Defining the Roles of the Iodothyronine Deiodinases: Current Concepts and Challenges

Abstract

As is typical of other hormone systems, the actions of the thyroid hormones (TH) differ from tissue to tissue depending upon a number of variables. In addition to varying expression levels of TH receptors and transporters, differing patterns of TH metabolism provide a critical mechanism whereby TH action can be individualized in cells depending on the needs of the organism. The iodothyronine deiodinases constitute a family of selenoenzymes that selectively remove iodide from thyroxine and its derivatives, thus activating or inactivating these hormones. Three deiodinases have been identified, and much has been learned regarding the differing structures, catalytic activities, and expression patterns of these proteins. Because of their differing properties, the deiodinases appear to serve varying functions that are important in regulating metabolic processes, TH action during development, and feedback control of the thyroid axis. This review will briefly assess these functional roles and others proposed for the deiodinases and examine some of the current challenges in expanding our knowledge of these important components of the thyroid homeostatic system.

This review assesses the functional roles of TH action and others proposed for the deiodinases and examines current challenges in expanding knowledge of these components of the thyroid homeostatic system.

The deiodination of T₄, T₃, and other iodothyronines is an integral component of thyroid hormone (TH) homeostasis. Catalyzed by three separate enzymes coded from three different genes, deiodination, depending on whether it occurs at the 5- or 5′-position on the iodothyronine molecule, serves to either activate or inactivate this class of compounds (Fig. 1). Thus, along with transport mechanisms that govern the flux of TH into and out of cells (1), the deiodinases function at a prereceptor level to influence both extracellular and intracellular TH levels and TH action.

Figure 1.

Reactions catalyzed by the iodothyronine deiodinases. T₄ is a relatively poor substrate for 5′-deiodination by the D1, whereas 5-deiodination by this enzyme most efficiently uses the sulfated derivatives (at the 4′-position) of T₄ and T₃ as substrates. 5D, 5-Deiodination; 5′D, 5′-deiodination.

Over four decades of research have brought us to the point where a great deal is known regarding the properties of these enzymes, referred to as the types 1, 2, and 3 deiodinases (D1, D2, D3) (2,3,4). In particular, studies using tissue homogenates and cell culture systems, as well as molecular cloning techniques, have defined to a considerable extent the structures, catalytic properties, expression patterns and regulatory mechanisms that pertain to these enzymes in a variety of mammalian and other vertebrate species. Key elements of this information for the human and rodent deiodinases are summarized in Table 1, including phenotypic features of deiodinase-deficient transgenic mice. Much of this information has recently been reviewed in detail elsewhere (2,3,4).

Table 1.

Characteristics of the iodothyronine deiodinases

Characteristic	D1	D2	D3
Reaction catalyzed	5 or 5′	5′	5
Substrate preference	5: T4S > T3S >> T3, T4 5′: rT3, rT3S > T2S >> T4	T4 > rT3	T3 > T4
Km (dithiothreitol as cofactor)	T4S (5): 0.3 μmol/liter	T4 (5′): 1 nmol/liter	T3 (5): 6 nmol/liter
	rT3 (5′): 0.06 μmol/liter		T4 (5): 37 nmol/liter
	T4 (5′): 2.3 μmol/liter
Molecular mass (kDa)	29	30	32
Selenocysteine	Present	Present	Present
Homodimer	Yes	Yes	Yes
Chromosomal location (human)	1p32-p33	14q24.3	14q32
Location	Liver, kidney, thyroid, pituitary	Pituitary, brain, BAT, thyroid,a heart,a skeletal musclea	Brain, skin, uterus, placenta, fetus
Activity in hypothyroidism	↓, Liver, kidney ↑, Thyroid	↑, All tissues	↓, Brain
Activity in hyperthyroidism	?, Liver, kidney ↑, Thyroid	↓, Most tissues ↑, Thyroida	↑, Brain
Inhibitors
Propylthiouracil	++++	+	+/−
Aurothioglucose	++++	++	++
Iopanoic acid	+++	++++	+++
Knockout phenotypes
Principal phenotypic features of deiodinase-deficient mouse	? Serum T4 and rT3 Normal serum T3 and TSH Enhanced fecal excretion of endogenous iodothyronines	↑Serum T4 and TSH Normal serum T3 and rT3 Pituitary resistance to T4 Impaired thermogenesis Impaired hearing Mild impairment of neurocognition	Increased perinatal mortality Growth retardation Impaired fertility in females and males Hyperthyroidism in the perinatal period Hypothyroidism in adulthood

^aHumans only. 

Of particular note are the catalytic actions of the D2, which is directed almost exclusively at the efficient conversion of T₄ to T₃ by 5′-deiodination (an activating reaction), and the action of the D3, which inactivates T₄ and T₃ by converting them to relatively inactive lesser iodothyronines (rT₃ and 3,3′-diiodothyronine, T₂) by 5-deiodination (5). In contrast, the D1 is able to catalyze both 5- and 5′-deiodination and can act on a variety of different iodothyronines, including those that have undergone sulfate conjugation of the hydroxyl group at the 4′-position on the outer ring (5,6).

Another notable feature of the deiodinases is that all are selenoenzymes; they contain the rare amino acid selenocysteine as a key residue at the active site (7). Substitution of cysteine for selenocysteine markedly impairs the catalytic efficiency of the deiodinases and substitution with any other amino acids renders the protein inactive (8,9,10). The presence of selenocysteine has implications beyond catalytic activity, in that the cellular processes for synthesizing selenoproteins are complex and inefficient (11) and may be impaired in some tissues by nutritional selenium deficiency (12,13,14).

