OUDE Hypoforum

Novel insights into thyroid hormones from the study of common genetic variation

Effects of thyroid hormones in individual tissues are determined by many factors beyond their serum levels, including local deiodination and expression and activity of thyroid hormone transporters. These effects are difficult to examine by traditional techniques, but a novel approach that exploits the existence of common genetic variants has yielded new and surprising insights. Convincing evidence indicates a role of type 1 iodothyronine deiodinase (D1) in determining the serum T4:T3 ratio and a role of phosphodiesterase 8B in determining TSH levels. In addition, studies of type 2 iodothyronine deiodinase (D2) variants have shown that thyroid hormones contribute to osteoarthritis and these variants influence Intelligence quotient alterations associated with iodine deficiency. Preliminary evidence suggests associations between TSH-receptor variants and fasting glucose level, D1 variants and insulin-like growth factor I production, and D2 variants and hypertension, psychological well-being and response to T3 or T4 treatment. Intriguingly, most of these associations are independent of serum thyroid hormone levels, which highlights the importance of local regulation of thyroid hormones in tissues. Future research might reveal novel roles for thyroid hormones in obesity, cardiovascular disease, osteoporosis and depression and could have implications for interpretation of thyroid function tests and individualization of thyroid hormone replacement therapy.

Key points
Various processes influence the action of thyroid hormones in different tissues, such as local deiodination and transport of thyroid hormones across cell membranes, independently of serum thyroid hormone levels
Studies of commonly inherited variations have shown that the genes encoding type 1 iodothyronine deiodinase and phosphodiesterase 8B are important determinants of baseline serum thyroid hormone and TSH levels
Polymorphisms that affect the thyroid hormone pathway influence osteoarthritis and a highly suggestive association has been reported between these polymorphisms and the neurodevelopmental response to iodine deficiency
Polymorphisms might influence fasting glucose level, insulin-like growth factor I level, hypertension, psychological well-being and response to replacement therapy; many of these effects are independent of serum T3/T4 levels
Studies of common genetic variation in large, diverse cohorts are likely to provide important novel insights into the role of thyroid hormones in health and disease

Introduction
Thyroid hormones have a role in a wide range of physiological processes from growth and development to homeostasis in the adult. Variation between individuals in the levels of thyroid hormones in tissues might, therefore, be anticipated to have important effects on many biological systems. Several studies indicated that such variation between individuals does exist, both in the supply of thyroid hormones and in many local factors that determine intracellular thyroid hormone levels separately from serum levels.1, 2 In this Review, we will discuss how important these differences are and how they can be studied.

Local regulation of T3 and T4 action
The thyroid gland releases a combination of T4 and T3 in a ratio of approximately 17:1.3 Conversion of T4 to T3 is catalyzed by type 1 and type 2 iodothyronine deiodinases (D1 and D2, respectively), which influence the relative balance of these hormones in the circulation (reviewed by Bianco et al.4). To enter or exit cells, thyroid hormones must be transported across the cell membrane by thyroid hormone transporters,5 which have different tissue distributions and ligand affinities. The properties of three thyroid hormone transporters (which, unlike many other transporting molecules, have a high affinity and consequently a high specificity for thyroid hormones) are summarized in Table 1. The activity of these transporters in any given tissue is likely to be a key factor in determining the effect of serum T3 concentration on intracellular T3 levels. T4 is also actively transported into cells, but besides the activity of transporters, its contribution to intracellular T3 levels is also determined by local and temporal variation in the activity of intracellular D2. Notably, the level of circulating free T4 is typically five times higher than that of T3, thus the potential contribution of serum T4 to intracellular T3 levels is very large. Studies in rats suggested that in tissues with very active uptake and local interconversion of thyroid hormones, such as the cerebral cortex, around 80% of intracellular T3 is derived from serum T4, whereas in other tissues (for example, in the kidneys) this fraction is as low as 13%.6

