Structure-Guided Approach to Relieving Transcriptional Repression in Resistance to Thyroid Hormone α

ABSTRACT Mutations in thyroid hormone receptor α (TRα), a ligand-inducible transcription factor, cause resistance to thyroid hormone α (RTHα). This disorder is characterized by tissue-specific hormone refractoriness and hypothyroidism due to the inhibition of target gene expression by mutant TRα-corepressor complexes. Using biophysical approaches, we show that RTHα-associated TRα mutants devoid of ligand-dependent transcription activation function unexpectedly retain the ability to bind thyroid hormone. Visualization of the ligand T3 within the crystal structure of a prototypic TRα mutant validates this notion. This finding prompted the synthesis of different thyroid hormone analogues, identifying a lead compound, ES08, which dissociates corepressor from mutant human TRα more efficaciously than T3. ES08 rescues developmental anomalies in a zebrafish model of RTHα and induces target gene expression in TRα mutation-containing cells from an RTHα patient more effectively than T3. Our observations provide proof of principle for developing synthetic ligands that can relieve transcriptional repression by the mutant TRα-corepressor complex for treatment of RTHα.

recruited to TRa (14), or treating TRa-PV mice with suberoylanilide hydroxyamic acid (SAHA), an HDAC inhibitor (15), ameliorates these phenotypes. Overall, these observations suggest that genetic or pharmacological disruption of the mutant TRa-CoR repression complex in RTHa might be beneficial.
Here, we have used biophysical (fluorescence anisotropy and circular dichroism) approaches to investigate the molecular properties of TRa mutants. Unexpectedly, we found that RTHa-associated TRa mutants could still bind T3. Validating this notion, we visualized T3 within the crystal structure of a prototypic TRa mutant, which was homologous to RTHa-associated mutant receptors. This finding prompted us to design, synthesize, and test TH analogues, identifying a lead compound which dissociated CoR from mutant human TRa more efficaciously than T3. This compound was more potent at preventing mutant human TRa-mediated developmental anomalies in a zebrafish model and induced greater TH target gene expression in patient-derived, TRa mutation-containing, primary cells studied ex vivo.

RESULTS
Biophysical analyses of TRa mutant-coregulator interactions. We studied three human TRa mutants (A382fs388X, F397fs406X, and E403X) with disruption or truncation of the receptor carboxy terminus ( Fig. 1A and B); these constitutively repress TH target genes, fail to activate their transcription in response to T3, and inhibit wild-type receptor action in a dominant negative manner (9,11), suggesting that their interaction with coregulator proteins is altered. To determine whether these mutations mediated failure of corepressor displacement, failure of coactivator recruitment, or a combination of the two, we used a fluorescence anisotropy (FA) assay to investigate the ability of wild-type and mutant TRa ligand-binding domains (LBDs) to bind to peptide motifs derived from corepressor and coactivator receptor interaction domains in the presence or absence of T3.
In the absence of T3, the interaction of both wild-type and mutant LBDs with the SMRT corepressor peptide is strong. Importantly, the mutant proteins A382fs388X and F397fs406X bind the corepressor with similar affinities to the unliganded wild-type receptor protein. In the absence of T3, neither the wild-type nor the mutant proteins exhibit detectable interaction with peptide from the GRIP1 coactivator ( Fig. 2A).
As anticipated, the presence of saturating levels of T3 decreases the affinity of the wildtype TRa LBD for the SMRT corepressor peptide by almost 10-fold and increases the affinity for the coactivator peptide by several orders of magnitude ( Fig. 2A). These changes in coregulator affinity underlie the ligand-induced molecular switch (Fig. 1B) which mediates the transition from repression to activation of target gene transcription by TR.
For the TRa mutants, we have delineated two molecular defects. First, their affinity for the corepressor in the presence of T3 is essentially unchanged; second, there is no T3-induced enhancement of coactivator interaction.
This behavior of TRa mutants as if the ligand were not present might suggest that the receptors are unable to bind to T3.
