Physiologic Significance of Epigenetic Regulation of Thyroid Hormone Target Gene Expression

Type 1 Mode of Action (Former Classic or Canonical Pathways)

Although T₄ is the predominant hormone produced by the thyroid gland, T₃ exerts the majority of known TH effects at the tissue and cellular level, including hypothalamic and pituitary negative feedback regulation. T₃ is mostly generated from the conversion of T₄ in the extrathyroidal tissues via the activity of iodothyronine deiodinase type 1 (D1) and type 2 (D2). In contrast, deiodinase type 3 (D3) converts T₃ and T₄ to relatively inactive metabolites as reverse T₃ (rT₃) and T₂ (3,3′-diiodothyronine). Therefore, in addition to the changes in the production of TH by the thyroid gland, the availability of T₃ in the tissues is also regulated by the intracellular activity of these enzymes [9]. The main pathway of TH action is at the genomic level via the regulating of gene expression by binding to the nuclear receptors of TH. TR belong to a superfamily of nuclear hormone receptors that act as ligand-regulated transcription factors. They bind T₃ with about a 10-fold higher affinity than T₄. Two genes encode for the 2 types of TR, namely THRA and THRB for TRα and TRβ, respectively. TR bind to TH-response elements (TRE) as either homodimers or heterodimers with retinoic X receptors (RXR). In the absence of T₃, unliganded TR interact with other molecules, most notably with transcriptional corepressors, such as silencing mediator of retinoic acid (SMRT) and nuclear corepressor-1 (NCoR1), suppressing the expression of the genes positively regulated by T₃ [10]. Upon ligand binding, TR undergoes a conformational change, attracting a new set of activator proteins, such as SRC-1 (p160/steroid receptor coactivator 1) and histone acetyltransferase (Table 1). Activation of theses enzymes induces modifications in the tridimensional structure of histones, particularly around the target gene promoters fostering the activity of RNA polymerases. In contrast to positively regulated genes, certain genes are negatively regulated by T₃. This is the case with the thyrotropin/thyroid-stimulating hormone (TSH) gene, critical for the feedback control of the TH axis. The TSH gene is downregulated by T₃ and overexpressed during T₃ deprivation. The mechanism of T₃-negative regulation is less well understood [11]. It may result from direct DNA binding of TR to TRE with repressor activity. But, irrespective of the mechanism, what seems clear is that unliganded TR, instead of operating as inactive, passive agents, repress gene expression. Therefore, it was not surprising that the targeted deletion of TRα and TRβ in mice failed to reproduce the effects of TH depletion.

Table 1.

Principal cofactors recruited by thyroid hormone receptors

View Table

Type 2–4 Modes of Action (Former Nongenomic Pathways)

While nuclear receptors of TH have proven useful for understanding the transcriptional changes on TH target genes, there is compelling evidence that several other mechanisms may also play a role. In fact, many actions of TH may occur too quickly to be mediated by enhanced or repressed target gene expression [12]. In addition, T₄ can be more potent than T₃ for some of these effects. Furthermore, increasing evidence shows that T₄ can directly modulate gene expression, depending on the TR subtype and other cellular cofactors. Together, these alternative mechanisms have been known as “nongenomic pathways” of TH action, an inaccurate designation. Expression of target genes can, in fact, be changed in response to TH without TR binding to DNA. Moreover, the direct action of T₃ in the mitochondria via 2 truncated TRα1 isoforms (p28–p43) is not necessarily extragenomic, because the existence of a mitochondrial genome is well ­established [11]. Finally, changes in the expression of TH target genes may be quickly induced by TH through histone modification that facilitates the activity of transcription factors and RNA polymerases. In order to encompass the diversity of the mechanisms involved in TH signaling, a more precise nomenclature was recently formulated that takes into consideration 4 types of cellular responses to TH [13]. Type 1, the canonical pathway, requires DNA binding to mediate the genomic response to TH. In the type 2 mode of action, TR do not bind directly to DNA and tether into DNA by protein-protein interactions that mediate the crosstalk with other transcriptional factors around genomic specific localizations. Type 3 responses are elicited by TR binding to cytosolic proteins, including regulatory proteins as the p85 regulatory subunit of phosphatidylinositol 3-kinase (PI3K). This, in turn, initiates intracellular responses that culminate in the expression of several target genes like those involved in glucose metabolism. Finally, type 4 responses can be triggered by the binding of TH, and preferentially T₄, to a membrane protein called integrin [14]. These actions initiated at the plasma membrane induce the expression of the genes that affect angiogenesis and cell proliferation via activation of the mitogen-activated protein kinase (MAPK) signaling cascade. Integrin αvβ3 is structurally unrelated and has no sequence homology with TR. Therefore, it is not a “receptor” in the classic sense, and this designation should be avoided. Although this 4-way model provides a broad explanation of TH action, it should be noted that the same stimulus may act through multiple mechanisms and some of these are coupled. Furthermore, epigenetic regulatory mechanisms may be required to bring transcriptional factors into close spatial proximity with the TH target genes.

