Noncanonical thyroid hormone signaling mediates cardiometabolic effects in vivo
Edited by David W. Russell, University of Texas Southwestern Medical Center, Dallas, TX, and approved November 15, 2017 (received for review April 25, 2017)
Significance
This study changes our understanding of how thyroid hormone acts. Thyroid hormone receptors are considered typical nuclear receptors that bind to DNA and, after binding, alter the expression of their target genes and regulate physiological responses. Nevertheless, we show that thyroid hormone still mediates important physiological effects in mice expressing mutant receptors that cannot bind DNA. These are predominantly linked to energy metabolism and include glucose and triglyceride concentrations, body temperature, locomotor activity, and heart rate. This study provides in vivo evidence that thyroid hormone receptors mediate physiologically relevant effects that are independent of DNA binding and direct activation of gene expression.
Abstract
Thyroid hormone (TH) and TH receptors (TRs) α and β act by binding to TH response elements (TREs) in regulatory regions of target genes. This nuclear signaling is established as the canonical or type 1 pathway for TH action. Nevertheless, TRs also rapidly activate intracellular second-messenger signaling pathways independently of gene expression (noncanonical or type 3 TR signaling). To test the physiological relevance of noncanonical TR signaling, we generated knockin mice with a mutation in the TR DNA-binding domain that abrogates binding to DNA and leads to complete loss of canonical TH action. We show that several important physiological TH effects are preserved despite the disruption of DNA binding of TRα and TRβ, most notably heart rate, body temperature, blood glucose, and triglyceride concentration, all of which were regulated by noncanonical TR signaling. Additionally, we confirm that TRE-binding–defective TRβ leads to disruption of the hypothalamic–pituitary–thyroid axis with resistance to TH, while mutation of TRα causes a severe delay in skeletal development, thus demonstrating tissue- and TR isoform-specific canonical signaling. These findings provide in vivo evidence that noncanonical TR signaling exerts physiologically important cardiometabolic effects that are distinct from canonical actions. These data challenge the current paradigm that in vivo physiological TH action is mediated exclusively via regulation of gene transcription at the nuclear level.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Thyroid hormone (TH) plays an essential role in organ development and metabolic homeostasis. The effects of 3,3′,5-triiodo-L-thyronine (T3), the active TH, are mediated by TH receptors (TRs) α and β. Canonical TR signaling controls gene expression in the nucleus directly. This mechanism was recently classified as type 1 TR-dependent TH signaling (Fig. 1A) (1): TRs bind to thyroid hormone response elements (TREs) in regulatory regions of target genes. In the absence of T3, TRs recruit corepressors with histone deacetylase activity, preventing transcription. T3 binding to TRs causes the release of corepressors, which are replaced by coactivators that engage histone acetylases and RNA polymerase II to activate gene transcription (2, 3). Thus, TRs act as hormone-dependent transcription factors. Consequently, the current paradigm of TH action attributes its physiological effects to the genes directly regulated via canonical TR signaling on DNA.
Fig. 1.
Beyond that, additional and distinct mechanisms of TR signaling have been demonstrated. T3 and TRβ can modulate intracellular second messenger signaling, e.g., the PI3K pathway (TR-dependent signaling of TH without DNA binding, type 3) (1, 4–7). A specific molecular mechanism for PI3K activation by TRβ has been identified and demonstrated to be essential for normal synapse development and plasticity in mice by mutating a specific tyrosine residue to phenylalanine in TRβ (7). Noncanonical signaling by TRα remains ill-defined, and its physiological importance is unknown. Such non-type 1 TH/TR action is considered noncanonical, because it is independent of DNA binding and TRE binding, does not require gene transcription or protein synthesis as the initial event, and may, therefore, also be more rapid (Fig. 1B).
To determine the physiological consequences of noncanonical TH/TR action in vivo, we generated TRαGS and TRβGS mice that lack canonical TR signaling by replacing glutamic acid (E) and glycine (G) in the first zinc finger of the DNA-binding domain of TRα and TRβ with glycine and serine (S). This GS mutation severely impairs TR binding to TREs and abrogates canonical TR signaling (8, 9). Since TRE binding is abolished, only the noncanonical TH actions are preserved in TR-GS–mutant mice. By contrast, the TR can mediate both canonical and noncanonical actions of TH in WT mice, whereas both actions are absent in TR-KO mice (Fig. 1C).
Therefore we hypothesized that a comparison of WT, TR-KO, and TR-GS mice would determine whether canonical or noncanonical TR signaling underlies physiological TH responses. A phenotypic difference between WT and TR-KO mice and similarity of TR-KO and TR-GS mice would indicate that canonical TR signaling is required for a particular phenotype (Fig. 1C). Conversely, if a phenotype is concordant between TR-GS and WT mice but discordant from TR-KO mice, then only noncanonical signaling would be required.
We also included TRβ147F mice, in which tyrosine 147 of TRβ is changed to phenylalanine (Y147F), as an additional control for TRβ signaling. Tyrosine 147 is crucial for TRβ-mediated activation of PI3K, and replacement by phenylalanine abrogates PI3K activation. Nevertheless, the classical TRE-mediated actions of TRβ are unaffected in TRβ147F mice (7). Thus, the TRβ147F mice are a specific negative control for noncanonical, PI3K-mediated TH/TRβ effects, as TRβ-mediated PI3K signaling is absent in these mice (Fig. 1C).
