRß in this region, we decided to investigate the effect of mutating Phe491 to methionine in ROR. This substitution would be highly conservative in structural terms but would abolish the extensive van der Waals interactions between Phe491 and Tyr504. We also undertook a more radical substitution at Phe491, replacing the residue with alanine, which caused considerable disruption in our model. Additionally, we mutated a highly conserved glutamate residue in AF2/helix 12 (Glu509) to alanine, which has been shown to cause transcriptional inactivation and a failure to recruit coactivators in other receptors.
|
activity as did the Glu509Ala mutation. We reasoned that the reduced activity of Phe491Met was due to inefficient recruitment of AF2. Accordingly it should be possible to rescue the mutant by coexpression of Gal-AF2, because the apparent increase in the local concentration should increase the rate of AF2 assuming the holoposition. We found that coexpression of the GAL-AF2 described above appreciably increased transcriptional activity of Phe491Met but actually decreased wild-type ROR transactivation and minimally increased activity for Glu509Ala (see Fig. 10A). We interpret these results as follows: a conformational equilibrium exists between apo- and holoforms of ROR (aporeceptor + AF2 
holoreceptor AF2). In wild-type receptor the equilibrium is shifted far toward the holoform because of the overlap of Phe491 and Tyr504. Hence, no further increase in transcriptional activity would be gained by increasing the local concentration of AF2. In contrast, Phe491Met would tend to shift the equilibrium in the direction of the apoform. In this case an increase in local AF2 concentration would shift the equilibrium back toward the active holoreceptor reconstituting receptor activity. Glu509Ala is unable to recruit cofactors because of a change in the electrostatic profile of AF2 itself rather than through an inability to secure AF2. Hence, additional AF2 will be unable to reconstitute receptor activity.
We also analyzed the two new mutants in vitro using the GST pull-down assay described above. Both Phe491Met and Glu509Ala greatly reduced coactivator interaction (see Fig. 10B). However, it should be noted that the Phe491Met mutant showed residual binding with GRIP-1, when compared with the transcriptionally inactive Glu509 mutant, while ablating interaction with other coactivators (i.e. reduced binding to the level of the GST control). The Glu509Ala mutant also abolished all coactivator interactions. These findings were reflected in the transcriptional activity of the mutants with Phe491Met showing reduced transcriptional activity compared with wild type, whereas Glu509Ala reduced transcriptional activity to background levels (see Fig. 10A).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
. Thus, transcriptional silencing of Lys339Ala and Ile353Ala can be easily rationalized in terms of a failure to interact with this class of LXXLL coactivator. Our modeling suggests structural reasons for this failure as follows:
Lys339Ala.
In our model this residue packs tightly between Leu693 and 694 of the LXXLL motif. The hydrophobic aliphatic portion of the side chain makes extensive van der Waals contacts with the side chains and main chains of Leu693 and 694. However, it makes no hydrogen bonds with the coactivator main chain. All three assays show reduced binding with coactivators for this mutant. This was most obvious with the transcriptional and mammalian twin hybrid assays in which the mutant was transcriptionally inactive. Substitution of alanine for lysine in ER (28) and VDR (29) also causes transcriptional inactivation. However, work by Mak et al. (29A ) showed a more conservative lysine-to-leucine mutant of ER still possessed transcriptional activity. The loss of activity by the ROR mutant is readily understandable in terms of disruption of an extensive set of van der Waals contacts between the receptor and leucines of the LXXLL motif. However, TRß Lys339
Ala has some residual transcriptional activity, which is surprising given the conservation in this region between TRß and ROR. The work of Mak et al. (29A ) shows that it is the aliphatic side chain of the lysine residue that is most important in driving the coactivator interaction, and it is significant that the adjacent residue to the mutated lysine in TRß is another lysine (Lys340), whereas in ROR and ER the equivalent residue is an arginine, which lacks a hydrophobic aliphatic moiety. In the TRß mutant, the adjacent lysine presumably complements the alanine mutation, whereas this is not possible in ROR and ER in which the equivalent residues are more polar and lack the long aliphatic lysine side chain.
Ile353Ala.