Short and Long-Term Effects of Fetal Hyperthyroidism

During the last 2 decades, we have studied an extended family, with RTHβ harboring a dominantly inherited mutation of THRB which leads to substitution of Arg243→Gln (R243Q) [15]. The mutant TR binds normally to T₃, but, once bound to DNA, the dissociation of the homodimers and corepressors is markedly impaired [16]. This suggests that the receptor DNA-binding domain can modulate the function of the hormone-binding domain via allosteric mechanisms. On the other hand, RTHβ patients are, in the majority of cases, heterozygous for mutations in the THRBgene, suggesting that the pathogenesis of the disease is driven by dominant negative inhibition of wild-type (WT) TRβ. Thus, the mutant TRβ does not act by reducing the amount of a functional TRβ (haploinsufficiency) but by interfering with the function of the WT TRβ (a dominant negative effect). The resulting insensitivity to TH due to the genetic defect of the TRβ is, most of the time, entirely compensated by an overactive thyroid gland which is stimulated by TSH with augmented biologic potency. Consequently, the serum levels of free TH are persistently increased. Fertility was not impaired in the affected couples regardless of whether women or men harbored the mutant THRB. Therefore, the availability of several generations of descendants to affected RTHβ mothers provided us with a model to investigate the effect of exposure to high TH levels during fetal life. Women harboring the THRB mutation had a significantly higher rate of miscarriage than those whose spouses were affected. Data obtained from children born to an affected parent show that only WT infants born to mothers with RTHβ present with a low birth weight. RTHβ mothers with lower TH levels due to prior ablative treatment of the thyroid gland give birth to WT infants of normal weight and nonsuppressed TSH. The same result was obtained by administering antithyroid medication to the affected mothers carrying a WT fetus [17].

In conclusion, these findings suggest that high maternal TH levels could induce a catabolic state during fetal life, similar to what happens in children and adults with uncontrolled hyperthyroidism. Mothers with RTHβ were not thyrotoxic and had no thyroid autoantibodies; we can thus conclude that the miscarriages and hyperthyroid features were the consequences of fetal hyperthyroidism. The main reason that high maternal TH levels affect only the WT progeny (without THRB gene mutation) of RTHβ mothers, is because, like their mothers, the fetuses harboring the mutation are protected from the toxic effects of TH due to the reduced effectiveness of TRβ. Table 2 illustrates the increase of body weight and thyroid function in a WT child born to an RTHβ mother. It shows that intrauterine exposure to high maternal levels of TH caused fetal and neonatal hyperthyroidism. After birth, there was a sharp decline in the serum levels of free T₄. However, the recovery of the infant hypothalamic-pituitary-thyroid (HPT) axis took about 8 months. During this time, the child was supplemented with L-T₄ to prevent neurodevelopmental constraints due to central hypothyroidism. Given the long-lasting effect of fetal hyperthyroidism in TSH suppression, we were unsure whether the abnormalities in the regulation of TSH secretion persist into adulthood.

Table 2.

Increase in body weight and evolution of thyroid function during the first year of life in a wild-type infant born to a mother with RTHβ

View Table

To study the functionality of the HPT axis in WT adults born to affected mothers, we evaluated the sensitivity to TH with the use of a standard sensitive test that measures the TSH response to thyroid-releasing hormone (TRH) after the administration of the biologically active L-T₃. The TSH response to TRH was found to be less suppressible when L-T₃ in WT adults born to a mother with high TH levels [18]. In contrast, the progeny of fathers with RTHβ did not display these consequences as their unaffected mothers had normal TH levels. This finding is consistent with that for reduced sensitivity to TH, suggesting that fetal hyperthyroidism causes life-long effects on the feedback regulation of the HPT axis.

