|Home | About | Journals | Submit | Contact Us | Français|
Some nuclear receptor (NR) ligands promote dissociation of radiolabeled bound hormone from the buried ligand binding cavity (LBC) more rapidly than excess unlabeled hormone itself. This result was interpreted to mean that challenger ligands bind allosteric sites on the LBD to induce hormone dissociation, and recent findings indicate that ligands bind weakly to multiple sites on the LBD surface. Here, we show that a large fraction of thyroid hormone receptor (TR) ligands promote rapid dissociation (T1/2 <2 hours) of radiolabeled T3 versus T3 (T1/2 ≈5–7 hours). We cannot discern relationships between this effect and ligand size, activity or affinity for TRβ. One ligand, GC-24, binds the TR LBC and (weakly) to the TRβ-LBD surface that mediates dimer/heterodimer interaction, but we cannot link this interaction to rapid T3 dissociation. Instead, several lines of evidence suggest that the challenger ligand must interact with the buried LBC to promote rapid T3 release. Since previous molecular dynamics simulations suggest that TR ligands leave the LBC by several routes, we propose that a subset of challenger ligands binds and stabilizes a partially unfolded intermediate state of TR that arises during T3 release and that this effect enhances hormone dissociation.
Nuclear Receptors (NRs) regulate gene expression in response to small signaling molecules . The NR family includes receptors for thyroid hormone (TH) , steroid hormones, vitamins A and D, cholesterol and fatty acid derivatives, heme, glucose and other molecules. Since NRs play widespread roles in development and disease, they are important targets for pharmaceutical discovery. TH receptors (TRs) are the subject of efforts to develop selective agonists to ameliorate aspects of metabolic syndrome without harmful effects on heart and antagonists to treat hyperthyroidism and other conditions [3, 4]. Improved understanding of mechanisms of NR ligand association and dissociation will provide insights into receptor function and could suggest ways to stabilize or destabilize bound hormone, improve antagonism and facilitate development of drugs that interact tightly and selectively with cognate NRs.
NRs harbor a single ligand binding cavity (LBC) whose location, relationship to gene activation and organization has been extensively studied [1, 5, 6]. X-ray structures of NR LBDs with agonists reveal the LBC is buried in the C-terminal ligand binding domain (LBD). Agonists promote packing of C-terminal helix (H) 12 against the LBD to complete a coactivator binding surface, activation function 2 (AF-2) [5, 7]. Close investigation of the LBCs of the two TRs (TRα and TRβ) revealed one subtype specific amino acid in the TR LBC involved in ligand contact (TRβN331/TRαS277) and it has been possible to exploit this difference to obtain TRβ selective ligands [3, 8]. X-ray structures also reveal that the buried pocket is flexible; the TR LBC can expand to accommodate a bulky 5′ iodine substituent in the parental form of TH, thyroxine (T4), and a bulky 3′ phenyl group in the TRβ selective agonist, GC-24 [6, 9, 10].
In contrast, mechanisms of ligand binding and dissociation from the LBC are only partly understood [11, 12]. X-ray structures of NR LBDs reveal that H12 can move to expose the LBC, and this probably constitutes one ligand escape route [5, 7, 11, 12]. However, our analyses of regions of instability in X-ray structures [13–15] and molecular dynamics simulations [11, 12] suggest that active TH (triiodothyronine, T3) can escape from the LBD in three ways: under H12 (Path I, described above); between H8 and H11 near the dimer/heterodimer surface at the H10/H11 junction (Path II); and through the H1-H3 loop (Path III). Our simulations also suggest that escape routes vary with ligand and receptor; T3 prefers Path III whereas the TRβ-selective GC-24 prefers Path I .
While the notion that there are multiple ligand escape paths awaits definitive verification, a number of data are consistent with this conclusion. Structural elements that permit ligand escape through each pathway are implicated in stable agonist binding [11, 12]. For Path I, suboptimal packing of TRβ H12 against the T4–LBD complex is associated with rapid ligand dissociation . Conversely, point mutations and coactivators that stabilize estrogen receptor (ER) or TR H12 in the active position reduce hormone dissociation rates [16–18]. For Path II (involving residues near the dimer surface), resistance to thyroid hormone syndrome (RTH) mutations that affect this region enhance T3 dissociation rates . For Path III, X-ray structural analysis of other RTH mutants reveals that increased T3 dissociation rate is associated with disorder in the H1-H3 loop [14, 15, 19].