To test directly whether the mutant receptors were able to bind ligand, we used a circular dichroism (CD) assay to monitor the thermal stabilities of the wild-type and TRa mutant LBDs in the presence and absence of saturating levels of T3 and coregulator peptides (Fig. 2B). Interestingly, in the absence of ligand or coregulator peptides, the A382fs388X and E403X TRa mutants exhibited slightly greater thermal stability compared to the wild-type receptor. This suggests that helix 12 in unliganded, wildtype TR may confer a destabilizing effect which is not seen in TRa mutants lacking this carboxy terminal.
Strikingly, CD studies showed that all three TRa mutants were able to bind to the ligand (T3) and showed significant thermal stabilization, similar to that of the wild-type receptor. Similarly, in the absence of T3, both wild-type and mutant TRa proteins were significantly stabilized by corepressor binding. Significantly, for the three TRa mutants, the presence of both T3 and corepressor induced further stabilization of proteins, strongly suggesting that both the coregulator peptide and the ligand can bind simultaneously to the receptor LBD (Fig. 2B).
Given that all three TRa frameshift/truncation mutants lack helix 12, which caps the ligand binding pocket, with the A382fs388X mutant TRa also lacking a significant portion of helix 11, which forms one side of the ligand binding pocket, the finding that all mutant LBDs were able to bind T3 was unexpected.
Since the CD experiments were performed with saturating T3 concentrations, we determined the T3-binding affinity of the wild-type and mutant receptors in a competition assay using radiolabeled ligand (Fig. 2C). This confirmed that all three mutants bind T3 with sub-micromolar affinity, approximately 100-fold weaker than the wildtype receptor. Together, these studies suggest that the mutations in these patientderived receptors exert their effects through an impairment in ligand-and coregulator-binding equilibria (Fig. 2D).
To assess the relative contribution of defective corepressor release versus coactivator recruitment to the inhibition of TH action by mutant TRa, we analyzed patientderived primary cells containing E403X TRa, a mutant which exhibits both defective corepressor dissociation and coactivator recruitment, versus patient-derived cells containing E403K TRa, a mutant with predominantly defective coactivator interaction (Fig.  3A). We observed that T3-dependent induction of KLF9, a TH target gene, was significantly more attenuated in E403X than in E403K mutant TRa-containing cells (Fig. 3B), supporting the concept that a receptor which cannot displace corepressor is more deleterious than one which simply cannot recruit coactivator. Importantly, electrophoretic mobility shift assays using either canonical (DR 1 4) or natural (human KLF9 promoter) thyroid response elements showed that these differences in transcriptional activity were not due to the differential DNA binding properties of E403X or E403K TRa mutants in the presence of natural (T3) or synthetic ligands (ES08, see below) ( Fig. 3C).
Structure of a ligand-bound mutant TRa. To understand how the TRa mutants might bind ligand, we sought to crystallize the human mutant TRa LBDs in the presence of T3 and complexed with corepressor peptides, which likely reflects the dominant negative mutant receptor species in vivo. Unfortunately, we failed to obtain diffraction quality crystals, likely due to the presence of additional amino acids following frameshift mutations, which we would expect to be disordered. Accordingly, to model the human TRa mutants, we crystallized an artificial mutant receptor (P393GX) which was truncated at the carboxy terminus of helix 11 ( Fig. 4A and C).
Importantly, the P393GX TRa shows similar molecular behavior to the human TRa mutants. In fluorescence anisotropy assays, P393GX TRa interacts with corepressor peptides with comparable affinity to the human TRa mutant LBDs, with this interaction becoming stronger in the presence of T3 (Fig. 5A). As is the case with natural human TRa mutants, both ligand and corepressor peptide increase the thermal stability of P393GX TRa, with the greatest stabilization occurring in the presence of both T3 and SMRTderived peptide, suggesting that P393GX LBD is able to bind T3 and SMRT corepressor simultaneously (Fig. 5B). Radiolabeled ligand binding studies confirmed T3 binding to P393GX mutant TRa with micromolar affinity (Fig. 2c), which was expected since the artificial mutant is more truncated than the natural human TRa mutants.
Small crystals of P393GX in complex with T3 were obtained which diffracted to 3.0 Å. The structure contains one P393GX molecule (amino acids 156 to 393) and one T3 molecule in the asymmetric unit ( Fig. 5C; Table 1). Much of the overall organization of the LBD is essentially the same as that of the wild-type receptor protein. However, despite having the potential to form a complete helix 11 (aa 363 to 393), in the P393GX mutant TRa this helix terminates at Ala 379, with the peptide backbone turning at this point to form an extended coil which runs antiparallel to the adjacent helix 5 toward the corepressor-binding site.