Here we provide in vivo evidence that noncanonical TR signaling is physiologically important and contributes significantly to TH action, specifically for the regulation of body temperature, oxygen consumption (VO2), locomotor activity, heart rate, and triglyceride and glucose concentrations, important parameters of cardiometabolic homeostasis and energy balance.
Results
The GS Mutation Abrogates Canonical TR Signaling in Vitro and in Vivo.
Canonical TR signaling consists of TR binding to TREs in the regulatory regions of target genes and subsequent activation of gene expression. ChIP-sequencing (ChIP-seq) analyses in mouse liver reported a DR+4 motif, a direct repeat of two 5′-AGGTCA-3′ half-sites in the same orientation separated by four nucleotides, as the most prevalent TRE in T3-induced genes (10–12). This consensus DR+4 TRE was used in a fluorescent EMSA to test DNA binding affinity of the TRα and TRβ variants (TRβ, TRβ125GS, TRβY147F, TRα, and TRα71GS). Only TRβ, TRβY147F, and TRα bound to the DR+4 probe; the TRβ125GS and TRα71GS receptors did not (Fig. 2A). The GS mutation abrogated DNA binding of both TRα and TRβ. Next, we tested the transcriptional activity of TRα71GS, TRβ125GS, and TRβY147F mutants in vitro using a TH-responsive luciferase reporter plasmid (DR+4-TKLuc) in comparison with WT TRs. Empty vector and TR mutants TRαG291R and TRβG345R served as controls. These mutant TRs cannot bind T3 and cannot activate gene transcription after T3 treatment (13). T3 increased luciferase activity eightfold with WT TRα, TRβ, and TRβY147F but not with empty vector or TRβG345R and TRαG291R (Fig. 2B). Luciferase activity was not increased by T3 with the TRα71GS and TRβ125GS mutants, demonstrating in vitro that the GS mutation in the DNA-binding domain abolishes the canonical TRE-mediated transcriptional activity of TRα and TRβ. This was confirmed by experiments with a common variant of the 3′ half-site (AGGACA) (Fig. S1 A and B).
Fig. 2.
To abolish the canonical TR action in vivo, we introduced the GS mutation into the murine WT Thra and Thrb gene loci, generating TRαGS (ThraGS/GS) and TRβGS (ThrbGS/GS) mice. We determined the expression of known TH-responsive genes in these mice in comparison with WT and the respective TR-KO mice (TRα0 and TRβ− mice; genotypes are Thra0/0 and Thrb−/−, respectively). As TRα is predominantly expressed in heart and TRβ in liver, we studied these two tissues. In heart, expression of Myh6 and Myh7 mRNAs in TRαGS mice differed significantly from their expression in WT mice and was comparable to that in TRα0 mice (Fig. 3A), demonstrating that the GS mutation in TRα abrogates canonical TH/TRα action in vivo. For TRβ-mutant mice, the different systemic TH concentrations in WT, TRβGS, and TRβ− mice had to be equalized. Thus, we determined TH-induced gene expression in hypothyroid animals without and with T3 treatment. In livers of WT mice, T3 induced expression of Dio1, Spot14, Me1, and Bcl3 mRNAs (Fig. 3B and Fig. S1 C and D). This induction was equally reduced in both TRβ− mice and TRβGS mice, demonstrating that the GS mutation in TRβ has the same effect on canonical TH/TR-mediated gene expression as the complete absence of TRβ in TRβ− mice.
Fig. 3.
To evaluate recruitment of TRβ mutants to DNA, we performed ChIP against TRβ and its heterodimerization partner retinoid X receptor alpha (RXRα) in liver tissue from T3-treated animals and evaluated the occupancy of known TR-binding sites previously identified in mouse liver tissue, e.g., in the Dio1, NcoR2, H13, and Strbp genes (Fig. 3C and Fig. S1E). Both WT and the TRβ147F mutant occupy these TR-binding sites together with RXRα, whereas the TRβGS mutant and TRβ− show little or no enrichment, confirming that the GS mutation in the DNA-binding domain of TRβ disrupts TRβ’s interaction with chromatin. Moreover, evaluation of histone 3 lysine 27 acetylation (H3K27Ac) at these binding sites showed that impaired occupancy with TRβ in TRβ− and TRβGS mice results in reduced histone acetylation (Fig. 3C and Fig. S1E). To evaluate the effect on histone acetylation genome-wide, we performed ChIP-seq against H3K27Ac in the liver of hypo- and T3-treated hyperthyroid WT and TRβ mutant animals. Sequenced tags were quantified at all previously identified DNase-accessible regions (12), and we identified more than 500 regions where H3K27Ac was differently regulated by T3 in WT animals (Fig. S1F). We quantified H3K27Ac at these differentially regulated regions in liver samples from hypo- and T3-treated hyperthyroid WT and TRβ− animals and animals carrying the GS and 147F TRβ mutants. Hierarchical clustering revealed that T3-induced H3K27Ac in animals with the 147F mutation is very similar to that observed in WT animals. By contrast, the TRβGS mutation or TRβ deletion resulted in attenuated H3K27Ac after T3 treatment (Fig. 3D). Consistent with this, H3K27Ac is increased in response to T3 at previously identified TR-binding sites in the Dio1 gene loci in WT and TRβ147F mice but not in TRβGS or TRβ− animals (Fig. 3E).