In spite of strong evidence for a single high affinity hormone binding site, early studies raised the possibility that NRs harbor auxiliary ligand binding sites that exert allosteric effects on bound hormone. Some NR interacting compounds displace bound high affinity ligands more rapidly than the high affinity ligand itself. Progesterone, for example, binds glucocorticoid receptors (GRs) with lower affinity than dexamethasone, and acts as an antagonist, yet displaces this higher affinity agonist more rapidly than dexamethasone . The mechanism of this effect is not clear, but it was proposed that progesterone binds to an undefined allosteric site to promote dexamethasone dissociation.
Recent evidence confirmed that there are multiple ligand binding sites on NR LBD surfaces. Several compounds bind to NR (including TR) AF-2 sites [21–23]. Other compounds were found at another location on the androgen receptor surface (BF-3) in X-ray screens and ligand binding to BF-3 may exert allosteric effects on AR AF-2 . Finally, the TR agonist GC-24 binds to at a location near the TR dimer/heterodimer surface at the junction of TRβ H10 and H11 .
In this paper, we show that several TR ligands (including GC-24) displace bound hormone at different rates and investigate this phenomenon. The effect is not related to ligand affinity, activity or size and does not appear to involve surface ligand interactions. Instead, several lines of evidence suggest that the challenger interacts with the LBC to displace bound hormone. We propose that challengers promote ligand release by binding partially unfolded conformational intermediates that occur in ligand release and blocking refolding of the hormone/receptor complex around labeled ligand. Implicit in this hypothesis is the concept that different ligands associate with TRs via different pathways.
Expression vectors for TRs (CMX-TRβ, CMX-TRα), TRβ mutants (CMX-TRβP419R, L422R, M423R, N331S) and the TRα mutant (TRαS277N) are described [8, 19]. TRs were expressed in TNT T7 Quick in vitro coupled transcription/translation kits, according to manufacturer’s protocols (Promega, Madison, WI).
T3 binding affinities were determined by saturation binding assays . Approximate amounts of TRs were determined by measurement of T3 binding activity in single point binding assays; TR preparations were incubated overnight at 4°C with 1 nM L-3,5,3′-125I-T3 (NEN Life Science Products) in 100μl binding buffer (400 mM NaCl, 20mM KPO4, pH 8, 0.5 mM EDTA, 1.0 mM MgCl2, 10% glycerol) containing 1mM monothioglycerol and 50μg calf thymus histones (Calbiochem). Bound 125I-T3 was separated from free ligand by gravity flow through a 2ml course Sephadex G-25 column (Pharmacia Biotech) and quantified on a γ-counter (COBRA, Packard Instruments, Meriden, CT). The number of binding sites per unit volume were calculated from specific activity of radiolabeled T3 (3824cpm = 1fmol). For saturation binding, 10–20 fmols of TR protein were incubated overnight at 4°C with varying concentrations of 125I-T3. Amount of 125I-T3 was verified by precount in each aliquot, prior to addition of protein. Next morning, bound vs. free 125I-T3 was determined by passage over the Sephadex G-25 column, as above. In these conditions, non-specific binding of 125I-T3 to unprogrammed reticulate lysates was negligible; > 1% observed in the presence of 20fmols TRs, as was residual binding of 1nM 125I-T3 obtained with a 1000-fold excess of unlabeled T3 (not shown). T3 applied to the column in the absence of TRs only dissociates after several hours of washing, and does not contribute to measurements of bound T3 (not shown). Thus, most (>99%) of labeled ligand that passes through the Sephadex G-25 column corresponds to TR bound to T3. Kd values were calculated by fitting saturation curves to the equations of Swillens using the GraphPad Prism program (GraphPad Software V3.03, San Diego, CA).