The T3 ligand could be placed unambiguously in the hydrophobic cavity of P393GX TRa LBD (Fig. 5D) and binds very similarly to its occupancy of wild-type receptor LBD ( Fig. 4D and E). However, T3 interactions with Met 388 and Phe 401 are lost and the orientation of His 381 is altered. Arg 384 makes a new interaction with the ligand in the P393GX LBD, donating a hydrogen bond to the 49 hydroxyl of the outer ring of T3. Interestingly, the lack of the carboxy-terminal end of helix 11 and the joining of the loop to helix 12 creates space for residues amino-terminal to helix 3 to form a short anti-parallel b-sheet, which is not present in the wild-type receptor.   Remarkably, given that so much of the receptor carboxy terminus is missing, the ligand binding cavity in P393GX mutant TRa is almost completely enclosed due to the rearrangement of the carboxy-terminal portion of helix 11. Indeed, the volume of the ligand-binding cavity in P393GX is only slightly greater than that of the wild-type receptor (%200 Å 3 versus %160 Å 3 , as calculated using POCASA [16]). There is a small opening to the surface of the receptor around the 49 hydroxyl and iodine of the outer ring of T3, suggesting that the mutant receptor could likely accommodate a larger ligand in these positions. Since P393GX mutant TRa is capable of binding T3 and corepressor (Fig. 5B), we surmise that if the corepressor peptide binds the receptor at the expected site, part of the rearranged helix 11 would likely be displaced to accommodate this interaction.
Novel thyroid hormone analogues displace corepressor from human TRa mutants. The finding that both natural and artificial TRa mutant LBDs retain the ability to bind T3 raised the possibility of designing novel ligands that are better able to displace corepressor, thereby alleviating transcriptional repression by mutant human TRa. We synthesized a series of ligands in which the hydroxyl group of T3 was modified with either ether, ester, or sulfonate ester linkages of varying size, hydrophobicity, and flexibility (Fig. 6A).
Novel ligands were tested at saturating concentrations in fluorescence anisotropy and thermal stability assays using wild-type and human TRa mutant LBDs. Compared  to the unliganded proteins, all compounds increased the thermal stability of wild-type and mutant LBDs, indicating their ability to bind these proteins. Strikingly, whereas T3 stabilized the wild-type TRa LBD better than the synthetic analogues, several of the novel ligands (including ES08) stabilized mutant TRa proteins more effectively than T3; this suggests that the additional groups in these modified ligands enabled them to better occupy the aberrant ligand-binding pockets of the mutant proteins ( Fig. 6B, lower panels). Fluorescence anisotropy assays measured the ability of ligands to perturb the binding of wild-type and human TRa mutant LBDs to corepressor peptide (Fig. 6B, upper panels). Many synthetic ligands promoted the dissociation of corepressor from wild-type TRa, but they did so less effectively than T3, presumably due to their inability to rearrange helix 12 into its active, corepressor-displacing conformation (17). However, two compounds, ES08 and ES09, mediated dissociation of corepressor from human TRa mutants more effectively than T3, particularly in the case of the mutant receptor E403X (Fig. 6B, upper panels). Competition assays with radiolabeled T3 confirmed that both wild-type and E403X TRa were able to bind ES08 with submicromolar affinity. For wild-type TRa, ES08 binding is much weaker than T3. However, importantly, the E403X mutant TRa binds T3 and ES08 with similar affinities (Fig. 7A).

C-terminal
Efficacy of TH analogues in mediating TRa-corepressor dissociation in cells. To compare the efficacy of T3 and synthetic ligands in promoting corepressor dissociation from TRa in cells, we tested them in cellular two-hybrid protein-protein interaction assays with co-expressed VP16-full length TRa and Gal4-corepressor fusions.
Both unliganded wild-type and mutant VP16-TRa fusion proteins exhibited strong interaction with Gal4-CoR fusions containing different isoforms of human NCoR (NCoRv and NCoR-d ) or SMRT (SMRT-a and SMRT-g) (18). T3 exposure readily dissociated all CoR isoform fusions from wild-type receptor but not from human TRa mutants. SMRT-« , a corepressor isoform lacking two receptor interaction motifs (S1 and S3) did not interact with either wild-type or mutant TRs (Fig. 8A).