TABLE 1 | Characteristics of the specific thyroid hormone transportersFrom the following article:
Full table
Figures and tables index
Download Power Point slide (91 KB) =
http://www.nature.com/nrendo/journal/v5 ... 19-pt1.ppt

Although in rats the ratio of T4 and T3 secreted from the thyroid gland and the dependence of serum T3 levels on the activity of D1 and D2 are different from those in humans, the above data suggest that various tissues are dependent on different amounts of local conversion of T4 to T3 in both species. Intracellular T3 levels are also dependent on the rate of inactivation by conversion to di-iodothyronine (T2) via the action of type 3 iodothyronine deiodinase (D3), and the rate of efflux from cells by transporters.7 From the cytoplasm, T3 enters the cell nucleus and binds to thyroid hormone receptors (THRs), and tissue-specific differences in the distribution and levels of the different isoforms of THRs, as well as of retinoic acid and other coregulator molecules, represent another level of local variation.8 Figure 1 summarizes the key steps between thyroid hormone release and its effects in individual tissues.

Figure 1 | The thyroid hormone pathway.
Abbreviations: D1, deiodinase 1; D2, deiodinase 2; D3, deiodinase 3; rT3, reverse T3; RXR, retinoid X receptor; T2, di-iodothyronine; THR, thyroid hormone receptor; TSHR, TSH receptor.
High resolution image and legend (32 KB)
Figures and tables index
Download Power Point slide (84 KB) = http://www.nature.com/nrendo/journal/v5 ... 19-pf1.ppt

This intricate system of local regulation allows cells and tissues to receive the appropriate amount of thyroid hormones, at an appropriate time in development, independent of serum thyroid hormone levels, and also provides a mechanism of autoregulation that protects tissues from variation in serum thyroid hormone levels. Note that the expression level of local regulating molecules is not static but can vary widely, depending on local thyroid hormone levels and other factors, such as the stage of development.

Implications of such regulation
From the endocrinologist's point of view, these new insights into local regulation of thyroid hormone action present several challenges. First, serum thyroid hormone levels cannot be assumed to indicate the thyroid hormone status of individual tissues. Second, if genetic differences between individuals determine the activity of elements of the thyroid hormone pathway (for example, transporters and deiodinases), the balance of serum T3 and T4 levels might require different interpretation from person to person. Finally, these findings imply that thyroid hormone replacement therapy might have to be individualized. In extreme conditions, such as severe hyperthyroidism or hypothyroidism, individual differences in local regulation become less important than the effects of very high or absent levels of serum thyroid hormones (although variation between individuals might have a role in the wide interindividual variation in the clinical presentation of these conditions). However, thyroid hormones have long-term effects on many processes—some of which endocrinologists were not previously aware (discussed below)—and in these processes, subtle interindividual variation in the thyroid hormone pathway over the lifetime of an individual is proving to be of increasing clinical importance.

Studies of common genetic variation
Common versus rare genetic variation
Recent advances in genetic research have increased our knowledge of the human genome and variation within it. In particular, our knowledge of single nucleotide polymorphisms (SNPs)—commonly inherited single-base changes that occur throughout the genome—has enabled us to obtain an improved map of the human genome and to study the genetic background of important biological parameters that have a polygenic basis, such as body weight or serum lipid levels.

Common genetic variation (mostly attributable to SNPs) usually causes minor alterations in function or activity, rather than mutations that lead to complete or near-complete inactivation of a single molecule in the thyroid hormone pathway. Major mutations (summarized in Table 2) are not only rare (most have not been observed in humans), but their effects are often both quantitatively and qualitatively different from those of common genetic variation, presumably because of the presence of compensatory pathways.