This side chain is equidistant from Leu690, His691, and Leu694, packing tightly with these three side chains. Its closest approach is to the 
-carbon of His691 of the modeled coactivator peptide which is 2.6 Å from the Ile -carbon. This residue forms a large part of the binding groove of the coactivator peptide and may play an important role in orientating the peptide on binding because it forms extensive van der Waals contacts with three of the peptide’s side chains. Alanine substitution abolishes all these contacts, and thus it is not surprising that the Ile353Ala mutant shows greatly reduced interaction for all coactivators and poor transcriptional activity.
Is ROR Truly an Orphan Receptor?
In addition to causing the repositioning of helix AF2 during receptor activation, ligands appear to play a more global role in receptor structure and make extensive contacts with the receptor core (6, 29, 30). Indeed, bound ligand seems to be completely buried in the majority of receptor structures in the Protein Data Base. When comparing TRß and ROR structures in the immediate vicinity of bound ligand, one is struck by the number of amino acid residues with small polar side chains that are replaced by large bulky hydrophobic amino acids as described in the modeling section. A similar phenomenon is apparent in model structures of the orphan receptors RVR and Reverb, both of which lack helix 12 and hence are very unlikely to bind ligands (25). This substitution could well be an adaptation to ligand-independent activity with structural roles normally occupied by bound ligand being complemented by bulkier hydrophobic side chains, some of which actually intrude into the space that would have been occupied by bound ligand. This hypothesis is strengthened by the inactivation of ROR by mutations in the ligand binding cavity investigated in our study.
Molecular modeling of the ROR LBD shows a distinct cavity in the equivalent region to ligand binding volumes in both RAR and TRß. This cavity seems to be too small to hold a retinoid-like ligand. However, given that there is a structural role for ligand stabilization of the LBD/receptor and conformational homogeneity, it may be that a small novel ligand further stabilizes the active ROR-coactivator peptide conformation. Indeed, Missbach et al. (31) have presented data suggesting that melatonin and thiazolidinediones (TZDs) are both ligands for ROR. Given the high degree of homology between TRß and ROR, it might be assumed that the two receptors would bind similar ligands. However, melatonin and TZDs bear little resemblance to T3, the natural ligand for TRß. Furthermore, we were unable to dock either melatonin or the TZD CGP53065 into the ROR LBD cavity without considerable disruption to the LBD. A similar procedure carried out for the antagonist R1881 docked into a model structure of the AR gave a satisfactory solution that was consistent with a subsequent x-ray crystallography structure (32, 33). To date, there have been no reports describing direct binding of melatonin or TZDs to the LBD of ROR, and the initial reports of the effects of TZD and melatonin on ROR mediate transcription remain controversial. Consequentially, it may well be that ROR is indeed a true orphan receptor and does not acquire ligands.
Agonist-Independent ROR Activity 1: Fixing AF2 in the Holoconformation
Repositioning of AF2 in nuclear receptors on ligand binding is widely regarded as being the conformational switch that toggles receptors between transcriptionally active and inactive states. Comparison of the structures of liganded and unliganded receptors clearly shows that there is a ligand-driven repositioning of helix 12 to pack against the signature sequence formed from helices 3–5 and helix 11 [see Wurtz et al. (29)]. Consistent with this, TRß acts as a repressor in the absence of ligands, and this repression is greatly enhanced by truncation of the AF2 helix (34, 35) Similarly, work by Harding et al. (36) showed that truncation of ROR caused it to repress expression of its natural promoter and recruit corepressors in vitro. Both phenomena point to the importance of the position of AF2 in determining whether receptors activate or silence transcription. Given the ability of ROR to recruit coactivator in a ligand-independent manner in a variety of contexts (in vitro with bacterially expressed protein; in yeast and in mammalian tissue culture with charcoal-stripped serum), it is highly likely that AF2 is constantly in the holo/transcriptionally active position folded against the helices forming the signature sequence. This view is supported by the data that we present here showing that the transcriptional activity of wild-type ROR is not synergized by coexpression of AF2, whereas receptor bearing a mutation designed to weaken its affinity for AF2 is effectively reconstituted. This, in turn, suggests increased stability of the active AF2 conformation in comparison to TRß.