Next, we compared the relative efficacy of T3 versus different TH analogues (each at 100 nM) in promoting corepressor dissociation from wild-type or mutant TRa using two-hybrid assays. Compared to the complete or partial dissociation of corepressor from wild-type receptor or TRa mutants with T3 exposure, most TH analogues exhibited comparable or inferior efficacy (Fig. 8B). Notably, when tested with a particular human mutant receptor, E403X TRa, a single compound, ES08, promoted greater dissociation of corepressor from mutant TRa than T3 did (Fig. 8B, highlighted). We confirmed that ES08 promotes greater corepressor dissociation from E403X mutant TRa than T3 does when tested over a range of ligand concentrations (Fig. 9A). Furthermore, compared to T3, ES08 promotes greater dissociation of other corepressor isoforms (NCoR-v , SMRT-a, and SMRT-g) from E403X mutant TRa (Fig. 9B). However, neither ES08 nor T3 was able to overcome the inability of E403X mutant TRa to recruit coactivator (Fig. 9C).
Efficacy of TH analogue in E403X mutant TRa-expressing zebrafish and patient-derived primary cells. We evaluated the efficacy of ES08 versus T3 in a zebrafish model of RTHa, as described previously (19). Following microinjection of zygotes with either wild-type receptor or E403X mutant TRa mRNAs, the embryos derived from E403X TRa mutant-injected zygotes exhibited multiple  developmental anomalies, consisting of abnormal indices of morphology (AMI), vascular malformation (VMI) or skeletal malformation (SMI), whereas embryos from wild-type TRa-injected zygotes were unaffected (Fig. 10). Despite exposure to T3 at high concentrations (2 to 20 mM), these developmental anomalies persisted in embryos derived from E403X TRa-injected zygotes, whereas their exposure to ESO8 (2 mM) prevented morphological, vascular, and skeletal malformations (Fig. 10C, F, and K). Furthermore, in E403X mutant TRaexpressing embryos, lower concentrations of ES08 (2 mM) induced greater expression of known thyroid hormone-responsive target genes than did high concentrations (2 to 20 mM) of T3 (KLF9 [Fig. 11A]; Dio3b [Fig. 11B]).  Finally, we compared the properties of ES08 versus T3 when incubated with primary, patient-derived, E403X mutant TRa-containing, inducible pluripotent stem cells. Compared to T3, ES08 at a higher concentration (2.5 mM) induced greater expression of KLF9, a known thyroid hormone-responsive target gene (Fig. 11C).

DISCUSSION
Severe RTHa is caused by a molecular mechanism involving the constitutive interaction of mutant TRa with corepressor, failure of corepressor release, and failure of coactivator recruitment in a ligand-dependent manner, resulting in silencing of gene transcription. Unexpectedly, given unmeasurable radiolabeled T3-binding to mutant TRa in previous assays (9), our biophysical studies together with radiolabeled T3-binding assays showed that these human TRa mutants retain hormonebinding ability. This was confirmed by visualizing T3 within the crystal structure of an artificial TRa mutant, prototypic of receptor mutants that commonly cause RTHa. This finding prompted the synthesis and evaluation of TH analogues, leading to the successful identification of a single compound which dissociates corepressor from human mutant TRa more effectively than T3 does. This synthetic ligand prevented phenotypic abnormalities in a zebrafish model of RTHa, and induced target gene expression in TRa mutation-containing cells from an RTHa patient, more efficaciously than T3.
The crystal structure of the P393GX TRa LBD reveals that the majority of the protein-ligand interactions are preserved in the mutant receptor despite the loss of much of H11 and H12. This accounts for why the mutant receptors retain the ability to bind thyroid hormone, albeit with substantially reduced affinity. The position of the carboxy-terminal portion of H11 in the structure appears to occlude the CoR binding region. This would appear to be an artifact of crystallization since we have shown experimentally that, like the natural TRa mutants, the P393GX receptor binds CoR both in the presence and in the absence of T3. We would predict that this region will not adopt a single conformation in solution.
