Insight Into Molecular Determinants of T3 vs T4 Recognition From Mutations in Thyroid Hormone Receptor α and β

Abstract Context The two major forms of circulating thyroid hormones (THs) are T3 and T4. T3 is regarded as the biologically active hormone because it binds to TH receptors (TRs) with greater affinity than T4. However, it is currently unclear what structural mechanisms underlie this difference in affinity. Objective Prompted by the identification of a novel M256T mutation in a resistance to TH (RTH)α patient, we investigated Met256 in TRα1 and the corresponding residue (Met310) in TRβ1, residues previously predicted by crystallographic studies in discrimination of T3 vs T4. Methods Clinical characterization of the RTHα patient and molecular studies (in silico protein modeling, radioligand binding, transactivation, and receptor–cofactor studies) were performed. Results Structural modeling of the TRα1-M256T mutant showed that distortion of the hydrophobic niche to accommodate the outer ring of ligand was more pronounced for T3 than T4, suggesting that this substitution has little impact on the affinity for T4. In agreement with the model, TRα1-M256T selectively reduced the affinity for T3. Also, unlike other naturally occurring TRα mutations, TRα1-M256T had a differential impact on T3- vs T4-dependent transcriptional activation. TRα1-M256A and TRβ1-M310T mutants exhibited similar discordance for T3 vs T4. Conclusions Met256-TRα1/Met310-TRβ1 strongly potentiates the affinity of TRs for T3, thereby largely determining that T3 is the bioactive hormone rather than T4. These observations provide insight into the molecular basis for underlying the different affinity of TRs for T3 vs T4, delineating a fundamental principle of TH signaling.

T hyroid hormones (THs) are indispensable for normal growth, development, and metabolism. T3 and T4 are the two major forms of TH. In 1952, it was recognized that T3 has greater biological potency than T4 (1)(2)(3)(4). This fundamental discovery led to the clinical concept that T4, despite being the most abundant circulating iodothyronine, functions as a prohormone, with T3 being the biologically active hormone. Since then, this paradigm has remained unchanged, although the molecular and structural mechanisms underlying this have not been investigated in detail.
The genomic actions of THs are exerted through binding to the three functional isoforms of TH receptors (TRs), namely TRa1, TRb1, and TRb2, which are highly homologous but have distinctive expression patterns (5)(6)(7). Mutations in TRa and TRb give rise to clinically distinct syndromes in humans, termed resistance to TH (RTH)a and RTHb, respectively (8)(9)(10)(11)(12)(13)(14). RTHb patients commonly present with goiter and tachycardia with abnormal thyroid function tests, including high serum free T3 (FT3) and free T4 (FT4) concentrations with normal or slightly increased TSH concentrations. The clinical phenotype of RTHa is distinct from RTHb and includes growth retardation, macrocephaly, constipation, intellectual disability, and anemia. In RTHa, thyroid function tests are typically characterized by high to high-normal FT3, low to low-normal FT4, low reverse T3, and normal TSH concentrations.
The greater biological activity of T3 vs T4 is explained by differences in affinity for the functional isoforms of TH receptors (TRs). The binding affinity of T4 to the TRs is 10-to 30-fold less compared with T3 (15)(16)(17). Previous crystallographic studies revealed that the ligand-binding pocket of TRb1 is able to accommodate both T3 and T4, although the helix (H)11-H12 loop is more loosely packed in the presence of T4 than T3 (16). These structural adaptations of TRb1, which are required to accommodate the larger T4 molecule, have been attributed to possible steric hindrance of its bulky 5 0 -iodine moiety with the surrounding amino acids, especially the Met residue located at position 310 in TRb1. Although no cocrystallization studies of TRa with T4 are available, a similar role for Met256 in TRa (the equivalent position of Met310 in TRb) has been suggested (18). However, no functional studies to support the relevance of these residues for the differences in affinity for T3 and T4 have been performed. Therefore, we combined structural modeling and in vitro approaches to determine the differential role of these Met residues in T3 vs T4 binding by TRs and characterized a newly identified TRa1-M256T and previously published TRb1-M310T mutations, which naturally occur in patients with RTH (19)(20)(21). We showed that these Met residues are of particular importance for the binding of T3 but not T4. This observation provides the underlying molecular and structural basis for the role of T4 as prohormone and T3 as bioactive hormone in a paradigm for TH physiology and daily clinical practice.

TRa-M256T identification
The TRa-M256T mutation in an RTHa patient was identified by exome sequencing and was confirmed by Sanger sequencing as previously described (12) after obtaining informed consent. This study was conducted following the Declaration of Helsinki principles and was approved by the Medical Ethical Committee of the Erasmus Medical Center, Rotterdam, Netherlands (MEC-2015-362).