Table 2 Effects of single-gene inactivations related to the thyroid hormone pathway in knockout mouse models and humans
Full table
Figures and tables index
Download Power Point slide (128 KB) = http://www.nature.com/nrendo/journal/v5 ... 19-pt2.ppt

Common polymorphisms as investigative tools
Up to 65% of baseline thyroid hormone and TSH levels are genetically determined,10, 11 and in any individual limited variation in thyroid hormone levels occurs over time.12 The identification of common genetic variants that alter thyroid hormone levels provides us with valuable scientific tools in two ways. Firstly, such functional variants provide powerful evidence to identify molecules that have a role in determining serum thyroid hormone levels in humans. Secondly, once a particular polymorphism has been shown to alter the balance of thyroid hormones, it can be used in large cohort studies to explore the effects of changes in thyroid hormone levels on a wide range of biological systems.

Iodothyronine deiodinases
Polymorphisms in the DIO1 gene (which encodes D1) have been associated with changes in the balance of thyroid hormones in the serum, typically raised T3, low T4 and low reverse T3 levels, but, interestingly, these changes are not associated with any difference in TSH levels. This finding implies that the net effect of these changes in serum thyroid levels are perceived by the hypothalamus and the pituitary gland as 'neutral'.13, 14, 15, 16 These data have been reported at significance levels greater than those required for genome-wide association studies (P = 3.6  10-13 for free T3:free T4 ratio and P = 2.1  10-9 for free T4).15 The associations are similar in patients who receive levothyroxine treatment and those who have an intact hypothalamus–pituitary–thyroid axis. By contrast, studies of polymorphisms in DIO2 and DIO3 have revealed no effects on thyroid hormone levels except in one study of DIO2, the results of which have not yet been replicated (Table 3).13, 14, 15, 17, 18, 19 In the case of DIO2, convincing evidence shows that at least some of the common variants that have been studied are indeed functional in that they have effects on clinical phenotypes (see below). Table 3 summarizes the findings in this area that have been reported so far.

Table 3 | Association of deiodinase SNPs and serum thyroid hormone parameters
Full table
Figures and tables index
Download Power Point slide (88 KB) = http://www.nature.com/nrendo/journal/v5 ... 19-pt3.ppt

Phosphodiesterase 8B
Arnaud-Lopez and colleagues20 performed genome-wide association studies to identify polymorphisms that are associated with altered TSH levels; their findings were confirmed by a meta-analysis that included multiple populations. The investigators found a strong association between rs4704397, an SNP in the gene that encodes phosphodiesterase 8B, and circulating TSH levels (P = 1.9  10-20 for a meta-analysis): individuals who carried one or two copies of the rare allele had 0.13 mIU/l or 0.26 mIU/l higher TSH levels, respectively, than those with no copies of it (T4 and T3 levels were not assessed in participants in this study).

Phosphodiesterase 8B is a protein that catalyzes the hydrolysis and inactivation of cyclic AMP and is found in the thyroid gland but not in the pituitary gland. Arnaud-Lopez and co-workers20 suggest that phosphodiesterase 8B could affect TSH levels indirectly (via feedback), by affecting the generation of T4 and T3 in the thyroid gland. The SNP rs4704397 and the DI01 SNPs mentioned above provide robust, novel tools to study the effect of changes in serum thyroid hormone and TSH levels on different body systems.

Thyroid hormone transporters
Several studies have reported associations between variants of thyroid hormone transporters and levels of thyroid hormones and TSH. Polymorphisms in MCT8 are associated with changes in free T4 level in men,21 polymorphisms in MCT10 with free T3 and TSH levels,21 polymorphisms in OATP1C1 with free T4, T3 and reverse T3 levels,22 and polymorphisms in OATP1B1 with free T4 and reverse T3 levels.23 However, these associations have only been studied in small populations and have not yet been replicated.24, 25

TSH receptor
One SNP in the TSH receptor (Asp727Glu, rs1991517) was associated with serum TSH level but not thyroid hormone levels in three studies that included more than 2,600 individuals in total.13, 26, 27 This polymorphism has been proposed to lead to an increased sensitivity of the TSH receptor to TSH, which results in decreased serum TSH level whereas T4 and T3 levels do not change.