Almost half (8/19) of the mutations causing RTHa which have been described hitherto are associated with a severe hypothyroid phenotype and disrupt the receptor carboxy terminus in a manner analogous to the TRa mutants we have studied here (9,11,(20)(21)(22)(23). To date, thyroxine treatment of such patients has been associated with a poor (10) or partial (11) clinical response. This may be because thyroxine therapy at conventional dosages does not lead to corepressor dissociation. New ligands that maximize the likelihood of corepressor dissociation from mutant TRa may be a more successful treatment.
Incomplete ligand-dependent release of corepressor from mutant TRa despite its ability to bind T3, therefore, provided the rationale for synthesizing TH analogues, identifying a compound, ES08, which mediated greater corepressor displacement from the E403X mutant. ES08 and ES09 are sulfonate ester T3 analogues bearing a biaryl or 9H-fluorene substituent, respectively. In the structure of ES08, it is possible that the change in geometry induced by the sulfonate ester link, together with the relative rigidity of the extension, sterically disrupts corepressor binding to the cleft on the TR LBD surface, thereby lowering its affinity for corepressor binding ( Fig. 7B and C). We speculate that the presence of part of helix 12 in E403X mutant TRa accounts for the ability of this analogue to displace corepressor more effectively from this particular TRa mutant.
One limitation of our study was the replacement of the phenol group of T3 with a sulfonate ester in ES08, disrupting the known interaction of its hydroxyl group with Histidine 381 in TR, which may have compromised its receptor-binding affinity. Accordingly, in the future, we will synthesize and test different ligands, restoring this phenol moiety and modifying T3 in the 59 position instead, since structural modeling suggests that such analogues would be better orientated to displace corepressor. Furthermore, although the synthetic analogue we have identified, ES08, was only effective with a specific TRa mutant, our studies suggest that it may be possible to generate ligands that can better displace corepressor from this general subclass of mutant receptor. As the comparison of the E403X and E403K mutants suggests, corepressor displacement is beneficial even in the absence of coactivator recruitment. It is striking that corepressor displacement, at high ES08 concentrations, has a significant beneficial effect in the zebrafish model. The recent recognition that a thyroid hormone analogue (triiodothyroacetic acid) which is used to treat resistance to thyroid hormone b acts via this mechanism (24) provides added justification for this approach.
Other approaches to relieving transcriptional repression by the mutant TRa-corepressor complex in RTHa have limitations. In a recent study, abrogating NCoR-TR interaction in TRa-PV transgenic mice did not reverse skeletal abnormalities, suggesting that transcriptional repression by mutant TRa complexed with other corepressors (e.g., SMRT) may mediate this phenotype (25). Similarly, targeting the TRa-corepressor repression complex pharmacologically with an HDAC inhibitor may not be effective if transcriptional repression operates via diverse complexes containing different HDACs which may not be sensitive to such inhibition (15). Histone deacetylases are also components of complexes with other nuclear receptors and transcription factors (26,27), and nonspecific inhibition of histone deacetylase enzyme activity may derepress these pathways, causing off-target effects.
Overall, our observations provide a proof of concept for the synthesis of designer ligands, targeting aberrant mutant TRa-corepressor interaction, to alleviate receptor dysfunction in resistance to thyroid hormone a.

MATERIALS AND METHODS
Expression and purification of TRa LBDs. The human WT, A382fs388X, F397fs406X, E403X, and P393GX LBDs (residues 148 to 410, 148 to 387, 148 to 402, and 151 to 393) were cloned into a pGEX2T (GE Healthcare) vector containing an amino-terminal glutathione S-transferase (GST) purification tag followed by a TEV protease cleavage site. TRa LBDs were expressed in E. coli Rosetta (DE3) (Novagen) by growing the transformed Rosetta (DE3) at 37°C in 2xTY until A 595 = 0.1, then inducing with 40 mM isopropyl-D-1-thiogalactopyranoside (IPTG) and growing overnight at 20°C. The bacterial cells were lysed by sonication in a buffer containing 1Â phosphate-buffered saline (PBS), 1 mM dithiothreitol (DTT) and complete EDTA-free protease inhibitor (Roche). The soluble protein was bound to glutathione-Sepharose (GE Healthcare), and washed with a buffer containing 1Â PBS, 0.5% Triton X-100, and 1 mM DTT. Next, the bound protein was washed with TEV cleavage buffer containing 20 mM Tris-HCl (pH 8), 100 mM NaCl, and 1 mM DTT. The GST tag was removed by incubation with TEV protease (100:1 molar ratio) overnight at 4°C.
