Design of thyroid hormone receptor antagonists from first principles

It is desirable to obtain TR antagonists for treatment of hyperthyroidism and other conditions. We have designed TR antagonists from first principles based on TR crystal structures. Since agonist ligands are buried in the fold of the TR ligand binding domain (LBD), we reasoned that ligands that resemble agonists with large extensions should bind the LBD, but would prevent its folding into an active conformation. In particular, we predicted that extensions at the 5′ aryl position of ligand should reposition helix (H) 12, which forms part of the co-activator binding surface, and thereby inhibit TR activity. We have found that some synthetic ligands with 5′ aryl ring extensions behave as antagonists (DIBRT, NH-3), or partial antagonists (GC-14, NH-4). Moreover, one compound (NH-3) represents the first potent TR antagonist with nanomolar affinity that also inhibits TR action in an animal model. However, the properties of the ligands also reveal unexpected aspects of TR behavior. While nuclear receptor antagonists generally promote binding of co-repressors, NH-3 blocks co-activator binding and also prevents co-repressor binding. More surprisingly, many compounds with extensions behave as full or partial agonists. We present hypotheses to explain both behaviors in terms of dynamic equilibrium of H12 position.     

Molecular dynamics simulations reveal multiple pathways of ligand dissociation from thyroid hormone receptors

Nuclear receptor (NR) ligands occupy a pocket that lies within the core of the NR ligand-binding domain (LBD), and most NR LBDs lack obvious entry/exit routes upon the protein surface. Thus, significant NR conformational rearrangements must accompany ligand binding and release. The precise nature of these processes, however, remains poorly understood. Here, we utilize locally enhanced sampling (LES) molecular dynamics computer simulations to predict molecular motions of x-ray structures of thyroid hormone receptor (TR) LBDs and determine events that permit ligand escape. We find that the natural ligand 3,5,3'-triiodo-L-thyronine (T(3)) dissociates from the TRalpha1 LBD along three competing pathways generated through i), opening of helix (H) 12; ii), separation of H8 and H11 and the Omega-loop between H2 and H3; and iii), opening of H2 and H3, and the intervening beta-strand. Similar pathways are involved in dissociation of T(3) and the TRbeta-selective ligand GC24 from TRbeta; the TR agonist IH5 from the alpha- and beta-TR forms; and Triac from two natural human TRbeta mutants, A317T and A234T, but are detected with different frequencies in simulations performed with the different structures. Path I was previously suggested to represent a major pathway for NR ligand dissociation. We propose here that Paths II and III are also likely ligand escape routes for TRs and other NRs. We also propose that different escape paths are preferred in different situations, implying that it will be possible to design NR ligands that only associate stably with their cognate receptors in specific cellular contexts.     

Selective modulation of thyroid hormone receptor action

Thyroid hormones have some actions that might be useful therapeutically, but others that are deleterious. Potential therapeutically useful actions include those to induce weight loss and lower plasma cholesterol levels. Potential deleterious actions are those on the heart to induce tachycardia and arrhythmia, on bone to decrease mineral density, and on muscle to induce wasting. There have been successes in selectively modulating the actions of other classes of hormones through various means, including the use of pharmaceuticals that have enhanced affinities for certain receptor isoforms. Thus, there is reason to pursue selective modulation of thyroid hormone receptor (TR) function, and several agents have been shown to have some β-selective, hepatic selective and/or cardiac sparring activities, although development of these was largely not based on detailed understanding of mechanisms for the specificity. The possibility of selectively targeting the TRβ was suggested by the findings that there are - and β-TR forms and that the TR-forms may preferentially regulate the heart rate, whereas many other actions of these hormones are mediated by the TRβ. We determined X-ray crystal structures of the TR and TRβ ligand-binding domains (LBDs) complexed with the thyroid hormone analog 3,5,3′-triiodithyroacetic acid (Triac). The data suggested that a single amino acid difference in the ligand-binding cavities of the two receptors could affect hydrogen bonding in the receptor region, where the ligand's 1-position substituent fits and might be exploited to generate β-selective ligands. The compound GC-1, with oxoacetate in the 1-position instead of acetate as in Triac, exhibited TRβ-selective binding and actions in cultured cells. An X-ray crystal structure of the GC-1-TRβ LBD complex suggests that the oxoacetate does participate in a network of hydrogen bonding in the TR LBD polar pocket. GC-1 displayed actions in tadpoles that were TRβ-selective. When administered to mice, GC-1 was as effective in lowering plasma cholesterol levels as T₃, and was more effective than T₃ in lowering plasma triglyceride levels. At these doses, GC-1 did not increase the heart rate. GC-1 was also less active than T₃ in modulating activities of several other cardiac parameters, and especially a cardiac pacemaker channel such as HCN-2, which may participate in regulation of the heart rate. GC-1 showed intermediate activity in suppressing plasma thyroid stimulating hormone (TSH) levels. The tissue/plasma ratio for GC-1 in heart was also less than for the liver. These data suggest that compounds can be generated that are TR-selective and that compounds with this property and/or that exhibit selective uptake, might have clinical utility as selective TR modulators.     