We then used transcriptome analysis to determine whether TRβ125GS and TRβ147F mutants possess residual transcriptional activity affecting known target genes or had acquired new target-gene specificity. Hypothyroid WT, TRβ−, TRβGS, and TRβ147F mice were treated with a single dose of T3, and hepatic gene expression was determined after 6 h by microarray analysis. Fig. 3F shows the results in a dendrogram with mice grouped by similarity of their gene-expression patterns, not by genotype. The gene-expression patterns in WT and TRβ147F mice were clearly different from those in TRβ− and TRβGS mice and formed their own branch in the hypothyroid and T3-treatment groups. Importantly, TRβ− and TRβGS mice were grouped together because of the similarity of their hepatic gene-expression patterns. In fact, T3-treated TRβ− and TRβGS mice could not be distinguished by gene-expression pattern. Thus, the GS mutation renders the TR transcriptionally nonfunctional in vivo.
Regulation of the Hypothalamic–Pituitary–Thyroid Axis Requires Canonical TRβ Signaling.
TH production and secretion is regulated by the hypothalamic–pituitary–thyroid (HPT) axis negative-feedback loop. Inhibition of thyroid-stimulating hormone (TSH) is mediated by TRβ (14–16) and depends on DNA binding of TRβ (9). In accordance with these reports, basal serum TSH was significantly higher in TRβ− and TRβGS mice than in WT mice (Fig. 4A). As a consequence of elevated TSH, serum thyroxine (T4) and T3 concentrations were also significantly higher in TRβ− and TRβGS mice than in WT mice (Fig. 4B and Fig. S2A). The combination of elevated circulating TH concentrations and elevated TSH demonstrates central resistance to TH due to the lack of the receptor in TRβ− mice and, importantly, a corresponding absence of canonical TRβ action in the HPT axis of TRβGS mice. In TRβ147F mice with intact DNA binding and canonical signaling, serum T3 and T4 concentrations and pituitary TSHβ expression were normal (7). TSH, T4, and T3 concentrations in TRα0 mice (devoid of TRα1 and TRα2) and TRαGS mice were not different from those in WT mice (Fig. 4 A and B and Fig. S2A).
Fig. 4.
Skeletal Development Requires Canonical TRα Signaling.
We measured growth curves from WT, TRα0, TRαGS, TRβ−, and TRβGS mice. Growth curves of TRβ− and TRβGS mice were not different from those of WT mice (Fig. S2 B and C). However, TRα0 and TRαGS mice had a similar delay in linear growth and gain in body weight (BW) compared with WT mice (Fig. 4 C–F). To investigate the underlying cause, skeletal analysis of TRα0 and TRαGS mice was performed after weaning at postnatal day 21 (P21). X-ray microradiography revealed a similar decrease in bone length and vertebral height in TRαGS and TRα0 mice, together with a reduction in bone mineral content in vertebrae but no difference in long bones (Fig. 5A and Fig. S3). Histological analysis of the growth plate revealed a delay in endochondral ossification similarly affecting both TRαGS and TRα0 mice (Fig. 5B). This delay comprised a decrease in the size of the secondary ossification center together with an increase in the reserve zone width and a decrease in the hypertrophic zone, features that are characteristic of impaired T3 action (17, 18). High-resolution micro-computed tomography (micro-CT) imaging demonstrated epiphyseal dysgenesis, increased trabecular bone mass, and reduced metaphyseal inwaisting consistent with a bone-modeling defect and delayed endochondral ossification (Fig. 5 C and D). These are all classic features of impaired T3 action in bone (18, 19). In summary, these data demonstrate an equivalent delay in skeletal development due to similar loss of TRE-mediated canonical TRα signaling in TRα0 and TRαGS mice. These findings establish that T3 actions in the skeleton are mediated by canonical actions of TRα on TREs.
Fig. 5.
Noncanonical TRβ Action Decreases Blood Glucose Concentration.
T3 treatment has been shown to reduce serum glucose concentration rapidly in lean and obese mice (20). We hypothesized that such a rapid TH effect could be mediated noncanonically by TRs and thus be present in WT and in TRαGS and TRβGS mice. We tested this hypothesis first in WT, TRβ−, TRβGS, and TRβ147F mice. Basal serum glucose after 1 h of fasting was not different among the TRβ genotypes (Table S1). T3 injection rapidly decreased blood glucose within 60 min in WT mice (Fig. 6A and Table S1). Strikingly, this hypoglycemic effect of T3 was abolished in TRβ− mice but was fully preserved in TRβGS mice. We conclude that the decrease in blood glucose after T3 treatment is mediated by TRβ, as it is absent in TRβ− mice. The similar decrease in blood glucose seen in both WT and TRβGS mice demonstrates that this effect is mediated by noncanonical TRβ signaling. Furthermore, this effect occurs within 60 min, which is likely too rapid to be dependent on RNA transcription and new protein translation. This conclusion is further supported by the failure of T3 to induce hypoglycemia in TRβ147F mice, which also demonstrates that PI3K activation by TRβ is required to lower the glucose concentration. By contrast, TRα is not involved in hypoglycemic response to T3, as it is unable to compensate for the lack of TRβ in TRβ− mice.
Fig. 6.
Noncanonical TRβ Signaling Maintains Normal Serum and Liver Triglyceride Concentration.
