The three-dimensional structures of antagonistic and agonistic forms of the glucocorticoid receptor ligand-binding domain: RU-486 induces a transconformation that leads to active antagonism

Here we describe the three-dimensional crystal structures of human glucocorticoid receptor ligand-binding domain (GR-LBD) in complex with the antagonist RU-486 at 2.3 A resolution and with the agonist dexamethasone ligand together with a coactivator peptide at 2.8 A. The RU-486 structure was solved in several different crystal forms, two with helix 12 intact (GR1 and GR3) and one with a protease-digested C terminus (GR2). In GR1, part of helix 12 is in a position that covers the co-activator pocket, whereas in the GR3, domain swapping is seen between the crystallographically identical subunits in the GR dimer. An arm consisting of the end of helix 11 and beyond stretches out from one molecule, and helix 12 binds to the other LBD, partly blocking the coactivator pocket of that molecule. This type of GR-LBD dimer has not been described before but might be an artifact from crystallization. Furthermore, the subunits of the GR3 dimers are covalently connected via a disulfide bond between the Cys-736 residues in the two molecules. All three RU-486 GR-LBD structures show that GR has a very flexible region between the end of helix 11 and the end of helix 12.     

The ligand-dependent interaction of mineralocorticoid receptor with coactivator and corepressor peptides suggests multiple activation mechanisms

We investigated the coregulator (coactivator and corepressor) interactions with the mineralocorticoid receptor (MR) that lead to activation and inhibition of the receptor in the presence of agonist and/or antagonist. Our results indicate that MR ligand binding domain (LBD) interacts strongly with only a few specific coactivator peptides in the presence of the agonist aldosterone and that these interactions are blocked by the antagonist eplerenone. We also discovered that cortisol, the preferred physiological ligand for the glucocorticoid receptor in humans, is a partial MR agonist/antagonist, providing a possible molecular explanation of the tissue-selective effects of glucocorticoids on MR. However, when we examined the coactivator and corepressor peptide interactions in the presence of cortisol, we found that MR bound with cortisol or aldosterone interacted with the same set of peptides. Thus, the partial agonism shown by cortisol is unlikely to be the result of differential interaction with known coactivators and corepressors. On the other hand, we have identified coactivator binding groove mutations that are critical for cortisol activation but not for aldosterone activation, suggesting that the two steroids induce different MR LBD conformations. In addition, we also show that cortisol becomes full agonist when S810L mutation is introduced in the LBD of MR. Interestingly, MR antagonists, such as eplerenone and progesterone, become partial agonist/antagonist of S810L but are still able to recruit LXXLL peptides to the mutant receptor. Together, these findings suggest a model to explain the MR activation by various ligands.     

Hepatic glucocorticoid receptor behaves differently when its hormone binding site is occupied by agonist (triamcinolone acetonide) or antagonist (RU486) steroid ligands

We have examined the influence of sulfhydryl (SH)-group modifying agents on the interaction of the rat liver glucocorticoid receptor (GR) with its known agonist triamcinolone acetonide (TA) and the newly synthesized antagonist mifepristone (RU486). In the freshly prepared cytosol, [3H]TA or [3H]RU486 bound to macromolecule(s) which sediment as 8-9 moieties: the binding of either ligand can be competed with radioinert TA or RU486. The presence of 2-10 mM dithiothreitol (DTT), beta-mercaptoethanol (beta-MER), and monothioglycerol (MTG) caused a 2-3 fold increase in the [3H]TA and [3H]RU486 binding to GR. Iodoacetamide (IA) and N-ethylmaleimide (NEM) decreased the agonist binding significantly. In contrast, the [3H]RU486 binding to GR increased by 50 percent in the presence of IA. IA and NEM inhibited the binding of the heat-transformed [3H]TA-receptor complex to DNA-cellulose by 70-90 percent whereas DNA binding of [3H]RU486-bound GR was inhibited only slightly. These results indicate that either a) the interaction of GR with the agonist or antagonist steroid ligands causes differential structural alterations, which are more readily detectable in the presence of SH-modifying agents or b) the agonist and the antagonist interact with distinct steroid binding sites.     