T3 association (kon) and dissociation (koff) rates were determined using methods similar to saturation binding assays, with the following modifications. For koff, TRs were incubated overnight with saturating (1nM) 125I-T3 at 4°C [9, 19]. Unlabeled T3 or challenger was added to a final concentration of 1 μM (1000-fold excess) the following morning and aliquots were taken at various times and applied to Sephadex G-25 columns to determine how rapidly 125I-T3 dissociates from TR. Binding curves and koff values were calculated using the GraphPad Prism one phase exponential decay model. For kon, unliganded TR preparations were added to binding buffer containing 1.5 nM 125I-T3 to a final concentration of 20 fmols TRs per 100 μl of buffer. 100 μl aliquots were then applied at various times to Sephadex G25 columns to separate bound from unbound T3. In these conditions, T3 is in excess of receptor, only about 10% of T3 present in the initial mix associates with the TR at equilibrium and the remainder remains unbound. Binding curves and kon values were calculated, where possible, by non-linear regression analysis using one and two phase association growth models with Graph Pad Prism Software. The program identifies the best fit (one/two phase) for each curve.
Binding of TR to TREs were assayed by mixing 20 fmols of 35S labeled TRs produced in a reticulocyte lysate, (TNT T7; Promega) with 10ng oligonucleotide and 1 μg poly(dI-dC) (Amersham Pharmacia Biotech) in final volume of 20 μl 1X binding buffer (25 mM HEPES pH 7.5, 50 mM KCl, 1 mM DTT, 10 μM ZnSO4, 0.1% NP-40, 5% glycerol). After 30′ incubation, the mixture was loaded onto a 5% nondenaturing polyacrylamide gel that was pre-run for 30 minutes at 200 V and run at 4°C for 120 minutes at 240 V, in a running buffer of 6.7 mM Tris (pH 7.5), 1 mM EDTA, and 3.3 mM sodium acetate. The gel was then fixed, treated with Amplify (Amersham Pharmacia Biotech), dried and exposed for autoradiography. TRs used in assay were quantified with 125I-T3 binding assay and SDS-PAGE analysis of 35S-TRs.
Full-length hRXRα was prepared in E. coli BL21 as a fusion with glutathione S-transferase (GST) as per the manufacturer’s protocol (Amersham Pharmacia Biotech). The bindings were performed by mixing glutathione-linked Sepharose beads containing 4 μg of GST fusion proteins (Coomassie Plus protein assay reagent, Pierce) with 1–2 μl of the 35S-labeled wild-type hTRβ in 150 μl of binding buffer (20 mM HEPES pH 7.5, 150 mM KCl, 25 mM MgCl2, 10% glycerol, 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, and protease inhibitors) containing 20 μg/ml bovine serum albumin for 1.5 h. Beads were washed three times with 200 μl of binding buffer, and the bound proteins were separated using 10% SDS-polyacrylamide gel electrophoresis and visualized by autoradiography.
We examined abilities of different TR ligands to displace bound hormone from the TRβ LBC in kinetic studies [14, 15, 19]. For these assays (in schematic in Fig. 1A), TR preparations were incubated overnight with saturating labeled T3 (1 nM) to allow stable hormone-TR complex formation and challenged with excess unlabeled ligand to prevent reassociation with radiolabeled T3. Displacement of radiolabeled hormone was monitored by size exclusion chromatography, which separates T3-TR complexes from free hormone.
TRβ selective agonists vary in effects on T3 displacement (Fig. 1B). The half-life (T1/2) of the TRβ-T3 complex at 4°C was 5–6 hours; consistent with previous measurements . Similar values were obtained when the TRβ-T3 complex was challenged with excess GC-1, a synthetic TRβ selective ligand that binds TRβ with similar affinity to T3 . However, T3 dissociated more rapidly in the presence of excess GC-24 (T1/2 varied between 30 and 120 minutes over these studies). Differential effects of GC-1 and GC-24 on T3 release were not related to affinities for TR; all three agonists displaced T3 with similar Kd values (0.1–0.15 nM inferred from Ki determinations, see Methods) in equilibrium hormone binding assays, in accordance with previous results (not shown).