To further investigate the mechanism of reduced sensitivity to TH in WT adults born to mothers with RTHβ, we studied a mouse model where TH levels were of the same magnitude as in humans who have the R243Q mutation of THRB. As in humans, mice born to dams with RTHβ also exhibited a reduced sensitivity to L-T₃ compared to those born to fathers with RTHβ. Their pituitary showed a significantly higher level of D3 mRNA [18]. D3 is a TH-inactivating enzyme that is usually expressed during fetal life. It is known to convert T₄ and T₃ into rT₃ and T₂, respectively. Therefore, increased D3 expression and, consequently, increased T₃ degradation seem to provide a more likely explanation for the reduced sensitivity of the anterior pituitary to administered L-T3. This observation provided a putative mechanism for the development of reduced sensitivity to TH via local reduction of TH-mediated feedback suppression of TSH.

Epigenetic Transmission of Reduced Sensitivity to TH

We followed the inheritance of this epigenetic phenomenon by studying 3 generations of WT adults descended from affected mothers whose parents and grandparents were not exposed to high levels of TH. Tests of TH sensitivity were repeated in adult WT individuals belonging to each of these generations. Reduced sensitivity to TH was documented in the subsequent generations of WT individuals descended from fathers but not from mothers with reduced sensitivity to TH [19]. Therefore, the phenotype persisted over 3 generations and was transmitted down the male line. Since the effect of intrauterine total body and gonadal exposure to elevated TH can potentially affect only the first (F1) and second (F2) generations, respectively, the results obtained from the third generation (F3) confirmed a true inheritance of the epigenetic effect (Fig. 2). This finding favors a transgenerational epigenetic inheritance which has been rarely documented in humans.

View Full Size

Reduced sensitivity to thyroid hormones (TH) is an epigenetic phenomenon transmitted along male line. F0 is the pregnant RTHβ mother. The wild-type (WT) embryo (F1) is directly exposed to high maternal levels of TH. This is also true for the developing germ cell lines in the testicles (male transmission) which may also affect the F2 generation (i.e., the grandmother’s influence). The F3 generation is therefore the first generation that has not been exposed to high levels of TH.

Citation: European Thyroid Journal 9, 3; 10.1159/000506423

• Download Figure
• Download figure as PowerPoint slide

The DIO3 Gene as a Candidate Gene

The mechanism of the transmission of this epigenetic phenomenon remains poorly clarified. It would require incomplete erasure of the parental epigenetic marks in the gametes to allow the transmission of the phenotype over multiple generations. This is known to occur in imprinted genes, such as the gene encoding D3, i.e., DIO3, which is predominantly expressed from the paternal allele, particularly in the central nervous system (CNS) [20]. Thus, the imprinted DIO3 appears a suitable candidate to mediate the transgenerational transmission of reduced sensitivity to TH along the male line. This effect may be mediated by a cell-specific increase of DIO3 expression. Overexpression of DIO3 could be the explanation for our finding of increased levels of D3 mRNA in the pituitaries of WT mice born to affected dams. DIO3 function is to inactivate TH, and it thus regulates the tissue-specific local availability of active TH. The absence of D3 expression, as observed in DIO3 KO mice, resulted in severe thyrotoxicosis in perinatal life, leading to central hypothyroidism due to the suppression of the HPT axis [21]. Finally, DIO3 transcription, particularly in the CNS, is highly sensitive to local hypoxic conditions and to the TH state, functioning as a negative feedback loop in response to increased levels of TH [22]. The question is: How is the effect of fetal hyperthyroidism in the DIO3 gene maintained throughout an entire life and over several generations?

Epigenetic Mechanisms

Methylation

Methylation is known to produce stable and persistent epigenetic markers. Genomic imprinting is associated with a differential methylation (imprint) between parental alleles in key regulatory sequences. The presence of methylcytosine in gene promoters and enhancers inactivates their transcription via 2 mechanisms [23]. First, the methylation of DNA itself may physically impede the binding of the transcriptional proteins to the gene. In the second, and probably more important mechanism, the methylated DNA may be bound by proteins known as methyl-CpG-binding domain (MBD) proteins. MBD proteins then recruit additional proteins to the locus, such as histone deacetylase and other chromatin-remodeling proteins that can modify histone, making DNA less accessible to transcription factors. For example, the expression of the sodium iodide symporter gene (SLC5A5), responsible for the uptake of iodine in the thyroid, was shown to be regulated by cytosine methylation of its promoter [24]. Therefore, in general, DNA methylation is associated with gene repression, while DNA demethylation promotes gene activation.