In silico prediction of TRa1-M256T function
The TRa1-M256T mutation bound to T3 and T4 was modeled into the wild-type (WT) TRa1 crystal structure (PDB-ID: 2H77) (22), and the M256T and M256A mutations were introduced using the side-chain substitution tool of the YASARA Structure Software (YASARA Bioscience GmbH, Vienna, Austria) (23) and processed as previously described (24).

Radioligand competitive binding assays
FLAG-TRa1 WT, M256T, and M256A receptor proteins were synthesized using the TnT® T7 Quick Coupled Transcription/Translation System (Promega, Leiden, Netherlands). The affinity for T3 and T4 of the receptors was determined by competitive binding assays as previously described (24)  Cell culture and transfection JEG-3 cells (ECACC Cat. no. 92120308, RRID:CVCL_ 0363; Sigma-Aldrich, Munich, Germany) were cultured and transfected as previously described (24,26). Given the absence of 5 0 -deiodinating activity in this cell type (27), there is no intracellular deiodination of T4 to T3, which allowed us to study the direct effect of T3 and T4 on transactivation. For transcriptional activity assays, WT or mutant receptors were coexpressed with luciferase reporter constructs containing direct repeat TH response elements (DR4-TRE) as well as pMaxGFP as a transfection control. We also coexpressed WT and TRa1-M256T in a 1:1 equimolar ratio to determine the effect of the mutant on WT function (i.e., the dominant-negative effect). For receptor-cofactor interaction (two-hybrid) assays, VP16-fused WT or TRa1-M256T were coexpressed with a luciferase reporter construct containing the Gal4 binding site (UAStkLuc), together with pSG424 expression vectors containing the Gal4DBD fused to the interacting domains of NCoR1 or SRC1 (11). After transfection for 24 hours, cells were stimulated with 0 to 10,000 nM T3 (Cat. no. T2877; Sigma-Aldrich) or T4 (Cat. no. T2376; Sigma-Aldrich) in DMEM/F12 medium supplemented with 0.1% bovine serum albumin for 24 hours.

Luciferase assays
Luciferase activity was measured as previously described (12,24). Data were expressed as percentage maximal response of WT stimulated by T3. EC50, IC50, and maximal response were calculated using GraphPad Prism program version 5.0 (GraphPad, La Jolla, CA). The results are shown as the mean 6 SEM of at least three independent experiments performed in triplicate.

Statistical analysis
Statistical differences of logKd, logIC50, and logEC50 values between groups were analyzed by Student t test or oneway ANOVA with Tukey post test. The percentage maximal response of mutants was compared with WT by one-sample t test. Statistical significance was considered at P , 0.05.

Protein modeling
The role of the Met256 in TRa1 function and the potential effect of this mutation on the affinity of T3 and T4 were predicted by in silico modeling. Given the absence of a T4-bound TRa crystal structure, we studied the structural organization of the domains surrounding the outer ring of TH in the available T3-(PDB ID: 1xzx) and T4-liganded (PDB ID: 1y0x) crystal structures of TRb1. In line with a previous report (16), we observed that the 5 0 position of the outer ring of T3 and T4 is flanked by Ile276 (H3), Met310 and Met313 (H6), His435 (H11), and Phe455 and Phe459 (H12) of TRb1. Together, these residues form a niche that allows the accommodation of T4 despite the presence of its bulky 5 0 -iodine. The same niche is present within the T3-liganded TRb1 crystal but is considerably smaller in the absence of the 5 0 -iodine. Met310 (corresponding to Met256 of TRa1) is in the closest structural proximity to the 5 0 -carbon of the outer ring and forms an extensive network of (hydrophobic) interactions that link H6, H11, and H12.
We next modeled a T4 molecule into the ligand binding pocket of the available T3-liganded TRa1 crystal structure (PDB-ID: 2H77) [ Fig. 2(b)]. Compared with the T3-liganded TRa1 structure [ Fig. 2(a)], a slight outward shift of H11 and H12 was observed in the T4-liganded model, which was accompanied by reorientation of side-chains of residues surrounding the 5 0 -iodine. This resulted in a loss of the direct hydrophobic interactions between Met256 and the outer ring and a less tightly packed structural organization of the ligand binding pocket. These changes were similar to those observed in the corresponding TRb1 crystal structures, validating the accuracy of the modeling procedure.
We subsequently modeled the M256T (shortening of side-chain, hydrophilic moiety) mutant in both T3-and T4-bound TRa1 structures and analyzed the impact on the conformation of the ligand binding domain and direct  Moreover, both mutants enlarged the niche surrounding the 5 0 position of T3 due to reorientation of various residue side-chains in H11 and H12 and the subsequent outward shift of these helices. As a result, the niche adopts a structural configuration that resembles the WT receptor in the T4-bound state. These changes were more pronounced for the M256T than the M256A, exemplified by the degree of reorientation of His381, which was previously implicated to interact with the phenolhydroxyl group of T3 (18) [ Fig. 2(g)]. In the case of T4, both mutations had little effect on structural organization [ Fig. 2(d), 2(f), and 2(h)]. Based on these in silico predictions, we hypothesized that both substitutions would have a greater impact on T3 than on T4 binding and action.