Hormone selectivity in thyroid hormone receptors

Separate genes encode thyroid hormone receptor subtypes TRalpha (NR1A1) and TRbeta (NR1A2). Products from each of these contribute to hormone action, but the subtypes differ in tissue distribution and physiological response. Compounds that discriminate between these subtypes in vivo may be useful in treating important medical problems such as obesity and hypercholesterolemia. We previously determined the crystal structure of the rat (r) TRalpha ligand-binding domain (LBD). In the present study, we determined the crystal structure of the rTRalpha LBD in a complex with an additional ligand, Triac (3,5, 3'-triiodothyroacetic acid), and two crystal structures of the human (h) TRbeta receptor LBD in a complex with either Triac or a TRbeta-selective compound, GC-1 [3,5-dimethyl-4-(4'-hydroy-3'-isopropylbenzyl)-phenoxy acetic acid]. The rTRalpha and hTRbeta LBDs show close structural similarity. However, the hTRbeta structures extend into the DNA-binding domain and allow definition of a structural "hinge" region of only three amino acids. The two TR subtypes differ in the loop between helices 1 and 3, which could affect both ligand recognition and the effects of ligand in binding coactivators and corepressors. The two subtypes also differ in a single amino acid residue in the hormone-binding pocket, Asn (TRbeta) for Ser (TRalpha). Studies here with TRs in which the subtype-specific residue is exchanged suggest that most of the selectivity in binding derives from this amino acid difference. The flexibility of the polar region in the TRbeta receptor, combined with differential recognition of the chemical group at the 1-carbon position, seems to stabilize the complex with GC-1 and contribute to its beta-selectivity. These results suggest a strategy for development of subtype-specific compounds involving modifications of the ligand at the 1-position.     

Low-resolution structures of thyroid hormone receptor dimers and tetramers in solution

High-resolution X-ray structures of thyroid hormone (TH) receptor (TR) DNA and ligand binding domains (DBD and LBD) have yielded significant insights into TR action. Nevertheless, the TR DBD and LBD act in concert to mediate TH effects upon gene expression, and TRs form multiple oligomers; however, structures of full-length TRs or DBD-LBD constructs that would clarify these influences are not available. Here, we report low-resolution X-ray structures of the TRbeta DBD-LBD construct in solution which define the shape of dimers and tetramers and likely positions of the DBDs and LBDs. The holo TRbeta DBD-LBD construct forms a homodimer with LBD-DBD pairs in close contact and DBDs protruding from the base in the same direction. The DBDs are connected to the LBDs by crossed extended D domains. The apo hTRbeta DBD-LBD construct forms tetramers that resemble bulged cylinders with pairs of LBD dimers in a head-to-head arrangement with DBD pairs packed tightly against the LBD core. Overall, there are similarities with our previous low-resolution structures of retinoid X receptors, but TRs exhibit two unique features. First, TR DBDs are closely juxtaposed in the dimer and tetramer forms. Second, TR DBDs are closely packed against LBDs in the tetramer, but not the dimer. These findings suggest that TRs may be able to engage in hitherto unknown interdomain interactions and that the D domain must rearrange in different oligomeric forms. Finally, the data corroborate our suggestion that apo TRs form tetramers in solution which dissociate into dimers upon hormone binding.     

Structural insights into ligand-binding pocket formation in Nurr1 by molecular dynamics simulations

Abstract

The nuclear receptor Nurr1 (NR4A2) has been identified as a potential target for the treatment of Parkinson’s disease. In contrast to most other nuclear receptors, the X-ray crystal structure of the Nurr1 ligand-binding domain (LBD) lacks any ligand-binding pocket (LBP). However, NMR spectroscopy measurements have revealed that the known Nurr1 agonist docosahexaenoic acid (DHA) binds to a region within the LBD that corresponds to the classical NR ligand-binding pocket (LBP). In order to investigate the structural dynamics of the Nurr1 LBD and to study potential LBP formation, the conformational space of the receptor was sampled using a molecular dynamics (MD) simulation. Docking of DHA into 50,000 LBD structures extracted from the simulation revealed the existence of a transient LBP that is capable to fully harbor the compound. The location of the identified pocket overlaps with the ligand-binding site suggested by NMR experiments. Structural analysis of the protein-ligand complex showed that only modest structural rearrangements within the Nurr1 LBD are required for LBP formation. These findings may support structure-based drug discovery campaigns for the development of receptor-specific agonists.     