Thyroid hormone receptor
Only one polymorphism affecting THR has been studied, in THRB, and no relationship was observed between this polymorphism and serum thyroid hormone levels.13 This finding contrasts with the observation that a major mutation of this gene leads to resistance to thyroid hormone (Table 2).

Common polymorphisms and clinical phenotypes
Common polymorphisms that affect the thyroid hormone pathway could influence biological systems in two ways: either by changing serum thyroid hormone levels or by changing the intracellular availability of T3. The genes described in the previous section (DIO1 and the gene encoding phosphodiesterase 8B) provide tools to study the former effect, whereas other genes, notably DIO2, provide tools to study the effects of changes in intracellular thyroid hormone levels, independently of changes in their serum levels. The effects of common genetic variation on biological systems need to be studied in larger cohorts than those needed to study their effects on thyroid hormone levels, as the more downstream an effect is from the effector molecules (for example, thyroid hormones), the less accurately it can be measured. As the majority of the latest findings in this area were obtained from relatively small cohorts, they require replication; nonetheless, these studies have provided intriguing insights (Box 1).

Box 1 | Effects of SNPs on clinical phenotypes
Full box = http://www.nature.com/nrendo/journal/v5 ... 9_BX1.html

Bone and joints
The first example in which a robust effect of thyroid hormone-related polymorphisms has been confirmed is osteoarthritis. Thyroid hormones are vital for normal bone development and maintenance.28 Meulenbelt and colleagues29 performed a genome-wide linkage scan and found an association between the DIO2 polymorphism rs225014 and generalized osteoarthritis. Further linkage scans in three other cohorts that included more than 4,500 individuals in total revealed associations between a haplotype containing the rare allele of rs225014 and the common allele of rs12885300 (also in DIO2) and osteoarthritis (P = 2.02  10-5). The authors proposed that impaired function of this gene could lead to relative hypothyroidism in tissues and the development of osteoarthritis, as D2-catalyzed conversion of T4 to T3 in the growth plate is required for chondrocyte differentiation and bone matrix synthesis. The rs225014 SNP is a coding-sequence polymorphism, located at a site of DIO2 that is important for D2 turnover because it affects ubiquitin-mediated degradation of D2 within cells.30

Polymorphisms in the TSH-receptor gene affect serum TSH levels13, 26 and TSH levels are related to bone formation.31 Van der Deure and co-workers27 assessed 1,089 individuals and found that those who carried the allele associated with the Glu727 variant of the TSH receptor had 2.3% higher femoral neck BMD (P = 0.03) and 12.6% lower TSH level (P = 0.04) that those who were homozygotes for the wild-type allele (Asp727). The association with BMD persisted when TSH was added as a covariate. These findings suggest that TSH might contribute to the regulation of bone formation through the action of TSH receptors.

Brain development and neurocognitive function
The developing brain is particularly sensitive to local T3 levels and D2 is likely to be responsible for maintaining brain T3 levels in iodine deficiency. This hypothesis is supported by the work of Guo and colleagues,32 who used a case–control design (which is different from that of other association studies mentioned here) to compare the prevalence of DIO2 SNPs in children who had mental retardation with that in children with normal intelligence quotients in a iodine-deficient region of China. The investigators found significant differences between the two groups in the frequency of the haplotype combination of rs225010 and rs225012 (P = 0.0005). These results suggest that individual differences in the tolerance of low iodine levels during development might be explained by genetic variation in DIO2.

A study that used MRI to assess morphological features of early Alzheimer dementia did not find any association between DIO1 or DIO2 polymorphisms and the volume of the most affected areas in this condition, the hippocampus and the amygdala.14

Liver function and IGF-I level
Adults with hypothyroidism are reported to have low insulin-like growth factor I (IGF-I) levels, whereas hyperthyroid individuals have normal or high IGF-I levels. The level of IGF-I normalizes on attainment of euthyroidism.33 Peeters et al.34 found an association between a haplotype of two DIO1 SNPs and increased IGF-I levels in 156 blood donors (P = 0.02) and 350 elderly men (P = 0.01), which suggests a possible local effect of D1 activity in the liver. In elderly men, this haplotype was also associated with increased lean body mass (P = 0.03) and improved muscle strength (P = 0.047). These findings suggest that the increase in IGF-I level was clinically relevant.