Further comparative modeling of ROR revealed another structural motif present in ROR that may contribute to increased stability of its holoform. Despite the high levels of identity between TRß and ROR in the region of AF2, there is a striking deviation at the loop between helix 11 and helix 12. Structural and sequence alignments show an insertion of three amino acids, Val499, Arg500, and Leu501. These residues form a tight turn with the bulky side chains on the exterior of the turn, which is stabilized by main chain hydrogen bonding similar to that found in the theoretical 3.10 helix conformation first described by Pauling (see Fig. 8). This turn may act as a molecular spring, keeping AF2 positioned against helix 11. Significantly, the three amino acids forming this turn are also present in the orphan receptor COUP-TF1 between helices 11 and 12. Like ROR, COUP-TF1 has yet to have any ligand interactions described for it. Furthermore, a similar motif is apparent in the PR at the N terminus of its AF2 helix. Recent work by Sack et al. (37) suggests the AF2 helix from the closely related AR may interact more tightly with unliganded receptor and that the short helical section may increase interaction with the receptor core. All three receptors have extensions to their helix 12 (AF2) regions. Connection between helices 11 and 12 is likely to be strengthened further by ring stacking interactions between Phe504 at the N terminus of helix 12 and Phe491 at the C terminus of helix 11. Significantly, the positional equivalent amino acid residues in TRß are Met442 and Phe551. Not only is the methionine residue incapable of forming a ring stacking interaction, it scarcely makes contact with the opposite residue instead making contact with bound T3. It is reasonable that these differences between ROR and TRß reflect the importance of repositioning helix 12 and how subtle changes in the immediate environment of AF2, rather than the nature of the helix itself, are able to dictate receptor activity. This is graphically illustrated by the constitutive activity of an ER mutant where mutation of Tyr537 to alanine, which is located in the flexible region between helices 11 and 12, causes the receptor to become constitutively active. Replacement of the bulky tyrosine side chain with the much smaller alanine would tend to increase the flexibility of the helix 11–12 hinge. Although this would doubtless incur an entropic loss in terms of being able to adopt a transcriptionally active position, any penalty could well be offset by increased accessibility of the AF2-helix to its active binding position. Similarly, Phe491 is replaced by Met442 in ER, a residue that undergoes a dramatic conformational change during binding of the antiagonist 4-hydroxytamoxifen (6). In addition to these two modeled features, ROR possesses an unusually long extension to helix 12, some 10 amino acids longer than TRß. Due to its proximity to the C terminus of the protein and a lack of suitable templates, it is not currently possible to model this extension with any certainty. However, it is highly likely that this extension contacts the rest of the receptor and hence this too might serve to stabilize the AF2 helix in the transcriptionally active holoposition.
Why ROR Does Not Need Ligands 2: Control of ROR by CAMIV Kinase
Work by Lau et al. (21) clearly demonstrates that ROR has a function in muscle differentiation. Furthermore, the phenotypes of the naturally occurring Staggerer mouse and artificial ROR knockout show that the receptor plays a role in the development of multiple tissues. If the receptor is constitutively active by virtue of permanent fixation of AF2, how can it be regulated? Kane and Means (8) have shown that the receptor is sensitive to calcium flux via the calcium calmodulin kinase CaMKIV, although the receptor itself is not phosphorylated by this kinase. This observation is conceptually consistent with the developmental role played by ROR because the key differentiation signals, vitamin D and thyroid hormone, both induce intracellular calcium fluxes (38, 39). This may, in turn, account for the apparent role played by ROR in Purkinje cell maturation in response to thyroid hormone. Significantly CaMKIV also potentiates transcription by the orphan receptor COUP-TF1, which also possesses the ValArgLeu tripeptide motif described above.
In conclusion, our analysis of in vitro and in vivo interaction between coactivators and binding interface mutants of ROR shows that this orphan receptor uses the nuclear receptor signature motif to recruit coactivator in a similar way to other receptors but does not require ligand binding for activation. We hypothesize that this reflects fixation of AF2 in the holoreceptor conformation described for ligand-bound receptors and that this fixation is brought about by structural variation in helix 11 and loop 11/12. Although we cannot rule out the existence of naturally occurring ligands for ROR, it seems unlikely that the receptor is able to bind a retinoid type molecule and has evolved away from the liganded receptor structure by preferential use of bulky hydrophobic amino acids as its core. Similarly we cannot rule out the existence of a repressive ligand as has been found for the receptor CAR (47). Notwithstanding this, the receptor does seem to possess a residual ligand-binding cavity, and it is quite possible that this could be used as a target for the development of small molecule antagonists of ROR action. However, rational design of these compounds will have to wait until a good physical structure of the ROR LBD exists.