Radiolabeled T3-binding assays. Homologous competitive binding assays were performed using triiodothyronine (T3) labeled with 125 I ( 125 I-T3, PerkinElmer). Recombinant WT, mutant, and artificial mutant TRa LBDs were incubated with 0.02 nM 125 I-T3 in binding buffer (20 mM Tris [pH 8], 50 mM KCl, 1 mM MgCl 2 , 10% glycerol, 5 mM DTT) in the presence of increasing amounts of unlabeled competing T3 (0 to 100 mM). Appropriate protein concentrations were determined experimentally to give 10% of the total radioactivity of the assay, securing a good signal (total binding) to noise (nonspecific binding) ratio and preventing the ligand depletion effect. Following 1 h of incubation at 37°C, bound T3 was separated from unbound T3 by passage through a filter membrane (Millipore HA Filters, 0.45 mm) under vacuum, followed by three washes with 2 mL of ice-cold binding buffer. Filters containing TR-bound 125 I-T3 were measured in a g-counter.
A competitive binding assay was performed following the same procedure in the presence of increasing amounts of unlabeled competing ES08 (0 to 10 mM) using recombinant WT and E403X TRa LBDs.
A half-maximal inhibitory concentration (IC 50 ), which indicates the amount of ligand that causes 50% inhibition of radioligand binding, was determined by plotting the radioactivity values obtained at every cold-competing T3 concentration, using the GraphPad Prism and a nonlinear regression analysis. Since the dissociation constant, Kd, and the inhibitor constant, K i , should be the same as the radioactive ligand and the competing ligand are the same, for this type of experiments the Kd of the binding is calculated by subtracting the concentration of radioligand from the IC 50 value obtained from the curve, as in the following equation: Kd = K i = IC 50 -[radioligand conc.].
Fluorescence anisotropy and circular dichroism. Two peptides were designed for use in the fluorescence anisotropy assay: an N-terminal, FITC-labeled, 14-aa-length peptide with a sequence based on the interaction domain 1 of the SMRT corepressor protein (RID1 residues 2346 to 2360: Ac-STNMGLEAIIRKALMG-NH 2 ), containing the corepressor NR recognition motif LxxxIxxx[I/L]; and a C-terminal, BODIPY-TMR-labeled, 16-aa-length peptide with a sequence based on the second NR interaction box of the GRIP1 coactivator protein (NID2 residues 686 to 700: Ac-KHKILHRLLQDSSC-NH 2 ), containing the coactivator NR recognition motif LxxLL.
Fluorescence anisotropy (FA) experiments were performed in black 96-well assay plates (Corning Life Sciences). Multiple titrations were performed using fixed concentrations of SMRT and GRIP1 peptides (5 nM) with increasing concentrations of TRa LBDs (0 to 5mm) in a final volume of 50 mL of assay buffer (1Â PBS, 0.01% [vol/vol] Triton X-100, 0.1 mg/mL bovine serum albumin [BSA]). For the assays in the presence of T3 or T3 analogues, increasing concentrations of the mixture protein:T3 or protein:T3 analogues in a 1:2 molar ratio were used. After incubation at room temperature for 5 min with slow shaking and centrifugation of the plates, the FA value was measured at each receptor concentration in a VICTOR X5 multilabel plate reader (Perkin Elmer, Singapore), using a 480-nm excitation filter and 535nm emission filters to measure FITC emission, and a 542-nm excitation filter and 572-nm emission filters to measure BODIPY fluorescence. Blank fluorescence values were subtracted in each polarization plane. FA values obtained at each protein concentration were used to generate saturation-binding curves that were subsequently used to calculate the equilibrium dissociation constant of the interaction (Kd), using Prism software (GraphPad) and nonlinear regression analysis.