Insulin resistance and diabetes mellitus
Previous studies have reported conflicting results on the association between insulin resistance and D2 polymorphisms, particularly rs225014. Three small studies showed associations between measures of insulin resistance and glucose utilization and rs225014 in both healthy individuals and people with diabetes mellitus.35, 36, 37 Furthermore, one of these studies demonstrated decreased activity of D2 in biopsy samples that were obtained from the thyroid gland and skeletal muscle in patients with diabetes mellitus who were homozygous for the rare allele (Ala92).36 Three large studies18, 38, 39 (total n >10,000), however, have demonstrated no association between rs225014 and any metabolic markers. Taken together, currently available data do not provide sufficient evidence of associations between this polymorphism and type 2 diabetes mellitus, BMI or insulin sensitivity.

In one study, which requires replication, a significant gene–gene interaction was observed between the rs225014 (Thr92Ala) polymorphism of DIO2 and the Pro12Ala polymorphism of the PPAR2 gene—a gene that is thought to have a role in determining insulin-resistant phenotypes. This interaction was associated with elevated systolic (P = 0.01) and elevated diastolic (P = 0.02) blood pressure and an increased risk of metabolic syndrome (P = 0.02) in 590 nondiabetic individuals. Among these patients, carriers of the 92Ala variant of rs225014 and the Ala12 variant of PPAR2 displayed the most severe symptoms.40

Peeters et al.37 have demonstrated associations between the 727Glu variant of the TSH receptor and levels of fasting glucose (P = 0.01), fasting insulin (P = 0.001), HbA1c (P = 0.002), leptin (P = 0.008) and homeostasis model assessment (HOMA) scores (P = 0.001). This study implies that in those who carry the 727Glu variant, the increased activity of the TSH receptor (which is known to be expressed in human adipose tissue) might cause increased adipogenesis, which subsequently increases leptin level and insulin resistance. However, as this study was performed in a population of elderly men, replication in large, diverse samples is required.

Hypertension
DIO2 is expressed in vascular smooth muscle cells and T3 acts as a vasodilator, hence relative tissue hypothyroidism could lead to hypertension. Gumieniak et al.19 found an increased frequency of the rare allele of the DIO2 polymorphism, rs225014, in 372 hypertensive individuals (odds ratio 2.11, P = 0.01). Another study, however, did not find any association between rs225014 and blood pressure,36 and a third one found an association only for rs225014 combined with the Pro12Ala polymorphism of PPAR2,40 hence the relation between this polymorphism and hypertension is yet to be clarified.

Thyroid hormone replacement therapy
In the brain, thyroid hormone levels are closely regulated by the action of thyroid hormone transporters and local deiodination.41, 42 Thyroid hormones must be first transported across the blood–brain barrier (by OATP1C1 and possibly by other transporters as well). T4 is then converted to T3 within astrocytes by D2, and T3 is transported into neurons by MCT8.7 Animal models have shown that this process can protect brain cells from changes in serum thyroid hormone levels very effectively.43