Serum triglyceride concentration is under TH control (21, 22). Previous studies have shown elevated serum triglyceride concentration in TRβ knockin mice with resistance to TH (TRβPV), indicating that TRβ mediates TH regulation of triglycerides (23). We therefore measured serum triglyceride concentrations in untreated WT, TRβ−, TRβGS, and TRβ147F mice. Triglyceride concentrations were significantly higher in TRβ− and TRβ147F mice than in WT mice (300 ± 61 and 264 ± 26 vs. 152 ± 23 mg/dL; P < 0.05) but not in TRβGS mice (123 ± 34 mg/dL; n.s.) (Fig. 6B). Cholesterol, albumin, and total protein concentrations were not different among WT, TRβ−, TRβGS, and TRβ147F mice (Table S2). These results demonstrate that in the absence of noncanonical TRβ action the serum triglyceride concentration is elevated in TRβ− and TRβ147F mice. A possible explanation for increased serum triglyceride concentration was increased triglyceride synthesis in the liver. Compared with WT mice, the expression of several key enzymes involved in triglyceride synthesis (Scd1, Me1, Fasn) was increased in livers of TRβ− and TRβ147F mice but was not increased or was increased to a lesser extent in TRβGS mice (Fig. 6 C and E and Fig. S4). Accordingly, we found higher triglyceride concentrations in the livers of TRβ− and TRβ147F mice but not in TRβGS mice, replicating the pattern found in serum (Fig. 6 D and F). The increased serum and hepatic triglyceride concentrations and hepatic triglyceride synthesis in TRβ− and TRβ147F mice and the similarity of the phenotype in TRβGS mice to that of WT mice demonstrates that the hypertriglyceridemia is due to the absence of noncanonical TRβ signaling in TRβ− and TRβ147F mice.
Noncanonical TRβ Signaling Contributes to Regulation of Body Temperature, Oxygen Consumption, and Locomotor Activity.
Body temperature homeostasis is crucially linked to TH, and deletion of either TRα or TRβ results in reduced body temperature, although not always significantly (24). By contrast, the mean body temperature of TRβGS mice was 0.9 °C higher than that of TRβ− mice and 0.5 °C higher than in WT mice (Fig. 7A), suggesting a noncanonical TRβ effect on body temperature. To support this conclusion further, we repeated body temperature measurements in TRβ WT, TRβ−, and TRβGS mice on a TRα0 genetic background. Body temperature in TRα0β− double-KO mice was markedly reduced compared with WT mice (to 34.9 °C vs. 37.0 °C) (Fig. 7A), in line with previous reports (24–26). However, the body temperature of TRβGS mice on the TRα0 background was ∼1 °C higher than in TRα0β− double-KO mice (Fig. 7A). These results demonstrate a TRβ-specific effect on energy metabolism that is independent of TRα and is mediated via noncanonical actions. We then determined VO2 by indirect calorimetry and compared the groups using linear regression models including body mass as a covariate. Average VO2 was increased in TRβGS mice and reduced in TRβ− mice compared with WT controls (Fig. 7 C and D and Table S3). Minimum VO2 as a proxy of resting metabolic rate was also increased in TRβGS mice. Food consumption did not differ between genotypes in this short period (Table S3). Furthermore, distance traveled was increased in TRβGS mice (Fig. 7B). Thus, the noncanonical TRβ-specific effects of TH increase the metabolic rate and locomotor activity.
Fig. 7.
Noncanonical TRα Signaling Contributes to Extrinsic Regulation of Heart Rate.
TRα is the predominant TR isoform in the heart, and regulation of heart rate is a well-known physiological effect of TH and TRα (26). We determined heart rate using a noninvasive ECG in untreated WT, TRα0, and TRαGS mice. Heart rate was significantly reduced in TRα0 mice compared with WT mice but not in TRαGS mice (Fig. 8A). The heart rate in TRαGS mice was normal despite similar reductions in TH target genes known to be involved in heart rate regulation in both TRα0 and TRαGS mice (Fig. 8B). The similar phenotype in WT and TRαGS mice demonstrates that noncanonical TRα signaling contributes to normal heart rate in TRαGS mice. To distinguish between intrinsic and extrinsic cardiac effects, we determined basal heart rate in isolated perfused hearts using a Langendorff apparatus. Ex vivo, the cardiac rhythm was similar in TRαGS and TRα0 mice (Fig. 8C), indicating that noncanonical TRα signaling controls heart rate by an extrinsic mechanism, likely via central regulation of the autonomous nervous system.
Fig. 8.
Discussion
The physiological role of TRs has been studied in vivo for more than 15 y by comparing WT with TR-KO mice. As both canonical and noncanonical, or type 1 and type 3 (1), TR signaling are present in WT mice, but both are absent in TR-KO mice, such a comparison could not distinguish between the two mechanisms. To separate both modes of TH/TR signaling, we abolished DNA binding of TRs in TRαGS and TRβGS mice. Strikingly, despite complete loss of canonical TR signaling, the TRαGS and TRβGS mice did not exhibit a TR-KO phenotype. Rather, for several established TH effects their phenotype was indistinguishable from WT mice. Thus, these results provide in vivo evidence for the physiological importance of noncanonical TR signaling. Furthermore, they demonstrate that noncanonical TR signaling can be mediated by both TRα and TRβ in an isoform- and tissue-specific manner. These findings have profound implications for the role of TRs in metabolism and physiology and explain the pathophysiology in diseases caused by the various TR mutations.
Canonical Actions of TRs.