Functional studies
We performed in vitro studies to test this hypothesis. In line with previous literature (15)(16)(17), competitive binding assays showed that the affinity for T4 of WT TRa1 was approximately sevenfold lower than for T3, indicated by a higher Kd of T4 than T3 [ Fig. 3(a); Table 1]. The TRa1-M256T mutant showed a ;40-fold lower T3 binding affinity than WT, whereas T4 affinity was unchanged [ Fig. 3(c); Table 1]. Also, the binding affinity of the TRa1-M256A mutant for T3 was selectively reduced (approximately sixfold) [ Fig. 3(e); Table 1].   To evaluate the impact of both mutations on the transcriptional activity, WT and mutant receptors were cotransfected with a reporter construct in which luciferase expression is under control of a TH response element into JEG-3 cells with increasing concentrations of T3 or T4. Equal expression of WT and both mutants was confirmed by immunoblotting nuclear extracts with anti-FLAG antibodies [ Fig. 3(b)]. In line with the binding assays and previous studies (16,17), the transcriptional activation assay showed that the EC50 of WT TRa1 induced by T4 was ;60-fold higher than that induced by T3 [ Fig. 3(b); Table 1]. The EC50 of TRa1-M256T was 100-fold higher for T3 but was unchanged for T4 compared with WT [ Fig. 3(d); Table 1]. The TRa1-M256A also selectively reduced transcriptional activity induced by T3 [ Fig. 3(f); Table 1]. The transcriptional activity was also reduced when WT and TRa1-M256T were coexpressed compared with WT expressed alone, suggesting a dominant-negative effect of this mutant (data not shown). In mammalian twohybrid assays compared with WT, the TRa1-M256T mutant also affected ligand-dependent interactions with the corepressor NCoR1 (fold increase IC50: ;80-fold for T3 and approximately sixfold for T4) and the coactivator SRC1 (fold increase EC50: ;90-fold for T3 and approximately sixfold for T4) [ Fig. 4(a)-4(d); Table 1]. Together, our results indicate that the mutations located at the Met256 of TRa1 have a differential impact on the binding and activation of the receptor by T4 vs T3.
We next investigated if this T3 vs T4 difference is present in other TRa mutants located outside the niche surrounding the 5 0 -iodine position. However, these naturally occurring mutations (D211G, A263S, and R384H) had a similar impact on T3-and T4-induced transactivation, and, as for WT TRa, the EC50 values for T4 exceeded those for T3 by ;30to 50-fold [ Fig. 5(a)-5(c)]. These transcriptional activation profiles were in contrast to the M256T mutant [ Fig. 5(d)], strongly indicating that only this mutant has a predominant impact on T3 affinity. To extend our findings to TRb, we studied the transcriptional activity of a corresponding mutation in TRb1 (TRb1-M310T). The EC50 of WT TRb1 induced by T4 was ;70-fold higher than that induced by T3 [ Fig. 6(a)], which was similar to WT TRa1. The T3-induced transcriptional response of TRb-M310T was greatly reduced, which contrasted with the T4-induced transcriptional activity (fold increase EC50: ;350-fold for T3 and approximately threefold for T4) [ Fig. 6(b)]. Figure 5. (a-c) The T4-induced transcriptional activity of three TRa1 mutations identified in RTHa patients is lower than that is induced by T3, which is similar to WT [ Fig. 2(d)] (mean 6 SEM of three experiments performed in triplicate). (d) The EC50 of T4 is ;30to 50-fold higher than the EC50 of T3, except for TRa1-M256T. ***P , 0.001 (one-way ANOVA with Tukey post test).