During preparation of this manuscript, Stehlin et al. (46) determined the structure of RORß in complex with stearic acid by x-ray crystallography. Prospects for rational design of specific ROR antagonists have been given a considerable fillip as a result of this publication, which also describes a homology model for ROR based on the RORß LBD structure. Stehlin et al. confirm the importance of K339 in coactivator recruitment and suggest that ROR has reduced ligand binding cavity compared with RORß due to substitution of bulky aromatic and aliphatic side chains. In contrast to our study, Stehlin et al. predicted a cavity of 568 Å for ROR compared with our estimate of 236 Å. This disparity in size reflects the difference in modeling templates used in the two studies and strengthens the possibility that ROR has a natural ligand. However, it should be noted that the homology model constructed by Stehlin et al. was based on a crystal structure derived from a receptor complexed with a ligand that did not activate transcription. Furthermore, RORß could not be cocrystallized with stearic acid in the absence of coexpressed coactivator peptide, suggesting that the natural ligand for RORß could be markedly different.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
model was constructed using the Modeler Module of InsightII (MSi) with the three-dimensional structure of TRß (PDBid: 1BSX) acting as a template. The resulting model was subject to energy minimization, Ramachandran analysis, and further quality checking with the WhatIf suite of programs (www.cmbi.kun.nl/swift/whatcheck/). Electrostatic potentials for this finalized coordinate file were then calculated with the DELPHI (40) implementation within Insight II and displayed using the surface modeler GRASP (41). Peptide docking was achieved using the program Sculpt (MDL) (42). Structures were ray traced with the freeware program "Persistence of Vision: POV-Ray" (http://www.povray.org) and the Macintosh patch "MacmegaPov" (http://users.skynet.be/smellenbergh).
Plasmids
Full-length pSG5-ROR (21) was subject to QuikChange mutagenesis (Stratagene, La Jolla, CA) according to the manufacturer’s instructions, using the mutant oligonucleotide pairs shown below. Mutations were confirmed by automated dideoxy sequencing. Full-length ROR mutants were excised from pSG5 by EcoRI digestion and ligated into pGAL0 (43) that had been predigested with EcoRI and treated with alkaline phosphatase. An EcoRV/HinDIII fragment was excised from VP16-ROR (amino acids 143–523) (21) and replaced with an analogous fragment from mutant pSG5-ROR to produce a panel of VP16-ROR mutants. Coactivator plasmids were a kind gift from Dr. Tom Chen with the exception of DRIP205, which was donated by Dr. Leonard P. Freedman as a pCDNA3.0 plasmid from which the coactivator was subcloned into pGEX and pGal0 by PCR amplification using the oligonucleotides shown below. The GAL-AF2 plasmid was constructed by oligonucleotide ligation. Briefly, oligos GMUQ 635 and 636 were phosphorylated with polynucleotide kinase and annealed before being ligated to EcoRI/Xba-digested GAL0.
