Hormone Kinetics and Production

Deiodination at the 5′ or 5 position accounts for approximately 80% of the daily disposal of T₄, with the other processes shown in Fig. 4-1 responsible for the remaining metabolism of this compound.⁵ Such approximations may underestimate the role of both deiodinative and nondeiodinative pathways, however. Many of the products of these reactions, including T₃ and rT₃, are present primarily in the intracellular compartment and may undergo degradation before they have a chance to exchange with the plasma pool. It thus is difficult in human kinetic studies that use plasma sampling techniques alone to assess the contribution of these processes to overall thyroid hormone metabolism. For example, it has been demonstrated that most of the thyronine (T₀) excreted in the urine is in the form of its acetic acid analogue,¹⁷ thus suggesting that deamination plays a more prominent role in thyroid hormone metabolism than is apparent from the very low circulating levels of Tetrac and Triac. Such concerns have led to the concept that “hidden pools” of thyroid hormone metabolites may be present in tissues.¹⁸

Table 4-1^19–21 provides estimates of various kinetic parameters of thyroid hormone production and metabolism in humans.^5,22,23 The high affinity of T₄ for plasma binding proteins, along with its greater production rate, accounts for its relatively high concentration in serum, as well as its long serum half-life. In contrast, T₃ and rT₃ are present at much lower serum concentrations because of their lower production rates, greater metabolic clearance rates, and lower affinity for thyroxine-binding globulin (TBG). In addition, these two triiodothyronines appear to reside primarily within the intracellular compartment; thus their volumes of distribution are significantly greater than that of T₄.

The rate of T₄ production remains remarkably constant in healthy adults, with alterations noted only during pregnancy and in the aged. Although not well documented, increased T₄ secretion probably occurs during pregnancy in response to several factors, including thyroidal stimulation by human chorionic gonadotropin, an increase in the size of the extrathyroidal T₄ pool resulting from increased TBG levels, and an increase in the rate of T₄ and T₃ degradation because of high levels of 5-deiodinase activity in the pregnant uterus and placenta.24,25 One clinical consequence is the need to increase the replacement dose of T₄ by 25% to 50% in hypothyroid women during pregnancy.²⁶ Thyroid function appears to be well preserved through age 80 in healthy individuals, although a slight decrease in T₄ production and clearance is noted after the age of 60 years, and free T₃ levels are lower in centenarians.^27,28 That being the case, the decreased T₄ requirements of elderly hypothyroid patients probably result more from the presence of chronic illness and the use of concurrent medications than from the aging process per se.^29,30

Deiodination and The Iodothyronine Deiodinases

The potential importance of deiodination to thyroid hormone action was first recognized nearly 50 years ago when Gross and Pitt-Rivers demonstrated that although T₃ was considerably more potent than T₄,³¹ it was present in the thyroid in much lower amounts.³² This observation suggested that T₃ was derived largely from T₄ by metabolism in extrathyroidal tissues. This thesis was later proved by Braverman et al., who demonstrated the presence of T₃ in the serum of athyreotic subjects injected with T₄.³³ Research over the past three decades has confirmed the physiologic importance of the 5′- and 5-deiodination processes, has defined the biochemical parameters of these enzymatic reactions, has determined important regulatory factors that influence deiodinase activities, and has identified key structural determinants of the proteins that catalyze these reactions.^34,35

Three deiodinase isoforms, termed D1, D2, and D3, are present in vertebrate species. These differ in their catalytic properties, patterns of tissue expression, and mechanisms of regulation.34,36,37 The 5′- and 5-deiodination reactions catalyzed by these enzymes can be considered broadly as activating and inactivating processes, respectively.