An important physiological role of TH is negative feedback in the HPT axis. TRβGS mice had a phenotype similar to that of TRβ− mice with lack of TRβ-mediated negative feedback in the pituitary and TH resistance (9), whereas the HPT axis in TRβ147F mice was unaffected (7). A higher TSH in TRβGS mice could indicate that they are slightly more resistant than the TRβ− mice. A potential but highly speculative explanation is cofactor squelching. TRβGS does not bind to DNA but is still capable of cofactor binding. This could, theoretically, reduce cofactor availability for TRα and reduce a minimal repressive TRα effect on TSH expression, resulting in even higher TSH concentrations than in TRβ-KO mice. Delayed longitudinal growth and bone mineralization are typical features of TRα0 mice and were replicated in TRαGS mice. Therefore negative regulation of the HPT axis in TRβGS mice and bone development in TRαGS mice are solely dependent on canonical TR/TRE-mediated transcriptional regulation. Nevertheless, a recently identified short variant of TRα, p30 TRα1, associates with the plasma membrane, increases nitric oxide and cGMP production, and activates the ERK and PI3K pathways in vitro (27). Furthermore, it has been suggested that p30 TRα may have a role in the regulation of adult bone formation. Here, we studied bone development in 3-wk-old mice and determined linear growth until cessation of the experiment at 10 wk of age. While skeletal development and linear growth were unequivocally canonical, we cannot exclude a contribution of noncanonical actions of TRα in adult bone remodeling.
Noncanonical Actions of TRβ.
Serum glucose decreased rapidly only in WT and TRβGS mice but not in TRβ− and TRβ147F mice despite intact canonical TRβ signaling in the TRβ147F mice. The phenotype in TRβGS mice demonstrates that noncanonical TRβ action alone suffices to generate physiological responses. Similarly, triglyceride synthesis and serum and liver triglyceride content were elevated in TRβ− and TRβ147F mice that lack noncanonical TRβ action, suggesting reduced consumption of triglyceride substrates. Body temperature was higher in TRβGS than in WT mice, probably as a consequence of elevated serum TH concentration due to the disruption of the HPT axis in these mice and the retained noncanonical action of TRβGS. Additionally, VO2 and distance traveled were higher in TRβGS mice. Increased energy metabolism, body temperature, and locomotor activity and lower triglyceride concentrations are physiological consequences of noncanonical TRβ action.
Noncanonical Actions of TRα.
Noncanonical TH effects on the cardiovascular system have been reported, especially a rapid decrease in blood pressure and PI3K signaling pathway activation in the heart (28–30). In TRαGS mice the heart rate in vivo was similar to that in WT mice, but, consistent with the expression of cardiac pacemaker genes, the ex vivo cardiac rate in TRαGS hearts was similar to that of TRα0 hearts. These results suggest that the noncanonical TH/TRα mechanism contributing to normal heart rate in vivo is not intrinsic to the heart. As ECG data were obtained from mice without anesthesia, it is possible that the response to stress or the activity of the sympathetic nervous system was reduced in TRα0 mice but preserved in TRαGS mice. Therefore we propose that noncanonical TH/TRα action influences heart rate via the autonomous nervous system and not via direct actions in the heart. Interestingly, this appears to be similar to the regulation of locomotor activity, which also is likely centrally mediated. Where and how TRα acts has yet to be determined.
Evolution of Noncanonical TH Signaling.
Our findings suggest that increased noncanonical TRβ signaling increases energy metabolism and body temperature. Support for this interpretation may be derived from evolution: TH stimulates thermogenesis in homeothermy, a feature of mammals. Interestingly, the tyrosine motifs in TRβ, which are required for noncanonical PI3K activation by TRβ, are found in all mammalian TRβ ortholog sequences but not in sequences from poikilothermic animals, such as alligator, clawed toad, or zebrafish (7). Hence, noncanonical TR signaling may be a recently acquired adaptation in evolution that provides mammals with a novel mechanism by which TH can regulate energy metabolism, body temperature, and activity. Development of noncanonical TR signaling relatively late in evolution, long after TRs assumed their canonical role as ligand-dependent transcription factors, may also explain why the physiological effects of canonical and noncanonical TR signaling are so clearly distinct from one another.
Negative Regulation of Gene Expression.
A persisting point of controversy is whether DNA binding of TRs is required for negative regulation of gene expression by TRs. TSH elevation in TRβ− and TRβGS mice clearly demonstrates that, in agreement with previous reports (9), inhibition of TSH expression in vivo requires DNA binding. Results from TRβGS mice indicate that noncanonical TH action may contribute to negative regulation. Scd1 expression in HepG2 cells and in mouse liver is negatively regulated by TH (Fig. S5) (31). Compared with WT mice, hepatic Scd1 expression was elevated in TRβ− mice but not in TRβGS mice. Therefore, negative regulation of Scd1 expression in vivo does not require DNA binding of TRβ. In microarray studies, however, we did not detect down-regulation of Scd1 within 6 h. Two explanations seem plausible: Either Scd1 is directly negatively regulated by TH/TRβ without DNA binding by an unknown mechanism that takes longer than 6 h or down-regulation of Scd1 in liver is only a secondary effect in response to noncanonical TRβ effects on metabolism, which also take more than 6 h. The fact that, in addition to Scd1, several other genes involved in triglyceride synthesis also were negatively regulated by T3 suggests the latter. Repression of these genes appears to be a coordinated response to reduced substrate availability for triglyceride synthesis, possibly due to increased metabolism. Repression of Myh7 appeared to be partially preserved in hearts of TRαGS mice. Expression of Myh7 is negatively regulated by miRNA miR-208a, which is encoded by an intronic sequence of Myh6. Repression of Myh7 by increased TH concentration is an indirect effect of Myh6 induction and, consequently, miR-208a expression. Our interpretation of the difference in Myh7 expression in TRα0 and TRαGS mice is that the relationship between Myh6 and Myh7 may not be linear. However, a partial noncanonical repressive effect of TRαGS cannot be ruled out.