Other TR ligands promoted rapid T3 release. We previously described GC-1 derivatives with bulky 5′ extensions that exhibit diverse activities, from full agonist to full antagonist, and bind TRs with a range of affinities [25, 26]. Of this series, eight of ten compounds displaced T3 more rapidly than native hormone (Fig. 2, GC-24 with a 3′ extension is also shown as a reference). This effect was not obviously related to ligand activity; rapid T3 release was observed with an agonist (NH-1), partial agonists (GC-14, NH-2, NH-6) and full antagonists (NH-3, NH-5, NH-7 and HY-4). Rapid T3 release was unrelated to affinity; the same phenomenon was observed with compounds that bind TRs with low and high affinity. For example, HY-4 (Kd=146 nM) displaced T3 as rapidly as NH-2 (Kd=0.52 nM).
TR interacting compounds that lack large extensions also varied in their ability to displace bound hormone (Fig. 3A). The synthetic TR agonist DIMIT displaced T3 at the same rate as T3 itself, but three other TH derivatives displaced T3 more rapidly than T3. These were: a) Triac, a low abundance active TH that binds TR with high affinity and is produced by deamination of thyroid hormone in the liver ; b) Thyroxine, T4, the parental form of thyroid hormone which binds TR with moderate affinity ; and c) Reverse T3, a product of thyroid hormone metabolism that binds weakly to TRs and acts as a partial agonist . Ligands that bind other NRs with high affinity, but not TRs, did not enhance T3 dissociation, including progesterone, testosterone, the synthetic androgen R1881, the mineralocorticoid receptor antagonist spironolactone and estradiol (Fig. 3B).
TRs exist as a mix of dimers and monomers in solution with T3 promoting monomer formation. Since the GC-24 surface binding site lies near the dimer surface , we examined the possibility that GC-24 interactions with this surface were involved in rapid T3 dissociation. In accordance with previous results [19, 28], mutant TRs that only form monomers (P419R, L422R, M423R) bound T3 with similar affinity to wild type TRs (data not shown). However, dimer surface mutations did alter ligand binding kinetics; T3 dissociated more rapidly from TRβL422R than wild type TRβ (Fig. 4A) and TRβL422R also exhibited increased rates of T3 association (Fig. 4B), with more than half the mutant TRs occupied by T3 within five minutes. Similar results were obtained with TRβP419R and TRβM423R that impair dimer and heterodimer formation (Fig. 4C and not shown). This resembles studies with ERs, which showed that estradiol dissociated more rapidly from monomers than homodimers . Thus, mutations in the TR dimerization surface affect ligand binding kinetics. However, GC-24 continued to promote rapid release of labeled T3 from each TR mutant that exists as an obligate monomer (Fig. 4C). This implies that GC-24 does not increase TR ligand dissociation by blocking residual TR-TR dimer interactions that occur in the presence of this ligand.
RXR-TR heterodimer formation involves the same TR surface that mediates homodimer formation, including residues implicated in GC-24 surface contact . However, by contrast to TR-TR homodimer formation, RXR-TR heterodimer formation is not affected by hormone [2, 19]. RXR did not affect the rate of radiolabeled T3 dissociation, in the presence of T3 or GC-24 (Fig. 5A). Control assays confirm that RXR-TRs to form in these conditions and that GC-24 did not disrupt heterodimer formation in these conditions, either in gel shift assays on DNA or pulldown assays in solution (Fig. 5B and 5C). Thus, assuming that RXR heterodimer formation does occlude the surface GC-24 binding site, our data suggest that GC-24 interactions at the TR surface are not necessary for increased rates of T3 release.
To test whether effects of ligands on T3 release were TR isoform-specific, we compared effects of different challengers on TRβ and TRα. Radiolabeled T3 dissociated from both TRs at similar rates T1/2 ≈ 4–5 hours; Fig. 6). Similar results were obtained with GC-1, even though this ligand binds more tightly to TRβ. By contrast, GC-24 (which binds TRβ about 40–100 times more tightly than TRα ) promoted more rapid release of radiolabeled T3 from TRβ than T3, but not from TRβ. Similar results were obtained with the weakly TRβ selective antagonist NH-3 (not shown).