Biochemical Characteristics

The biochemical properties of the deiodinases are outlined in Table 4-2.^38–45 They are in essence oxidoreductases in that they catalyze the substitution of hydrogen for iodine on the iodothyronine substrate. No other catalytic properties of these enzymes have yet been identified, nor do other known enzymes possess deiodinase activity. The enzymes are remarkable in terms of their substrate specificity and the precise location of the iodine removed.¹⁶ D1 is unique in that it can catalyze either 5′- or 5-deiodination, depending on the reacting substrate. Thus rT₃ is efficiently deiodinated only at the 5′ position by D1. In contrast, T₄ and T₃ are poor substrates for this enzyme unless they are first sulfated. This reaction markedly enhances their rate of 5-deiodination and further reduces their susceptibility to 5′-deiodination.⁴⁶ That D1 is relatively inefficient in converting T₄ to T₃ presents a considerable paradox, given that a significant proportion of T₃ production has traditionally been believed to occur in the liver and other D1-expressing tissues (vide infra).

D2 catalyzes only 5′-deiodination and very efficiently converts T₄ to T₃.¹⁶ The T₃ thus formed is a poor substrate for 5′-deiodination and is not further metabolized by this enzyme. In contrast, rT₃ is a good substrate for D2 as well as D1 and frequently is used in research assays to quantitate 5′-deiodinase activity. D3 exclusively catalyzes 5-deiodination¹⁶ and thus serves to convert T₄ and T₃ to rT₃ and 3,3′-T₂, respectively—metabolites with little affinity for the nuclear thyroid hormone receptors.

The deiodinases all require the availability of reduced thiol cofactors for efficient catalytic cycling.³⁹ These cofactors presumably function to displace iodine from an enzyme intermediate formed during the reaction and thus to regenerate the active deiodinase.⁴⁷ In in vitro assay systems using tissue homogenates or cellular subfractions, dithiothreitol typically is added as a cofactor. This small, nonnative four-carbon dithiol efficiently supports deiodination of all three enzymes. Kinetic data derived from broken cell preparations using dithiothreitol as cofactor have demonstrated a Michaelis-Menten constant (K_m) value for D1 of approximately 2.3 µmol/L (for T₄ 5′-deiodination), whereas D2 and D3 manifest much lower K_m values in the nanomolar range (see Table 4-2).³⁹ Based on this analysis, D1 sometimes is referred to as a “high K_m” enzyme, whereas D2 is said to catalyze a “low K_m” 5′-deiodination process. This distinction is somewhat spurious, however, because the kinetic properties of the deiodinases are clearly dependent on the thiol cofactors used in the assay system.⁴⁸ Thus, when the native thiol cofactor glutathione or thioredoxin is used, K_m values for D1 in the nanomolar range are obtained. Unfortunately, the cofactor system supporting deiodination in intact cells remains unknown,⁴⁹ and evidence suggests that glutathione and thioredoxin may not serve this role.⁵⁰ Thus the physiologic significance of the kinetic parameters derived in vitro remains uncertain.

Inhibitors

In addition to differences in their catalytic properties, the deiodinases show differential susceptibilities to certain inhibitors³⁹ (see Table 4-2). Most notable is the marked sensitivity of D1 to the antithyroid drug propylthiouracil (PTU). This agent forms an inactive complex with D1 by binding covalently to its active site. Notably, D2 and D3 show little or no susceptibility to inhibition by PTU. The related thioureylene drugs carbimazole and methimazole have no inhibitory effect on deiodination.⁴⁷

Gold compounds such as aurothioglucose are known to react with the active site selenocysteine residue in glutathione peroxidase and impair its activity.⁵¹ This compound also inhibits all three deiodinase isoforms, with D1 again showing a significantly greater sensitivity to this effect.^52–54 Other drugs that affect thyroid hormone metabolism include the iodinated radiographic contrast agents iopanoic acid (Telepaque) and sodium ipodate (Oragrafin). These small phenolic compounds act as substrate analogues and inhibit all three deiodinases in a competitive manner. To date, no selective inhibitors of D2 or D3 have been described—a situation that has significantly limited experimental investigations into their physiologic roles.