Activation of rat liver glucocorticoid receptor bound to the antiglucocorticoid RU38486

The kinetics of steroid binding to rat liver glucocorticoid receptor (GR) and receptor denaturation were dependent upon the nature of the molecule occupying GR. Both the agonist [triamcinolone acetonide (TA)] and the antagonist (Ru38486) however competed for the same saturable binding site. Despite opposing physiological action, both steroid analogues permitted receptor activation as evident by binding to DNA-cellulose and 9S to 4S shift on sucrose gradient sedimentation. It therefore seems necessary to reevaluate a current notion that antagonist action of RU38486 in rat liver is a result of impaired receptor activation.     

Glucocorticoid ligands specify different interactions with NF-kappaB by allosteric effects on the glucocorticoid receptor DNA binding domain

Glucocorticoids inhibit inflammation by acting through the glucocorticoid receptor (GR) and powerfully repressing NF-kappaB function. Ligand binding to the C-terminal of GR promotes the nuclear translocation of the receptor and binding to NF-kappaB through the GR DNA binding domain. We sought how ligand recognition influences the interaction between NF-kappaB and GR. Both dexamethasone (agonist) and RU486 (antagonist) promote efficient nuclear translocation, and we show occupancy of the same intranuclear compartment as NF-kappaB with both ligands. However, unlike dexamethasone, RU486 had negligible activity to inhibit NF-kappaB transactivation. This failure may stem from altered co-factor recruitment or altered interaction with NF-kappaB. Using both glutathione S-transferase pull-down and bioluminescence resonance energy transfer approaches, we identified a major glucocorticoid ligand effect on interaction between the GR and the p65 component of NF-kappaB, with RU486 inhibiting recruitment compared with dexamethasone. Using the bioluminescence resonance energy transfer assay, we found that RU486 efficiently recruited NCoR to the GR, unlike dexamethasone, which recruited SRC1. Therefore, RU486 promotes differential protein recruitment to both the C-terminal and DNA binding domain of the receptor. Importantly, using chromatin immunoprecipitation, we show that impaired interaction between GR and p65 with RU486 leads to reduced recruitment of the GR to the NF-kappaB-responsive region of the interleukin-8 promoter, again in contrast to dexamethasone that significantly increased GR binding. We demonstrate that ligand-induced conformation of the GR C-terminal has profound effects on the functional surface generated by the DNA binding domain of the GR. This has implications for understanding ligand-dependent interdomain communication.     

Conserved aromatic residues in the transmembrane region VI of the V1a vasopressin receptor differentiate agonist vs. antagonist ligand binding.

Despite their opposite effects on signal transduction, the nonapeptide hormone arginine-vasopressin (AVP) and its V1a receptor-selective cyclic peptide antagonist d(CH2)5[Tyr(Me)2]AVP display homologous primary structures, differing only at residues 1 and 2. These structural similarities led us to hypothesize that both ligands could interact with the same binding pocket in the V1a receptor. To determine receptor residues responsible for discriminating binding of agonist and antagonist ligands, we performed site-directed mutagenesis of conserved aromatic and hydrophilic residues as well as nonconserved residues, all located in the transmembrane binding pocket of the V1a receptor. Mutation of aromatic residues of transmembrane region VI (W304, F307, F308) reduced affinity for the d(CH2)5[Tyr(Me)2]AVP and markedly decreased affinity for the unrelated strongly hydrophobic V1a-selective nonpeptide antagonist SR 49059. Replacement of these aromatic residues had no effect on AVP binding, but increased AVP-induced coupling efficacy of the receptor for its G protein. Mutating hydrophilic residues Q108, K128 and Q185 in transmembrane regions II, III and IV, respectively, led to a decrease in affinity for both agonists and antagonists. Finally, the nonconserved residues T333 and A334 in transmembrane region VII, controlled the V1a/V2 binding selectivity for both nonpeptide and cyclic peptide antagonists. Thus, because conserved aromatic residues of the V1a receptor binding pocket seem essential for antagonists and do not contribute at all to the binding of agonists, we propose that these residues differentiate agonist vs. antagonist ligand binding.     