Interestingly, TR-isoform specific effects of GC-24 on T3 dissociation were partly sensitive to mutation of the buried LBC . T3 dissociation was not altered by mutations that reverse the LBC subtype specific residue (TRβN331S and TRαS277N) in response to excess T3 or GC-1 (Fig. 6). However, the TRα S277N mutant (which converts the LBC to that of TRβ) exhibited more rapid T3 dissociation than native TRα in the presence of GC-24. The converse TRβN331S mutant (which converts the TRβ LBC cavity to that of TRα) did not reduce T3 dissociation rates. Thus, a strongly TRβ-selective challenger ligand (GC-24) exerts TRβ selective effects on release of a non-TR isoform selective hormone, T3, and this effect is partly sensitive to mutation of the buried pocket.
Finally, we assessed association rates of T4 with TRs. This ligand binds TR with relatively low affinity vs. T3 and dissociates rapidly from both TRs yet also displaces radiolabeled T3 more rapidly from TRβ (Fig. 3) and TRα (not shown) than T3 . Fig. 7 reveals that T4 associates very rapidly with TRs; whereas half the TRs were occupied with T4 within two minutes, more than fifty minutes were needed to obtain similar levels of TR occupancy with T3. Thus, a compound that displaces T3 rapidly from TR also associates rapidly with the TR.
In this study, we examined the basis of an observation that was made in the 1970s , ligands (challengers) that bind NRs with low affinity displace radiolabeled bound ligands more rapidly than non-labeled versions of bound hormone itself. Since early hypotheses suggested that the low affinity challenger interacts with an undefined allosteric site to promote hormone release, and emerging evidence confirms that NR ligands weakly interact with the LBD surface at functionally important sites, we tested whether this phenomenon could be observed with TRs and whether we could understand the effect in terms of recent evidence about TR structure, function, ligand interaction and dynamics.
A large subset of ligands that bind to TR displace bound T3 more rapidly than T3 itself. Generally, T1/2 for the TR-T3 complex varied between 5–7 hours in response to T3 challenge. Of seventeen TR interacting compounds investigated, twelve displaced T3 with T1/2 from 20 minutes to 2 hours. There is no obvious correlation between effects of TR ligands and their affinity for TR, activity or molecular weight. More rapid T3 dissociation was observed with compounds that bind TRβ tightly (GC-24, Kd = 0.07nM) or weakly (rT3, Kd = 393nM), with agonists (GC-24, NH-1, Triac, T4 and rT3), partial agonists (GC-14, NH-2, NH-6 and NH-8) and antagonists (NH-1, NH-3, NH-5, NH-7 and HY-4) and with compounds that are of similar size to T3 (Triac and rT3) or contain bulky extension groups (GC-24, GC-14, the NH series, HY-4 and T4). However, compounds that bind to other NRs did not enhance T3 dissociation relative to T3 challenger, including one compound (progesterone) which displaces dexamethasone rapidly from GR even though it interacts weakly with the GR LBC . Thus, our data suggest that only compounds that bind to the TR LBC enhance T3 dissociation rates.
Since one of the ligands that rapidly displaces T3 from the LBC, GC-24, was found at a site in the vicinity of the TR-TR dimerization and TR-RXR heterodimerization surface , we examined the possibility that surface interactions could influence T3 dissociation rates. However, RXR, which should occlude the site through heterodimer formation, fails to alter effects of GC-24 on T3 dissociation rates. Moreover, mutations in the TR dimer-/heterodimer surface enhance T3 dissociation rates, but do not abolish GC-24 effects. In addition, we have not observed other compounds that promote rapid T3 dissociation (including Triac) at surface sites in our structures . Thus, we do not think that surface ligand binding to the dimer surface explains rapid T3 dissociation.
Why does T3 dissociation rate vary with different challenger ligands? Three lines of evidence suggest that challenger ligands interact with the buried LBC to promote T3 release. As mentioned above, there is no correlation between affinity of the challenger ligand for TR and its ability to displace T3 in kinetic studies with TRβ, but only challenger ligands that are T3 analogues are effective. Ligands that interact with other NRs fail to enhance T3 dissociation. Moreover, two TRβ selective challengers (GC-24 and NH-3) promote rapid T3 dissociation (relative to T3) from TRβ but not TRα and this effect is partly sensitive to mutation of the TR LBC. Finally, T4, which binds TRs with low affinity, induces rapid T3 dissociation and associates with the TR much more rapidly than T3. This finding suggests that a ligand that displaces T3 rapidly from TRs also binds rapidly to TRs. Presently, we have not been able to examine association rates of other ligands due to the lack of availability of radiolabeled compounds but we predict that variations in ligand association rates will be common. Together; our data implies that rapid T3 dissociation observed with selected challenger ligands is associated with processes involved in binding of these ligands to the buried LBC.