Thermal unfolding of proteins was monitored by CD spectroscopy over a wavelength range of 200 to 250 nm, using a Chirascan Spectrometer (Applied Photophysics) equipped with a temperature controller (Quantum Northwest TC125). CD spectra were measured from samples in 1-mm-path-length quartz cuvettes, using a scanning speed of 100 nm/min, a spectral bandwidth of 1 nm, and a response time of 1 s. Secondary structure of the proteins was assessed by visual inspection of CD spectra from 200 to 250 nm. The thermal denaturation or unfolding profile of the proteins was characterized by measuring the ellipticity changes at 222 nm induced by a temperature increase from 20 to 90°C with steps of 1 degree.
Peptide synthesis. SMRT and GRIP1 peptides were synthesized using a CEM Liberty 1 Automated Microwave Peptide Synthesizer on a 0.05-mmol scale, using solid-phase peptide synthesis (SPPS). For this technique, an N-protected C-terminal amino acid residue is anchored to an amino resin; in this case, H-Rink Amide Chem Matrix resin.
The amino acids were loaded onto this resin in a sequential manner from the C terminus to the Nterminus by repetitive cycles. Amino acids were Fmoc-protected and solubilized in dimethylformamide (DMF) to a concentration of 0.2 M. After every amino acid was loaded, they were deprotected using 20% piperidine in DMF to remove the Fmoc group. Following this, 0.5 M O-(1H-6-chlorobenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HCTU) in DMF was used as an activator, with 3 M N,Ndiisopropylethylamine (DIPEA) in NMP (N-methyl-2-pyrrolidone) as an activator base. The deprotection and amino acid coupling reactions were repeated in a linear manner for each amino acid to build the peptide sequence from the C terminus to the N terminus.
After synthesis of the full peptide sequences, the resin was removed and washed. The SMRT peptide was incubated with fluorescein isothiocyanate (FITC) in 5-fold molar excess for 5 h on a shaker at room temperature. The FITC-SMRT peptide and the GRIP1 peptide were cleaved from the resin by incubating the resin in 1 mL of TFA:TES:H 2 O at room temperature for 3 h. The peptides were precipitated using cold diethyl ether and then centrifuged at 3,500 Â g for 5 min. Supernatants were discarded, and the pellets were washed twice more with diethyl ether. After the 3rd ether wash, the peptides were freeze-dried and left overnight before purification using semi-prep high-performance liquid chromatography (HPLC). The fractions containing peptide from HPLC were pooled and a sample was submitted for LC-MS to determine purity.
BODIPY-TMR C 5 malemide (Invitrogen) was coupled to GRIP1 peptide through an N-terminal cysteine. Next, 90 mM peptide was incubated with a 5-fold molar excess of BODIPY in a 1-mL reaction with constant stirring for 2 h in darkness at room temperature. The purification of the labeled peptide from free dye was performed using a PD-10 column (GE Healthcare) pre-equilibrated with 1Â PBS containing 0.5% Tris(2-carboxyethyl)phosphine hydrochloride. Eluted fractions were concentrated to 50 mL using an Amicon centrifugal concentrator.
Crystallisation, structure determination, and refinement. The P393GX LBD was mixed with a 5fold molar excess of T3 (Sigma) and concentrated up to 9 mg/mL. Crystallization trials were initially conducted using commercial screens (Molecular Dimensions) into MRC 96-well sitting drop crystallization plates using 100 nL of protein sample and precipitant. Hexagonal crystals up to 20 mm in length were grown via sitting drop vapor diffusion at room temperature using 0.2 M NaCl, 0.1 M Tris (pH 8.5), and 1 M lithium sulphate. Cryoprotectant solution containing 20% glycerol, in addition to the other crystallisation buffer components, was added to the crystals, which were then rapidly frozen at 100 K using liquid nitrogen. Data were collected at the microfocus beamline I-24 at the Diamond Light Source (UK). The structure was solved by molecular replacement using wild-type TRa LBD (PDB code 2h79) as a search model in Phaser. The preliminary model was rebuilt iteratively by multiple rounds of refinement and building using Refmac5 and Coot to an R free of 22% and an R work of 18%. The final model contains one molecule of P393GX and one molecule of T3 in the asymmetric unit. The final model has 95.73% residues in the favored region, 4.27% in the allowed regions, and none in the outlier region of the Ramachandran plot.