Discussion
Although the notion of T4 and T3 being the precursor and active hormones, respectively, is widely recognized in the clinical and scientific communities, the molecular and structural basis of this dogma has received little attention. In this study, we highlight the crucial role of residue Met256 of TRa1 and Met310 of TRb1 in determining the differential bioactivity of T3 vs T4 using a novel mutant (TRa1-M256T) identified in an RTHa patient and a mutant at the corresponding position (TRb1-M310T) identified in RTHb patients (19)(20)(21). In contrast to WT TRa or TRb and mutations involving other residues, mutations at these Met residues selectively affected binding and transactivation of TR by T3. These observations emphasize the key role of these residues in designating T4 as the prohormone and T3 as the major bioactive hormone.
In line with previous reports (15)(16)(17), our results showed that T3 has a higher binding affinity for WT TRa1 and stimulates receptor activity with a higher potency than T4. Previous structural studies in TRb1 have suggested that the lower affinity for T4 is caused by decreased packing of the ligand binding domain in the presence of T4 vs T3, which particularly allows oscillation of H12 between liganded and unliganded states, resulting in a higher ligand dissociation rate (16). Here, we extend these observations by showing that the ligand binding domain of T3-liganded TR has a similar decrease in packing as observed in T4-liganded WT receptors upon substitution of Met256 in TRa1 or Met310 in TRb1 by Thr. In contrast, these substitutions hardly changed the predicted structure of T4-liganded mutant receptors. Based on these models, we postulated that the extensive (hydrophobic) interactions of Met with surrounding residues are key in stabilizing interhelical interactions (e.g., between H6, H11, and H12), which facilitate the tight packing of the ligand binding domain as observed in T3-liganded receptors. Moreover, we observed a direct interaction between Met and the 5 0 position of the outer ring of T3, which was not formed with T4. This suggests that Met256 in TRa1 and Met310 in TRb1 have a critical role in achieving optimal folding and enthalpy in T3-liganded receptors, whereas their role in T4 binding is of less importance.
This in silico prediction was confirmed by in vitro studies indicating that TRa1-M256T selectively affected binding affinity for T3 as well as cofactor interactions and transcriptional activity of the T3-stimulated receptor. These properties seemed specific for the M256T mutant because the transactivation potency of T3 and T4 with TRa mutants identified in other RTHa patients [D211G (26), A263S, and R384H (28)] was affected equally. Additional testing of the naturally occurring mutation at the corresponding residue in the TRb1 (M310T) (19)(20)(21) further substantiated the specificity of the findings.
Thr substitution at position 256 in TRa1 or 310 in TRb1 not only alters the binding space but also affects the hydrophobicity of the ligand-binding pocket. Therefore, we tested the artificial TRa1-M256A mutant, which reduces the size of the side-chain but maintains the hydrophobic property of the ligand-binding pocket. Indeed, functional studies showed that TRa1-M256A also selectively impairs T3 binding affinity and T3-induced transcriptional activity, whereas T4 binding and activity are maintained. Although the effect of the TRa1-M256T mutation in our functional and structural models was slightly greater than that of TRa1-M256A, these findings support the notion that loss of the specific properties of Met, rather than the unfavorable impact of the hydrophilic moiety of Thr on the hydrophobic environment, are mainly responsible for the differential impact on T3 vs T4 signaling. Based on our studies and on a previous report (16), we propose that Met256 in TRa1 and Met310 in TRb1 are crucial residues that determine specific affinity for T3 vs T4. Thr and Ala substitution at these Met positions significantly affected the hydrophobic interactions with T3 and altered the niche accommodating the outer ring of T3 to a "T4-bound" configuration, both resulting in a reduced binding affinity of the mutants for T3. In contrast, because the ligand binding domain of T4-liganded receptors already exhibit looser packing without direct interaction(s) between Met and the T4 molecule, mutations in the Met residue are better tolerated.
No unique phenotype was discernible in the newly identified M256T RTHa patient when compared with other cases of RTHa harboring missense mutations in the THRA gene (25,26,(28)(29)(30) or in patients carrying TRb-M310T (19)(20)(21) when compared with other RTHb cases reported in the literature. These findings indicate that, although mutations at Met256-TRa1/Met310-TRb1 residues preserve T4 binding to mutant receptor proteins, this property is not sufficient to prevent patients from developing features of RTH, implying that the phenotype of RTH is linked primarily to defective T3 rather than T4 binding by mutant TRs.
This study provides in vitro evidence for the importance of Met256 in TRa1 and Met310 in TRb1 in ligand recognition. Our studies highlight the relevance of this Met residue in TRs for discrimination between T3 and T4, providing the molecular basis for the role of T4 as prohormone and T3 as bioactive hormone.