Tissue Patterns of Expression

An intriguing feature of the deiodinases is their pattern of tissue expression (see Table 4-2). The liver, kidney, thyroid gland, and pituitary gland express high levels of D1, with lesser amounts found in the brain.^16,55,56 In contrast, D2 is most abundant in the pituitary gland and brown adipose tissue (in rodents), with significant amounts also noted in the central nervous system.¹⁶ Studies using in situ hybridization and immunocytochemistry have demonstrated that D2 expression in the brain appears to be confined to certain subpopulations of astroglial cells, such as the tanycytes lining the third ventricle of the hypothalamus, which suggests that they serve as the principal site of T₃ production in this tissue.^57–59 In humans, D2 expression appears more widespread; the mRNA for this enzyme has been noted in the heart, skeletal muscle, and thyroid gland,^54,60 as well as in coronary artery smooth muscle cells.⁶¹

In adult mammals, D3 expression occurs primarily in the brain, with significant amounts also present in the skin.62,63 The brain is thus the only tissue that expresses all three deiodinases, although these are generally found in different subregions of this tissue and in different cell types. Thus, as opposed to the expression of D2 by glial cells, D3 is expressed primarily in neurons.^64,65

Finally, as detailed below, deiodinase expression patterns during pregnancy and development are of critical importance, with high levels of D3 activity present in the uterus, the placenta, and several fetal tissues.66,67

Structural Characteristics

Important structural features of the deiodinases have been deduced from the results of molecular cloning experiments.³⁷ All the enzymes have a molecular mass of approximately 29 to 32 kDa, function as homodimers,^43–45 and are selenoproteins in that they contain the uncommon amino acid selenocysteine as the reactive residue in the catalytic cleft (Fig. 4-2).⁶⁸ The importance of this amino acid to enzymatic activity is demonstrated by experiments in which selenocysteine has been replaced by cysteine. Such a substitution decreases the catalytic efficiency of the mutant protein to less than 1% of that of the native enzyme.⁶⁹ In the case of D3, the cysteine mutant also demonstrates an altered substrate preference.⁷⁰ It is remarkable that substitution of selenocysteine for a serine residue in a monoclonal antibody raised against T₄ confers on the protein deiodinase activity with catalytic properties similar to the D1, including sensitivity to PTU inhibition.⁷¹

The importance of selenocysteine to deiodination likely derives from its being ionized at physiologic pH, thus serving as a much more potent nucleophile than cysteine.⁷² Incorporation of selenocysteine into the deiodinases and other selenoproteins occurs at the time of translation and is directed by a specific stem-loop structure in the 3′-untranslated region of the mRNA termed a selenocysteine insertion sequence.⁷³ A unique tRNA (Sec-tRNA[Sec]), a specific RNA binding protein, and a specialized elongation factor also are required for efficient synthesis of selenoproteins.^74,75

In earlier studies, it was suggested that a 29 kDa (p29) nonselenoprotein, almost identical in sequence to a member of the Dickkopf protein family (Dkk3), was a substrate binding subunit of the D2.⁷⁶ However, Montero-Pedrazuela et al. have demonstrated a marked difference in spatial expression patterns in the brain of the p29 and the D2 selenodeiodinase,⁷⁷ and as recently reported, mice deficient in the Dkk3 protein do not have any consistent alterations in thyroid hormone status or D2 activity in their tissues.⁷⁸ These findings strongly suggest that the p29 protein is not relevant to D2 activity or to thyroid hormone homeostasis.

The overall amino acid identity of the three deiodinase isoforms is less than 30%. However, a high degree of homology is present in the regions of the selenocysteine and a conserved histidine residue (Fig. 4-3A). These areas show significant structural similarities to the thioredoxin-fold family of proteins and to α-l-iduronidase, respectively.⁷⁹ Based on these structural homologies, Callebaut et al.⁷⁹ and others have reported mutagenesis studies that define further the structure-function correlates of the deiodinase isoforms, including the identification of a key serine residue in the D1 active catalytic site that appears to confer sensitivity to PTU.⁸⁰