Site-Directed Mutagenesis Oligonucleotides
GM607 ACAGAAGCTATAGCATATGTGGTGGAG Q332A forward
GM608 ctccaccacatatgctatagcttctgt Q332A reverse
GM609 atacagtatgtggcagagtttgccaaa v335A forward
GM610 tttggcaaactctgccacatactgtat v335A reverse
GM611 cagtatgtggtggcatttgccaaacgc e336A forward
GM612 gcgtttggcaaatgccaccacatactg e336A reverse
GM613 gtggagtttgccgcacgcattgatgga k339A forward
GM614 tccatcaatgcgtgcggcaaactccac k339A reverse
GM615 tgtcaaaatgatgcaattgtgcttcta q352A forward
GM616 tagaagcacaattgcatcattttgaca Q352A reverse
GM617 caaaatgatcaagcagtgcttctaaaa i353A forward
GM618 ttttagaagcactgcttgatcattttg i353A reverse
GM639 ACAGAAAAGCTAATGGCAATGAAAGCAATATACCCAGAC F491M forward
GM640 GTCTGGGTATATTGCTTTCATTGCCATTAGCTTTTCTGT F491M reverse
GM639 ACAGAAAAGCTAATGGCAATGAAAGCAATATACCCAGAC F491A forward
GM640 GTCTGGGTATATTGCTTTCATTGCCATTAGCTTTTCTGT F491A reverse
GM641 TCTCTAGAGGTGGTGGCTATCAGAATGTGCCGT F365A forward
GM642 ACGGCACATTCTGATAGCCACCACCTCTAGAGA F365A reverse
GM643 GCGAGCCCCGATGTCGCCAAGTCCCTAGGTTGT F391A forward
GM644 ACAACCTAGGGACTTGGCGACATCGGGGCTCGC F391A reverse
GM645 GCCATCAAGATTACAGCAGCTATCCAGTATGTG E329A forward
GM646 CACATACTGGATAGCTGCTGTAATCTTGATGGC E329A reverse
GM647 CGCATTGATGGATTTGCGGAGCTGTGTCAAAAT M345A forward
GM648 ATTTTGACACAGCTCCGCAAATCCATCAATGCG M345A reverse
GM621 CCATTATACAAGGCATTGTTCACTTCA E509A forward
GM622 tgaagtgaacaatgccttgtataatgg E509A reverse
GM623 cgacttcatttttagccattatacaag 
AF2 forward
GM624 cttgtataatggctaaaaatgaagtcg AF2 reverse
Gal AF2 Oligonucleotides
GM635 AATTCCATTTTCCTCCATTATACAAGGAGTTGTTCACTTCAGAATTTGAGCCAGCAATGCAAATTGATGGGT AF2 forward
GM636 CTAGACCCATCAATTTGCATTGCTGGCTCAAA TTCTGAAGTGAACAACTCCTTGTATAATGGAGGAAAATGG AF2 reverse
DRIP205 Subcloning Primers
CGTCGACATATGAGTTCTCTCCTGGAACGGCTCCAT DRIP205 forward
ATCGATTTAGGATAAGAGGAACTCGGCCAGGGTGCT DRIP205 reverse
Cell Culture and Transfections
DMEM supplemented with 10% FCS, glutamine 300 µg ml-1, kanamycin 100 µg ml-1, and fungizone 1 µg ml-1 (omitted in transfection medium) was used for cell culture. Charcoal-stripped serum or serum-free medium was used for transient transfections. Transcriptional assays were performed using the plasmid G5E1bLUC as a reporter for transcriptional activity. Protein expression was monitored by transfecting wild-type and mutant receptors bearing N-terminal FLAG tags. Western blotting and subsequent immunodetection gave an indication of the effects of mutations on transcriptional and translational efficiency of the various constructs.
Transcription Assays
JEG3 human choriocarcinoma cells were transfected at 80% confluence in 12-well plates with 1 µg G5E1bluc and 0.33 µg ROR plasmids using the DOTAP/DOSPER (Roche Diagnostics Australia, Brisbane, Queensland, Australia) procedure as described previously (44). Luciferase activity was assayed using a Luclite kit (Packard Instruments, Meriden, CT) according to the manufacturer’s instructions. Briefly, cells were washed once in PBS and resuspended in 150 ml of phenol red-free DMEM and 150 µl of Luclite substrate buffer. Cell lysates were transferred to a 96-well plate, and relative luciferase units were measured for 5 sec in a Trilux 1450 microß luminometer (Wallac, Inc., Gaithersburg, MD).
Mammalian Twin Hybrid Assays
These were carried out essentially as for the transcription assays except that VP16-ROR and mutants were cotransfected with GAL-coactivator fusion plasmids (0.33 µg of each plasmid + 1 µg G5E1bluc).