Individuals who receive thyroid hormone replacement therapy are unable to respond to low serum T3 levels by increased production of T3 in the thyroid gland. Insufficient compensation of low T3 level in some individuals might have detrimental effects especially in those tissues where close regulation of thyroid hormone levels is particularly important, such as the brain.6 These effects might explain the fact that patients who receive thyroid hormone replacement therapy have reported decreased well-being.44, 45 To investigate this phenomenon further, our research group has studied the effects of 16 tag SNPs (representative SNPs in a region of the genome with high linkage disequilibrium, which were chosen from the 'HapMap' haplotype database of the human genome) in the three deiodinase genes in 552 individuals on levothyroxine treatment. We found that SNPs in DIO2, particularly rs225014, were associated with impaired psychological well-being (P = 0.02).46 We found no association between SNPs in DIO1 or DIO3 and well-being, which is consistent with the fact that D2 is the only activating deiodinase that is present in the brain.6 The detrimental effect of DIO2 polymorphisms on well-being might result from a reduced ability to upregulate D2 function in the brain in response to low local T3 levels, which would be particularly apparent in individuals who receive T4 replacement alone, as these patients have a decreased T3:T4 serum ratio.47, 48 In a small study that included 141 patients on T4 replacement, Appelhof et al.49 observed a similar trend towards an association between rs225014 and impaired well-being, although it had low statistical significance (P = 0.11 to 0.13, with limited power due to sample size). Van der Deure et al.50 investigated the effect of polymorphisms in OATP1C1, which encodes a brain-specific thyroid hormone transporter, on well-being in the same 141 patients and showed an association between rs10770704 and increased fatigue and depression. Interestingly, the investigators found no association between this polymorphism and results in neurocognitive tests in these individuals, which suggests that thyroid hormones affect well-being and cognitive function via distinct mechanisms.

The effect of polymorphisms on the response to levothyroxine monotherapy or combination T4 and T3 therapy is still a controversial area, as clinical trials have not shown any benefit from combination therapy as compared with monotherapy.51 In 552 patients (270 on T4 and T3, 282 on T4 only) we found a significantly better response to T4 and T3 therapy than to T4 therapy in those with the rare allele of rs225014 (which was associated with the lowest baseline scores for well-being on T4 therapy) as measured by the general health questionnaire (GHQ-12; P = 0.03), thyroid symptom questionnaire (P = 0.03) and satisfaction scores (P = 0.02). Combination therapy was associated with a mean improvement in GHQ-12 scores of 2.3 points after 3 months of treatment.46 Consistent with these findings, another study has demonstrated that patients with thyroid cancer who carried rs225014 required a higher dose of levothyroxine to suppress TSH than patients with the same TSH levels who did not carry this polymorphism. This finding might reflect decreased D2 activity in the pituitary gland of patients with rs225014.52

In two relatively small studies, no association was found between either version of thyroid hormone replacement therapy and D2 or OATP1C1 polymorphisms.49, 50 All these studies retrospectively genotyped the participants, which means that these studies were not set up to investigate genotypic differences and their results have to be confirmed in prospective studies that are randomized by the genotype of participants.

Conclusions
Genetic association studies have increased our understanding of thyroid hormone regulation and action. The D1 deiodinase seems to have a key role in determining the balance of T3 and T4 in the serum and phosphodiesterase 8B has been demonstrated to determine TSH levels. A robust association has been observed between DIO2 polymorphisms and generalized osteoarthritis, and one study suggests an association of these polymorphisms with the ability of the brain to compensate for iodine deficiency. Furthermore, possible associations have been reported between TSH-receptor polymorphisms and osteoporosis and fasting glucose, DIO1 polymorphisms and IGF-1 production, and DIO2 polymorphisms and hypertension, psychological well-being and response to thyroid hormone treatment. Intriguingly, in the majority of these body systems, the effects of genetic variation were seen independently of serum thyroid hormone levels, highlighting the importance of local regulation of thyroid hormone in tissues. For the endocrinologist, genetic association studies provide new tools to examine the long-term effects of small changes in thyroid function on different tissues as well as to deepen our knowledge of thyroid hormone replacement therapy. Our current views of the importance of thyroid hormone and how to replace it seem likely to be challenged by surprising new findings in the near future.

Review criteria
PubMed was searched for articles published between 1970 and August 2008, using the search terms "thyroid", "genetic", "polymorphism" and "SNP".