As described in the Introduction, we have suggested that there are multiple ligand entry and exit pathways for TRs and other NRs and we propose that our findings can be explained in terms of differential utilization of entry and exit pathways (Fig. 8). Our MD simulations suggest that the TR LBD is a highly mobile protein and that bound T3 is constantly probing potential escape routes on the receptor surface [11, 12]. Usually, escape routes close before T3 release, but T3 can also escape from partially unfolded intermediates before the LBD refolds into the active state. We have also found that preference of escape route varies with receptor oligomeric state and ligand [11, 12, 29]. Thus, for standard T3 dissociation assays (Fig. 8A), we predict that the TR-T3 complex constantly rearranges and unfolds to open ligand exit routes (a). At this point, labeled T3 can dissociate (b) and be sequentially replaced by unlabeled T3 (c), or the hormone-receptor complex refolds (a′). We propose that alternate ligands with preferences fordifferent entry/exit routes bind the partially unfolded TR intermediate complex before T3 leaves, blocking T3 re-entry into the LBC and promoting rapid hormone dissociation (Fig. 8B). Accordingly, our previous simulations with T3 and GC-24 reveal strong energetic differences in pathway utilization; whereas T3 prefers Path III (through the H1-H3 loop) GC-24 prefers Path I (under the H12 loop). In this event, the second ligand will inhibit refolding of the T3-TR complex and enhance T3 dissociation through step b.
Our model suggests explanations for several puzzling observations. First, it explains how the challenger interacts with an inaccessible LBC to promote T3 release; TR will partly unfold to expose entry routes to the pocket. Second, it explains why effects of the challenger ligands correlate poorly with their affinities for TR; the key interaction involves a partially unfolded TR and not the native receptor observed in X-ray structures. Finally, our model explains why T4 associates with TRs more rapidly than T3; different ligands bind TRs in different ways. Our model does not predict detailed molecular events involved in ligand escape or conformations of partially unfolded intermediate states. However, we think that this model accounts for many aspects of previous observations about ligand release.
Our data also support to the notion that there may be different modes of ligand escape from TRs and that patterns of TR ligand association and dissociation resemble other NRs. As discussed earlier, ER dimerization reduces hormone dissociation rates and our MD simulations suggest that this may be because pathways of ligand release are occluded in the dimer . We find here that ligand association and release rates are elevated in TR mutants that are obligate monomers. Perhaps dimerization selectively occludes TR ligand association and release pathways either directly (Path II involves H8 and H11 near the dimer surface) or indirectly through stabilization of LBD conformation. Improved understanding of relationships between NR LBD surfaces and hormone binding to the buried LBC could help us explain variations in ligand release pathways and exploit these findings in drug design.
We do not think that effects observed in this paper will prove to be physiologically relevant, as circulating T3 and T4 concentrations are much lower than 1μM used to obtain radiolabeled T3 displacement in these assays (see reference ). However, given that high affinity for the TR LBC does not always correlate with the ability of challenger ligands to rapidly displace bound ligand, it may be interesting to consider the possibility that some compounds which bind TRs with low affinity, but are present in cells at high concentrations, could regulate T3 dissociation rate. Given that T3 is an unusual amino acid derived from tyrosine, it may be interesting to measure effects of physiological amino acids on TR ligand binding kinetics.
A large fraction of available TR ligands trigger release of bound T3 from the buried LBC more rapidly than an excess of T3 itself. While previous explanations of this phenomenon suggested that such ligands interact with a poorly defined allosteric interaction site, our data suggest that the challenger interacts with the LBC to promote ligand release, implying that it binds to a partially unfolded TR intermediate. This hypothesis suggests that different ligands associate with, and dissociate from, the TR LBD in different ways.
This work was supported by NIH grants DK41482 and DK51281 to JDB. JDB is deputy director and consultant to Karo Bio AB, a Biotechnology company with commercial interests in nuclear receptors.