TR regulates the transcription of the gene encoding de novo DNA methyltransferase (DNMT3a). The expression of DNMT3a can be induced by exogenous T₃ administration. Thus, exposure to high levels of TH may increase the methylation of susceptible sequences of DNA, particularly the CpG dinucleotide sequence. Imprinted genes, such as DIO3, are expressed in a parent-of-origin-specific manner controlled by allele-specific DNA methylation marks established in the gametes. They are frequently organized in clusters in the genome, and several imprinted clusters harbor CCCTC-binding factor (CTCF)-dependent insulators involved in the regulation of monoallelic expression. Insulators are short nucleotide sequences of DNA that establish connections between promoters and enhancers of the gene. When CTCF, an 82-kDa protein with 11 zinc fingers, binds to an insulator sequence, the crosstalk between the enhancer and the gene promoter is hampered and transcription of the gene is blocked (Fig. 3). In this sense, methylation of the imprinting control region enhances DIO3 expression of the paternal allele by preventing gene repression via insulator-binding [25]. It is still not clear why the insulators of the maternal allele of DIO3 gene are less susceptible to methylation. Interestingly, the promoters and enhancers of both alleles of DIO3 are relatively resistant to methylation. In conclusion, DNA methylation is an active and efficient epigenetic mechanism of TH target gene regulation.

View Full Size

The imprinting control region of the gene encoding deiodinase type 3 (DIO3) contains multiple DNA sequences called insulators, which bind the transcriptional repressor CCCTC-binding factor (CTCF). The insulators are methylated when paternally inherited and unmethylated when derived from the mother. CTCF can only bind to unmethylated sequences of maternal origin. Binding of CTCF to insulators leads to chromatin loop formation in a manner that blocks communication between the enhancer and the promoter of DIO3, repressing its expression.

Citation: European Thyroid Journal 9, 3; 10.1159/000506423

• Download Figure
• Download figure as PowerPoint slide

MicroRNAs

miRs are involved in another mechanism of epigenetic regulation of the TH targeted genes. TH is shown to regulate the levels of miR-206/miR-133b in human skeletal muscles, miR208a in the heart, and miR21 and miR181d in the liver [27]. In a mouse model of myocardial infarction, Dio3 is upregulated in cardiomyocytes to create a local hypothyroid condition to increase the regenerative capacity by switching to the fetal gene program. Dio3 seems to be the target of miR-214. Given that miR-214 is negatively regulated by TH, it can be hypothesized that miR-214 and Dio3 form a negative feedback loop in cardiomyocytes and perhaps in several other tissues, including the pituitary tissue. In conclusion, miRs are newly recognized mediators of TH target gene expression regulation at a posttranscriptional level.

Epigenetic modification is also required for the T3-dependent activation of sarcoplasmic reticulum Ca2⁺ ATPase (SERCA2a) [28]. SERCA2a is responsible for intracellular calcium (Ca²⁺) homeostasis and contractility of the heart. SERCA2 removes Ca²⁺ from the cytosol and stores it in the sarcoplasmic reticulum, leading to diastolic relaxation. Expression of SERCA2 is largely dependent on direct regulation by T₃ as well as by the inhibition of its counterpart phospholamban (PLB) [29]. The expression of the gene encoding SERCA2 is positively regulated by TH and could be increased with T₃ treatment. In contrast, PLB is negatively regulated by T₃. Thus, T₃ is able to relax the heart and enhance cardiac output by reducing the amount of cytosolic Ca²⁺ by increasing SERCA2 expression. Furthermore, the cardiac myocytes contain myosin heavy chain (MHC)α and MHCβ, that are positively and negatively regulated by T3, respectively.

Finally, sirtuin-mediated epigenetic mechanisms are also involved the regulation of TH target genes [30]. SIRT1 (the first of 7 different sirtuin enzymes) is a nicotinamide adenine dinucleotide (NAD)-dependent deacetylase that removes the acetyl groups from proteins, such as histones, thereby affecting the expression of TH target genes. The deacetylation of histone is associated with gene repression and requires NAD⁺. Therefore, the activity of SIRT1 is connected to the ratio NAD⁺/NADH that reflects the metabolic status of the cell. SIRT1 and TRβ1 were found to physically interact with each other and SIRT1 enhances the TH-responsiveness in hepatic cells. The SIRT1-TRβ1 interaction may be protective for reducing TH-mediated transcriptional activation when general transcription needs to be reduced during times of nutrient or energy stress. In conclusion, SIRT1 is a gene-specific TRβ1 coregulator, and the TRβ1/SIRT1 interaction could play an important role in the regulation of the liver metabolic response. The physiologic relevance of the negative regulation of SIRT1 by TH remains unclear. In Table 3, we summarize the epigenetic regulation of some of the TH-responsive genes that have a broad metabolic impact.

Table 3.

Epigenetic regulation of the thyroid hormone (TH) target genes with a broad metabolic impact

View Table