In Vitro Interaction Assays (GST Pull Down)
Interaction assays were performed as described by Lau et al. (21) and Sartorelli et al. (45). Briefly, GST fusion plasmids were transformed into competent Escherichia coli BL21 and grown in TYP medium containing 50 µg/ml ampicillin. Fusion protein production was induced in late-log phase cultures by addition of 0.2 mM isopropyl-ß-D-thiogalactopyranoside followed by a further 4-h growth at ambient temperature. After induction, bacteria were lysed by treatment with 1 mg/ml lysozyme for 1 h at room temperature and subjected to two freeze-thaw cycles in dry ice/ethanol. Liberated DNA was sheared by gentle sonication with a LabSonic microtip (B. Braun Biotech Intl., Allentown, PA) at setting 4 for 30 sec. Purification of fusion proteins was by affinity chromatography on glutathione agarose. Lysates were bound to equilibrated beads and washed extensively in NETN-2 buffer (0.5% NP-40, 1 mM EDTA, 4 mM MgCl2, 10% glycerol, 20 mM Tris, pH 8.0, 100 mM NaCl) containing a complete protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN). pSG5-ROR and mutants were in vitro translated and 35S-labeled using the TNT rabbit reticulocyte lysate kit produced by Promega Corp. (Madison, WI) according to supplied instructions. Glutathione agarose beads loaded with GST-coactivator or GST alone were incubated with 35S-labeled proteins for 3 h at 4 C with gentle rocking in NETN-2 buffer with protease inhibitors and blocking agents (BSA and EtBr). Beads were washed three times with NETN-2, and bound proteins were released by denaturation in SDS-PAGE buffer at 75 C. Solubilized proteins were resolved on 10% SDS-PAGE and subjected to autoradiography with scintillant amplification as described in Ref. 41 .
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: AF2, Activation function 2; CaMKIV, calmodulin-dependent kinase IV; DRIP, VDR-interacting protein; GRIP, GR-interacting protein; GST, glutathione-S-transferase; LBD, ligand-binding domain; ROR, retinoid orphan-related receptor; TZD, thiazolidinedione.
Received for publication July 1, 2001. Accepted for publication December 21, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
, a novel family of orphan hormone nuclear receptors. Genes Dev 8:538–553 in thymocyte survival and lymphoid organ development. Science 288:2369–2373 (ROR) is essential for lymphoid organogenesis and controls apoptosis during thymopoiesis. Proc Natl Acad Sci USA 97:10132–10137-deficient mice. Proc Natl Acad Sci USA 95:3960–3965 in staggerer mice. Nature 379:736–739[CrossRef][Medline]
function in bone metabolism. Proc Natl Acad Sci USA 97:9197–92021 vector in muscle cells impairs differentiation: ROR1 directly interacts with p300 and myoD. Nucleic Acids Res 27:411–420. Mol Endocrinol 13:1550–1557. Nature 395:137–143[CrossRef][Medline]
with potent antiarthritic activity. J Biol Chem 271:13515–135221 and ß1. J Endocrinol 167:219–227[Abstract]
1 and ß1 with an identical truncation in the extreme carboxyl-terminal tau4 domain. Mol Cell Endocrinol 138:95–104[CrossRef][Medline]
, an orphan nuclear receptor required for cerebellar development. Mol Endocrinol 11:1737–1746: evidence that the AF-1 region functions in a cell-specific manner. Nucleic Acids Res 24:264–271. Mol Endocrinol 12:248–262This article has been cited by other articles:
 |
|
 |
|
  H. Odawara, T. Iwasaki, J. Horiguchi, N. Rokutanda, K. Hirooka, W. Miyazaki, Y. Koibuchi, N. Shimokawa, Y. Iino, I. Takeyoshi, et al. Activation of Aromatase Expression by Retinoic Acid Receptor-related Orphan Receptor (ROR) {alpha} in Breast Cancer Cells: IDENTIFICATION OF A NOVEL ROR RESPONSE ELEMENT J. Biol. Chem., June 26, 2009; 284(26): 17711 - 17719. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Du, C. Huang, B. Zhou, and S. F. Ziegler Isoform-Specific Inhibition of ROR{alpha}-Mediated Transcriptional Activation by Human FOXP3 J. Immunol., April 1, 2008; 180(7): 4785 - 4792. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Das, M. Schapira, M. Tomic-Canic, R. Goyanka, T. Cardozo, and H. H. Samuels Farnesyl Pyrophosphate Is a Novel Transcriptional Activator for a Subset of Nuclear Hormone Receptors Mol. Endocrinol., November 1, 2007; 21(11): 2672 - 2686. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K D S. A. Wansa and G. E O Muscat TRAP220 is modulated by the antineoplastic agent 6-Mercaptopurine, and mediates the activation of the NR4A subgroup of nuclear receptors J. Mol. Endocrinol., June 1, 2005; 34(3): 835 - 848. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
And 
statistics are plotted for each of the amino acids in the model structure. These statistics represent N-C