Role of Noncanonical PI3K Activation.
The best-studied noncanonical action of TRβ is rapid activation of PI3K (5–7). We therefore studied the TRβ147F mouse as a control, because in this mouse model the activation of PI3K by TRβ is selectively abrogated. In the blood glucose response to T3 and serum and liver triglyceride concentrations, the phenotype of TRβ147F mice was similar to that of TRβ− mice, demonstrating that these effects are indeed mediated by TRβ activation of PI3K. As insulin also signals via PI3K to decrease blood glucose by increasing the translocation of glucose transporter 4 (GLUT4) to the plasma membrane, we hypothesize that insulin and noncanonical TRβ signaling converge at the PI3K pathway. This mechanism of TRβ action would explain previous observations in which T3 treatment increased glucose uptake into the thymus in rats independent of protein synthesis (32), increased glucose uptake into L6 muscle cells in a PI3K-depentent manner in vitro (33), and increased the translocation of GLUT4 in skeletal muscle of rats (34). TRα also activates signaling pathways (29, 35), possibly as a short p30 TRα variant that indirectly activates ERK and PI3K after T3 binding and nitric oxide and cGMP production (27) or by direct protein–protein interaction between TRα and PI3K (29, 35, 36). The noncanonical mechanism by which TRα influences heart rate remains unknown.
In conclusion, noncanonical, DNA-independent TR signaling contributes significantly to the physiological actions of TH, and these noncanonical actions appear to predominantly regulate energy homeostasis. These findings demonstrate that TRα and TRβ mediate both canonical and noncanonical signaling in vivo, establishing a paradigm shift for TH action.
Materials and Methods
Extended and detailed material and methods are provided in Supporting Information. All animal studies were approved by the local authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen). All tests performed at the German Mouse Center (GMC) were approved by the responsible authority of the Regierung von Oberbayern. TRα and TRβ WT and KO and TRβ147F mice were genotyped as previously described (7, 24, 37). TRβ147F mice were provided by D.L.A. (7). TR-GS mice were generated via the zinc finger nuclease approach (CompoZr Targeted Integration-Kit-AAVS; Sigma-Aldrich) (38, 39). Mice were rendered hypothyroid by maintenance on a low-iodine diet and administration of methimazole and perchloride via the drinking water. Animals were treated with experiment-specific concentrations of T3 by i.p. injections. Serum TSH was measured as previously described (40), and TH serum concentrations were determined by ELISA (DRG Diagnostics). Bone morphology and histology were investigated with standardized methods described in detail in Supporting Information and refs. 18 and 19. Body temperature was determined with a rectal probe, and blood glucose was measured with a Contour XT glucometer (Bayer). Indirect calorimetry was performed at the GMC (Munich) according to standardized in-house protocols. Heart rate was determined either with an ECG in fully conscious mice or ex vivo via Langendorff apparatus. Experimental procedures for transfection, luciferase reporter assay, fluorescent EMSA, qRT-PCR (41), Oil red O staining, and immunoblotting have been published previously and are described together with fluorescent DNA probes, primers, and antibodies in Supporting Information. Microarray was done with a MG-430_2.0 gene chip (Affymetrix). ChIP and ChIP-seq were performed as previously described (12, 42) with minor modifications explained in Supporting Information. Microarray and ChIP-seq data have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (43) and are accessible through the following GEO Series accession numbers: GSE93864 (microarray) and GSE104877 (ChIP-seq). Statistical analysis was performed with GraphPad Prism6, and differences were considered significant when P < 0.05.
Data Availability
Data deposition: Data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (accession nos. GSE93864 for microarray data and GSE104877 for H3K27 ChIP-seq Illumina sequencing).
Acknowledgments
We thank B. Gloss at the National Institute of Environmental Health Sciences for providing breeding pairs of TRβ147F mice. Illumina sequencing was performed by Ronni Nielsen with support from the VILLUM Center for Bioanalytical Sciences at the University of Southern Denmark. The TR C1 antibody was provided by Sheue-yann Cheng of the NIH. This work was supported by Deutsche Forschungsgemeinschaft Grants MO1018/2-1/2 (to L.C.M.), FU356/7-1 (to D.F.), and KO 922/17-1/2 (to J.K.) within the priority program Thyroid Trans Act SPP 1629; by Charité funds (J.K.); and in part by NIH Grant R37DK15070 (to S.R.), the Seymour J. Abrams Fund for Thyroid Research, Wellcome Trust Strategic Award 101123, and Wellcome Trust Project Grant 094134 (to J.H.D.B. and G.R.W.). Studies at the GMC were supported by German Federal Ministry of Education and Research Infrafrontier Grant 01KX1012 (to M.H.d.A.). L.G. was supported by grants from the Lundbeck Foundation and Novo Nordisk Foundation. D.L.A. was supported by Award Z01-ES102285 from the Intramural Program at the National Institute of Environmental Health Sciences of the NIH. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestive and Kidney Diseases or the NIH.
Supporting Information
Supporting Information (PDF)
- Download
- 812.74 KB
Dataset_S01 (DOCX)
- Download
- 27.61 KB
References
1
F Flamant, et al., Thyroid hormone signaling pathways. Time for a more precise nomenclature. Endocrinology 158, 2052–2057 (2017).
2
PM Yen, Physiological and molecular basis of thyroid hormone action. Physiol Rev 81, 1097–1142 (2001).