In addition, all deiodinases contain a hydrophobic region near the N-terminus,⁸¹ and subcellular fractionation studies have confirmed that these enzymes are integral membrane proteins.^82,83 Such studies, along with fluorescent microscopy of epitope-tagged enzymes, have demonstrated that the D2 is located in the endoplasmic reticulum,^82,84 whereas a significant fraction of the D3 has been reported to be in the plasma membrane.⁸⁵ The location of the D1 may vary in different tissues; in renal epithelial cells, the D1 has been localized to the basolateral plasma membranes,⁸⁶ whereas in the liver, it appears to reside in the endoplasmic reticulum (Pallud, Croteau, and St. Germain, unpublished observations). Studies by Friesema et al.⁸⁷ demonstrate that the catalytic effectiveness of all three deiodinases is markedly increased when specific thyroid hormone transporters are coexpressed with the deiodinases in cultured cells. This finding indicates that the active catalytic sites of these enzymes are located in the intracellular compartment.

Genetic Considerations and Knockout Mouse Models

The three deiodinase isoforms are coded in mammals by three different genes located in humans on chromosomes 1 (D1) and 14 (D2 and D3), as indicated in Table 4-2.^40–42 Although their genetic structure appears relatively simple in that they contain only four, three, and a single exon, respectively (Fig. 4-3B),^40,88,89 surprising complexities that may influence expression patterns have been identified.^90,91 For example, the DIO3 gene locus also expresses antisense transcripts from the DIO3os gene coded on the opposite DNA strand, and in the mouse fetus, the D3 is imprinted; transcription preferentially occurs from the paternal allele.⁹¹ In humans the DIO3 is also located within an imprinted region of chromosome 14.⁹¹ This pattern of organization and expression of the D3 often is observed for genes important in development.⁹²

To enhance our understanding of the physiologic roles of the deiodinases, genetic knockout models of each of the isoforms have been developed and partially characterized (Table 4-3).^93–100 D1-deficient mice show no general phenotypic abnormalities, although serum T₄ and rT₃ levels are modestly elevated, and the excretion patterns of iodothyronines in the feces and of iodine in the urine are significantly altered.⁹³ D2-deficient mice are viable and fertile in a protected laboratory setting. However, observed impairments in hearing,⁹⁶ thermogenesis,⁹⁵ and neurocognition⁹⁷ in these mutants would likely result in lethality in the wild. These animals also manifest elevated T₄ and thyroid-stimulating hormone (TSH) levels and show a marked selective resistance to the feedback effects of T₄, demonstrating the importance of D2 in pituitary thyrotroph regulation.⁹⁴ This contrasts with D1-deficient mice, in which TSH levels are normal.⁹³ This implies that the D1 expressed in the pituitary is not involved in TSH regulation.

A striking finding in the D1- and D2-deficent mouse models is that serum T₃ levels are normal.93,94 Indeed, in mice cross-bred to yield a combined D1 and D2 deficiency, the serum T₃ level is only minimally decreased (V. A. Galton, unpublished observations).¹⁰¹ This finding is both unexpected and difficult to explain, given that 5′-deiodination in peripheral tissues is supposed to contribute 80% of the T₃ production in humans and 45% in rats, with the remainder secreted directly from the thyroid.^34,102 Given that the combined D1/D2 knockout mouse has been unequivocally shown to lack all 5′-deiodinating capacity, it is clear that at least in the rodent, thyroidal production and secretion of T₃, perhaps combined with a decreased rate of T₃ clearance from D1 deficiency, can compensate and maintain serum T₃ levels that are essentially normal in the face of an inability to convert T₄ to T₃ in all tissues. A surprising corollary to this observation is that a D1 deficiency does not appear to significantly compromise T₃ secretion from the rodent thyroid, a tissue in which this enzyme is highly expressed and previously was thought to be essential for T₃ production.^102,103 This suggests that under physiologic conditions, D1 acts primarily as a “scavenger” enzyme that prevents the buildup of lesser iodothyronines (rT₃, T₂s, T₁) and sulfated derivatives in the thyroid gland, blood, and tissues.