Dissociation of steroid receptor coactivator 1 and nuclear receptor corepressor recruitment to the human glucocorticoid receptor by modification of the ligand-receptor interface: the role of tyrosine 735

Within the human glucocorticoid receptor (GR) steroid binding pocket, tyrosine 735 makes hydrophobic contact with the steroid D ring. Substitution of tyrosine735 selectively impairs glucocorticoid transactivation but not transrepression. We now show, using both mammalian two-hybrid and glutathione-S-transferase pull downs, that such substitutions reduce interaction with steroid receptor coactivator 1, both basally and in response to agonist binding. Using a yeast two-hybrid screen we identified one of the three nuclear receptor interacting domains (NCoR-N1) of nuclear receptor corepressor (NCoR) as interacting with the GR C terminus in an RU486-specific manner. This was confirmed in mammalian two-hybrid experiments, and so we used the NCoR-N1 peptide to probe the GR C-terminal conformation. Substitution of Tyr735phe, Tyr735val, and Tyr735 ser, which impaired steroid receptor coactivator 1 (SRC1) interaction, enhanced NCoR-N1 recruitment, basally and after RU486. RU486 did not direct SRC1 recruitment to any of the GR constructs, and dexamethasone did not allow NCoR-N1 recruitment. Using a glutathione-S-transferase pull-down approach, the NCoR-N1 peptide was found to bind the full-length GR constitutively, and no further induction was seen with RU486, but it was reduced by dexamethasone. As both SRC1 and NCoR are predicted to recognize a common hydrophobic cleft in the GR, it seems that changes favorable to one interaction are detrimental to the other, thus identifying a molecular switch.     

Mapping the rho1 GABA(C) receptor agonist binding pocket. Constructing a complete model

Gamma-aminobutyric acid (GABA) is the major inhibitory neurotransmitter in the mammalian brain. The GABA receptor type C (GABA(C)) is a ligand-gated ion channel with pharmacological properties distinct from the GABA(A) receptor. To date, only three binding domains in the recombinant rho1 GABA(C) receptor have been recognized among six potential regions. In this report, using the substituted cysteine accessibility method, we scanned three potential regions previously unexplored in the rho1 GABA(C) receptor, corresponding to the binding loops A, E, and F in the structural model for ligand-gated ion channels. The cysteine accessibility scanning and agonist/antagonist protection tests have resulted in the identification of residues in loops A and E, but not F, involved in forming the GABA(C) receptor agonist binding pocket. Three of these newly identified residues are in a novel region corresponding to the extended stretch of loop E. In addition, the cysteine accessibility pattern suggests that part of loop A and part of loop E have a beta-strand structure, whereas loop F is a random coil. Finally, when all of the identified ligand binding residues are mapped onto a three-dimensional homology model of the amino-terminal domain of the rho1 GABA(C) receptor, they are facing toward the putative binding pocket. Combined with previous findings, a complete model of the GABA(C) receptor binding pocket was proposed and discussed in comparison with the GABA(A) receptor binding pocket.     

Molecular interactions of nonpeptide agonists and antagonists with the melanocortin-4 receptor

The melanocortin-4 (MC4) receptor is a potential therapeutic target for obesity and cachexia, for which nonpeptide agonists and antagonists are being developed, respectively. The aim of this study was to identify molecular interactions between the MC4 receptor and nonpeptide ligands, and to compare the mechanism of binding between agonist and antagonist ligands. Nonpeptide ligand interaction was affected by mutations that reduce peptide ligand binding (D122A, D126A, S190A, M200A, F261A, and F284A), confirming overlapping binding determinants for peptide and nonpeptide ligands. The common halogenated phenyl group of nonpeptide ligands was a determinant of F261A and F284A mutations' affinity-reducing effect, implying this group interacts with the aromatic side chains of these residues. All affected compounds contain this group, the mutations reduced binding of 2,4-dichloro-substituted compounds more than 4-chloro-substituted-compounds, and F284A mutation eliminated the affinity-enhancing effect of 2-chloro-substitution. F261A and F284A mutations reduced the affinity of antagonists more than agonists, suggesting that the stronger ligand interaction with these residues, the lower the ligand efficacy. Supporting this hypothesis, F261A mutation increased the efficacy of nonpeptide antagonist and partial agonist ligands. D122A and D126A mutations reduced nonpeptide ligand interaction. Removing the ligands' derivatized amide group eliminated the effect of the mutations. Interaction of agonists, which bear a common amine within this group, was strongly reduced by D126A mutation (550-3300-fold), suggesting an electrostatic interaction between the amine and the acidic group of D126. These postulated interactions with aromatic and acidic regions of the MC4 receptor are consistent with a molecular model of the receptor. Furthermore, the strength of interaction with the aromatic pocket, and potentially the acidic pocket, controls the signaling efficacy of the ligand.