3
TM Ortiga-Carvalho, AR Sidhaye, FE Wondisford, Thyroid hormone receptors and resistance to thyroid hormone disorders. Nat Rev Endocrinol 10, 582–591 (2014).
4
NM Storey, JP O’Bryan, DL Armstrong, Rac and Rho mediate opposing hormonal regulation of the ether-a-go-go-related potassium channel. Curr Biol 12, 27–33 (2002).
5
X Cao, F Kambe, LC Moeller, S Refetoff, H Seo, Thyroid hormone induces rapid activation of Akt/protein kinase B-mammalian target of rapamycin-p70S6K cascade through phosphatidylinositol 3-kinase in human fibroblasts. Mol Endocrinol 19, 102–112 (2005).
6
NM Storey, et al., Rapid signaling at the plasma membrane by a nuclear receptor for thyroid hormone. Proc Natl Acad Sci USA 103, 5197–5201 (2006).
7
NP Martin, et al., A rapid cytoplasmic mechanism for PI3 kinase regulation by the nuclear thyroid hormone receptor, TRβ, and genetic evidence for its role in the maturation of mouse hippocampal synapses in vivo. Endocrinology 155, 3713–3724 (2014).
8
N Shibusawa, AN Hollenberg, FE Wondisford, Thyroid hormone receptor DNA binding is required for both positive and negative gene regulation. J Biol Chem 278, 732–738 (2003).
9
N Shibusawa, et al., Thyroid hormone action in the absence of thyroid hormone receptor DNA-binding in vivo. J Clin Invest 112, 588–597 (2003).
10
S Ayers, et al., Genome-wide binding patterns of thyroid hormone receptor beta. PLoS One 9, e81186 (2014).
11
P Ramadoss, et al., Novel mechanism of positive versus negative regulation by thyroid hormone receptor β1 (TRβ1) identified by genome-wide profiling of binding sites in mouse liver. J Biol Chem 289, 1313–1328 (2014).
12
L Grøntved, et al., Transcriptional activation by the thyroid hormone receptor through ligand-dependent receptor recruitment and chromatin remodelling. Nat Commun 6, 7048 (2015).
13
A Sakurai, et al., Generalized resistance to thyroid hormone associated with a mutation in the ligand-binding domain of the human thyroid hormone receptor beta. Proc Natl Acad Sci USA 86, 8977–8981 (1989).
14
D Forrest, et al., Recessive resistance to thyroid hormone in mice lacking thyroid hormone receptor beta: Evidence for tissue-specific modulation of receptor function. EMBO J 15, 3006–3015 (1996).
15
RE Weiss, et al., Thyrotropin regulation by thyroid hormone in thyroid hormone receptor beta-deficient mice. Endocrinology 138, 3624–3629 (1997).
16
ED Abel, RS Ahima, ME Boers, JK Elmquist, FE Wondisford, Critical role for thyroid hormone receptor beta2 in the regulation of paraventricular thyrotropin-releasing hormone neurons. J Clin Invest 107, 1017–1023 (2001).
17
JH Bassett, GR Williams, Role of thyroid hormones in skeletal development and bone maintenance. Endocr Rev 37, 135–187 (2016).
18
JH Bassett, et al., Optimal bone strength and mineralization requires the type 2 iodothyronine deiodinase in osteoblasts. Proc Natl Acad Sci USA 107, 7604–7609 (2010).
19
JH Bassett, et al., Thyroid hormone receptor α mutation causes a severe and thyroxine-resistant skeletal dysplasia in female mice. Endocrinology 155, 3699–3712 (2014).
20
Y Lin, Z Sun, Thyroid hormone potentiates insulin signaling and attenuates hyperglycemia and insulin resistance in a mouse model of type 2 diabetes. Br J Pharmacol 162, 597–610 (2011).
21
U Petersson, T Kjellström, Thyroid function tests, serum lipids and gender interrelations in a middle-aged population. Scand J Prim Health Care 19, 183–185 (2001).
22
L Johansson, et al., Selective thyroid receptor modulation by GC-1 reduces serum lipids and stimulates steps of reverse cholesterol transport in euthyroid mice. Proc Natl Acad Sci USA 102, 10297–10302 (2005).
23
O Araki, H Ying, XG Zhu, MC Willingham, SY Cheng, Distinct dysregulation of lipid metabolism by unliganded thyroid hormone receptor isoforms. Mol Endocrinol 23, 308–315 (2009).
24
K Gauthier, et al., Genetic analysis reveals different functions for the products of the thyroid hormone receptor alpha locus. Mol Cell Biol 21, 4748–4760 (2001).
25
L Wikström, et al., Abnormal heart rate and body temperature in mice lacking thyroid hormone receptor alpha 1. EMBO J 17, 455–461 (1998).
26
PE Macchia, et al., Increased sensitivity to thyroid hormone in mice with complete deficiency of thyroid hormone receptor alpha. Proc Natl Acad Sci USA 98, 349–354 (2001).
27
H Kalyanaraman, et al., Nongenomic thyroid hormone signaling occurs through a plasma membrane-localized receptor. Sci Signal 7, ra48 (2014).
28
BM Schmidt, et al., Nongenomic cardiovascular effects of triiodothyronine in euthyroid male volunteers. J Clin Endocrinol Metab 87, 1681–1686 (2002).
29
Y Hiroi, et al., Rapid nongenomic actions of thyroid hormone. Proc Natl Acad Sci USA 103, 14104–14109 (2006).
30
JA Kuzman, KA Vogelsang, TA Thomas, AM Gerdes, L-Thyroxine activates Akt signaling in the heart. J Mol Cell Cardiol 39, 251–258 (2005).