The extent to which humans may be able to compensate for at least a partial impairment in 5′-deiodinating capacity has recently been reported. Although to date no mutations in the deiodinase genes in humans have been described, Dumitrescu et al.¹⁰⁴ have described two families with partially inactivating mutations in the selenocysteine insertion sequence-binding protein2 (SBP2), which is essential for the synthesis of selenoproteins, including the deiodinases.⁷⁴ Affected individuals demonstrate short stature and delayed bone age along with modest abnormalities in serum thyroid hormone levels, including increases in total T₄ and rT₃.¹⁰⁴ Notably, total serum T₃ levels were only minimally reduced, whereas TSH levels were at the upper part of the reference range and proved relatively resistant to suppression by administered T₄, but not T₃. These findings are very similar to those observed in knockout mice with a combined deficiency in D1 and D2, as noted previously. Indeed, a deficiency in D2 activity was demonstrated in cultured skin fibroblasts from affected patients.¹⁰⁴

Mice deficient in D3 present the most severe phenotype, likely as a result of the importance of this enzyme during the developmental period. Observed abnormalities include significant perinatal mortality, growth retardation, impaired fertility, and altered systemic thyroid hormone levels during the perinatal period (transient hyperthyroidism) and in adulthood (moderate hypothyroidism).99,100 Multiple abnormalities, including impaired response of the thyroid gland to TSH and of the pituitary to TRH, have been observed at all levels of the thyroid axis in the D3-deficient adult mouse. The response to hypothyroidism of TRH mRNA production in the hypothalamus and of pituitary TSH production is also markedly impaired. Many of these abnormalities are similar to those observed in human infants born to mothers with poorly controlled hyperthyroidism during pregnancy^105,106 or to rodents treated with large doses of thyroid hormone during the perinatal period.^107,108 These findings demonstrate that strict control of thyroid hormone levels during development by D3 is critical for proper development of the thyroid axis.

Regulation

The deiodinases are regulated by multiple hormones, growth factors, and environmental and nutritional factors.³⁴ Foremost among these are the thyroid hormones themselves; alterations in thyroid status induce profound changes in enzyme activity (see Table 4-2). Hypothyroidism is associated with a marked decrease in D1 and D3 levels, whereas D2 activity increases severalfold. Opposite changes occur in hyperthyroidism. These changes result from both pretranslational and posttranslational mechanisms.³⁴ For example, D1 and D3 mRNA levels are increased in the hyperthyroid state,^53,109 and D2 mRNA is decreased.⁵⁴ Hyperthyroidism also results in rapid downregulation of D2 activity by ubiquitination of the D2 protein.^110,111 This induces a reversible conformational change in the enzyme’s dimerization structure, which results in reversible loss of activity and eventually may target the enzyme for proteasomal degradation.¹¹² Considerable progress has been made in defining the molecular mechanisms involved in transcriptional control of the deiodinases; this topic has been reviewed recently by Gereben et al.³⁷ Of particular interest has been the observation by Simonides et al.¹¹³ that hypoxia induces D3 expression in several cell culture systems, as well as in ischemic myocardium.

Other important regulatory effects on deiodinase activity are noted in the thyroid gland, where TSH and thyroid-stimulating immunoglobulins stimulate both D1 and D2 activity⁶⁰; in brown adipose tissue, where cold exposure,¹¹⁴ bile acids,¹¹⁵ or intracerebroventricular administration of leptin¹¹⁶ markedly stimulates D2 activity; and in the liver, where nutritional deprivation, diabetes, tumor necrosis factor, and other cytokines decrease D1 activity (see also below).^109,117,118 In addition, D2 activity in the brain displays a significant diurnal variation.¹¹⁹ Although the physiologic significance of this remains to be determined, in some seasonal breeding species, rhythmic alterations in D2 and D3 activity in the central nervous system, perhaps mediated by melatonin,¹²⁰ appear to be integral responses to changes in the photoperiod.^121,122