The loss of an electrostatic contact unique to AMPA receptor ligand binding domain 2 slows channel activation

Ligand-gated ion channels undergo conformational changes that transfer the energy of agonist binding to channel opening. Within ionotropic glutamate receptor (iGluR) subunits, this process is initiated in their bilobate ligand binding domain (LBD) where agonist binding to lobe 1 favors closure of lobe 2 around the agonist and allows formation of interlobe hydrogen bonds. AMPA receptors (GluAs) differ from other iGluRs because glutamate binding causes an aspartate-serine peptide bond in a flexible part of lobe 2 to rotate 180° (flipped conformation), allowing these residues to form cross-cleft H-bonds with tyrosine and glycine in lobe 1. This aspartate also contacts the side chain of a lysine residue in the hydrophobic core of lobe 2 by a salt bridge. We investigated how the peptide flip and electrostatic contact (D655-K660) in GluA3 contribute to receptor function by examining pharmacological and structural properties with an antagonist (CNQX), a partial agonist (kainate), and two full agonists (glutamate and quisqualate) in the wildtype and two mutant receptors. Alanine substitution decreased the agonist potency of GluA3(i)-D655A and GluA3(i)-K660A receptor channels expressed in HEK293 cells and differentially affected agonist binding affinity for isolated LBDs without changing CNQX affinity. Correlations observed in the crystal structures of the mutant LBDs included the loss of the D655-K660 electrostatic contact, agonist-dependent differences in lobe 1 and lobe 2 closure, and unflipped D(A)655-S656 bonds. Glutamate-stimulated activation was slower for both mutants, suggesting that efficient energy transfer of agonist binding within the LBD of AMPA receptors requires an intact tether between the flexible peptide flip domain and the rigid hydrophobic core of lobe 2.     

Structural rearrangements in the thyroid hormone receptor hinge domain and their putative role in the receptor function

The thyroid hormone receptor (TR) D-domain links the ligand-binding domain (LBD, EF-domain) to the DNA-binding domain (DBD, C-domain), but its structure, and even its existence as a functional unit, are controversial. The D domain is poorly conserved throughout the nuclear receptor family and was originally proposed to comprise an unfolded hinge that facilitates rotation between the LBD and the DBD. Previous TR LBD structures, however, have indicated that the true unstructured region is three to six amino acid residues long and that the D-domain N terminus folds into a short amphipathic alpha-helix (H0) contiguous with the DBD and that the C terminus of the D-domain comprises H1 and H2 of the LBD. Here, we solve structures of TR-LBDs in different crystal forms and show that the N terminus of the TRalpha D-domain can adopt two structures; it can either fold into an amphipathic helix that resembles TRbeta H0 or form an unstructured loop. H0 formation requires contacts with the AF-2 coactivator-binding groove of the neighboring TR LBD, which binds H0 sequences that resemble coactivator LXXLL motifs. Structural analysis of a liganded TR LBD with small angle X-ray scattering (SAXS) suggests that AF-2/H0 interactions mediate dimerization of this protein in solution. We propose that the TR D-domain has the potential to form functionally important extensions of the DBD and LBD or unfold to permit TRs to adapt to different DNA response elements. We also show that mutations of the D domain LXXLL-like motif indeed selectively inhibit TR interactions with an inverted palindromic response element (F2) in vitro and TR activity at this response element in cell-based transfection experiments.     

Dynamic stabilization of nuclear receptor ligand binding domains by hormone or corepressor binding

We have developed a novel assembly assay to examine structural changes in the ligand binding domain (LBD) of the thyroid hormone receptor (TR). Fragments including the first helix of the TR LBD interact only weakly with the remainder of the LBD in the absence of hormone, but this interaction is strongly enhanced by the addition of either hormone or the corepressor NCoR. Since neither the ligand nor the corepressor shows direct interaction with this helix, we propose that both exert their effects by stabilizing the overall structure of the LBD. Current models of activation of nuclear hormone receptors focus on a ligand-induced allosteric shift in the position of the C-terminal helix 12 that generates the coactivator binding site. Our results suggest that ligand binding also has more global effects that dynamically alter the structure of the receptor LBD.