31
K Hashimoto, et al., Human stearoyl-CoA desaturase 1 (SCD-1) gene expression is negatively regulated by thyroid hormone without direct binding of thyroid hormone receptor to the gene promoter. Endocrinology 154, 537–549 (2013).
32
J Segal, SH Ingbar, In vivo stimulation of sugar uptake in rat thymocytes. An extranuclear action of 3,5,3′-triiodothyronine. J Clin Invest 76, 1575–1580 (1985).
33
A Gordon, H Swartz, H Shwartz, 3,5,3′ Triiodo-L-thyronine stimulates 2-deoxy-D-glucose transport into L6 muscle cells through the phosphorylation of insulin receptor beta and the activation of PI-3k. Thyroid 16, 521–529 (2006).
34
EL Brunetto, SdaS Teixeira, G Giannocco, UF Machado, MT Nunes, T3 rapidly increases SLC2A4 gene expression and GLUT4 trafficking to the plasma membrane in skeletal muscle of rat and improves glucose homeostasis. Thyroid 22, 70–79 (2012).
35
X Cao, F Kambe, M Yamauchi, H Seo, Thyroid-hormone-dependent activation of the phosphoinositide 3-kinase/Akt cascade requires Src and enhances neuronal survival. Biochem J 424, 201–209 (2009).
36
F Furuya, et al., Ligand-bound thyroid hormone receptor contributes to reprogramming of pancreatic acinar cells into insulin-producing cells. J Biol Chem 288, 16155–16166 (2013).
37
K Gauthier, et al., Different functions for the thyroid hormone receptors TRalpha and TRbeta in the control of thyroid hormone production and post-natal development. EMBO J 18, 623–631 (1999).
38
ID Carbery, et al., Targeted genome modification in mice using zinc-finger nucleases. Genetics 186, 451–459 (2010).
39
X Cui, et al., Targeted integration in rat and mouse embryos with zinc-finger nucleases. Nat Biotechnol 29, 64–67 (2011).
40
J Pohlenz, et al., Improved radioimmunoassay for measurement of mouse thyrotropin in serum: Strain differences in thyrotropin concentration and thyrotroph sensitivity to thyroid hormone. Thyroid 9, 1265–1271 (1999).
41
SA Bustin, et al., The MIQE guidelines: Minimum information for publication of quantitative real-time PCR experiments. Clin Chem 55, 611–622 (2009).
42
M Siersbæk, et al., High fat diet-induced changes of mouse hepatic transcription and enhancer activity can be reversed by subsequent weight loss. Sci Rep 7, 40220 (2017).
43
R Edgar, M Domrachev, AE Lash, Gene expression omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res 30, 207–210 (2002).
44
HH Samuels, F Stanley, J Casanova, Depletion of L-3,5,3′-triiodothyronine and L-thyroxine in euthyroid calf serum for use in cell culture studies of the action of thyroid hormone. Endocrinology 105, 80–85 (1979).
45
A Dobin, et al., STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
46
S Heinz, et al., Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell 38, 576–589 (2010).
47
MI Love, W Huber, S Anders, Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550 (2014).
48
R Kolde, Pretty Heatmaps. R Package, Version 1.0.8. Available at https://cran.r-project.org/web/packages/pheatmap/index.html. Accessed August 9, 2017. (2015).
49
MW Pfaffl, A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 29, e45 (2001).
50
HA Hildebrandt, et al., Kinetics and signal activation properties of circulating factor(s) from healthy volunteers undergoing remote ischemic pre-conditioning. JACC Basic Transl Sci 1, 3–13 (2016).
Information & Authors
Information
Published in
Classifications
Copyright
© 2017. Published under the PNAS license.
Data Availability
Data deposition: Data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (accession nos. GSE93864 for microarray data and GSE104877 for H3K27 ChIP-seq Illumina sequencing).
Submission history
Published online: December 11, 2017
Published in issue: December 26, 2017
Keywords
Acknowledgments
We thank B. Gloss at the National Institute of Environmental Health Sciences for providing breeding pairs of TRβ147F mice. Illumina sequencing was performed by Ronni Nielsen with support from the VILLUM Center for Bioanalytical Sciences at the University of Southern Denmark. The TR C1 antibody was provided by Sheue-yann Cheng of the NIH. This work was supported by Deutsche Forschungsgemeinschaft Grants MO1018/2-1/2 (to L.C.M.), FU356/7-1 (to D.F.), and KO 922/17-1/2 (to J.K.) within the priority program Thyroid Trans Act SPP 1629; by Charité funds (J.K.); and in part by NIH Grant R37DK15070 (to S.R.), the Seymour J. Abrams Fund for Thyroid Research, Wellcome Trust Strategic Award 101123, and Wellcome Trust Project Grant 094134 (to J.H.D.B. and G.R.W.). Studies at the GMC were supported by German Federal Ministry of Education and Research Infrafrontier Grant 01KX1012 (to M.H.d.A.). L.G. was supported by grants from the Lundbeck Foundation and Novo Nordisk Foundation. D.L.A. was supported by Award Z01-ES102285 from the Intramural Program at the National Institute of Environmental Health Sciences of the NIH. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestive and Kidney Diseases or the NIH.
Notes
This article is a PNAS Direct Submission.
Authors
Competing Interests
The authors declare no conflict of interest.
Metrics & Citations
Metrics
Citation statements
Altmetrics
Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFGet Access
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to get full access to it.