The purpose of this brief review is to step back, for a moment, from what is a large body of experimental data and try to provide a perspective of the field in terms of what we have learned about the physiological roles of these enzymes and the challenges we face in trying to define further their function and bring this knowledge into the clinical arena.

What We Know: Concepts Well Established

The deiodinases play essential roles in coordinating TH action during vertebrate development

TH are critical signaling molecules that help to orchestrate the developmental programs of all vertebrate species (15,16). The deiodinases, particularly the D2 and D3, are expressed widely and in a dynamic, coordinated fashion during development, and it is now clear that they are important for regulating the circulating and tissue levels of TH during this period (17). As such, they play a critical role in developmental processes, and this has been demonstrated in several species.

Amphibian metamorphosis is a particularly dramatic example of the importance of TH in development and illustrates the key roles played by the deiodinases (18,19,20). In the absence of TH, metamorphosis of the tadpole into the mature frog does not occur (21). Furthermore, using iopanoic acid to block D2 activity in Rana catesbeiana tadpoles renders T₄, but not T₃, ineffective in inducing metamorphic changes, demonstrating the importance of this enzyme in tadpole development (22). Additional studies have clearly demonstrated that the diverse expression patterns of the D2 and D3 during tadpole development are exquisitely coordinated to regulate TH action at the local cellular level in a manner that ensures the correct timing of the metamorphic changes taking place in various tissues (20,23,24).

The D3 has also been demonstrated to be critically important for regulating the degree of exposure of the mammalian fetus and neonate to TH (25,26,27). A deficiency of TH during mammalian development, such as occurs with severe iodine deficiency, can lead to life-long and catastrophic neurological impairment and result in the syndrome of cretinism (28,29,30). Yet, almost from the moment of implantation until birth, the mammalian fetus is surrounded by tissues (i.e. the decidua and placenta) that express extraordinarily high levels of D3 and appear to be instrumental in maintaining circulating and tissue levels of fetal TH exceedingly low relative to maternal or adult levels (31,32,33). This has led to the concept that TH levels during development are tightly regulated within narrow limits to avoid either an excess or a deficiency of these compounds and that the D3, along with the D2, is largely responsible for this precise level of control.

That being the case, one might predict that a D3 deficiency would have profound consequences with regard to fertility and developmental processes. Characterization of a transgenic D3-deficient [D3 knockout (D3KO)] mouse demonstrates that this is the case (34,35). Both male and female D3KO mice have impaired fertility such that breeding for the purpose of maintaining a colony of transgenic animals is best accomplished by mating heterozygous parents. There is also significant perinatal mortality of the D3KO pups, and both pups and adults are significantly growth impaired. In addition, the developmental programming of the thyroid axis is markedly perturbed in the D3KO mouse, presumably by the overexposure of the animal to excessive levels of TH in utero and during the first 2 wk of perinatal life. Thus, an initial phase of hyperthyroidism gives way in the latter part of the perinatal period to a moderate state of hypothyroidism that persists into adulthood, where functional abnormalities in the hypothalamus, pituitary gland, and thyroid gland have all been demonstrated (36). Notably, these abnormalities of the thyroid axis resemble those observed in human children born to mothers with poorly controlled hyperthyroidism during pregnancy (37,38) and also rodents administered exceeding high doses of TH transiently in the immediate neonatal period (i.e. neo-T₄ syndrome) (39,40,41,42). Although much remains to be learned regarding the exact mechanisms leading to the phenotypic abnormalities in the D3KO mouse, the studies to date clearly demonstrate the proof of principle regarding the importance of the D3 during development.

The deiodinases regulate TH action within selected tissues during both development and in adulthood

The concept that the D2 and D3 provide a means of regulating local concentrations of active TH, thereby individualizing TH action within a subset of cells within a tissue, has been demonstrated by direct functional studies in diverse species and systems. Thus, in an example again taken from the metamorphosing tadpole, Marsh-Armstrong et al. (43) have demonstrated that the inactivating D3 is expressed in the dorsal ciliary marginal zone of the eye where it inactivates TH, thus inhibiting proliferation of these cells and allowing for the asymmetric growth of the retina that is characteristic of the frog. Direct confirmation of their hypothesis employed the use of the deiodinase inhibitor iopanoic acid and the expression of a D3 transgene in elegant studies that assessed the rates of cellular proliferation and anatomic changes in the tadpole retina.

Convincing evidence that the deiodinases play critical roles in modulating development in selective mammalian tissues has also been derived from the study of deiodinase-deficient mouse models. For example, Campos-Barros et al. (44) have demonstrated a striking transient spike in D2 activity in the cochlea of wild-type mice between postnatal d 2 (P2) and P12 of life, resulting in a T₃/T₄ ratio in this organ at d P8 of 1.76, compared with the corresponding serum hormone ratio of 0.02, a nearly 100-fold enrichment in the T₃ content of this organ relative to that of serum. This suggested that a D2-mediated increase in the local T₃ content might be critical for cochlear development and hearing. This hypothesis was confirmed when genetically modified mice completely devoid in D2 activity (D2KO mouse) were found to have markedly impaired hearing, a trait that could be rescued by the judiciously timed administration of high doses of T₃ via the nursing dam (45). More recently, it has been found that D3 is also expressed in the mouse cochlea starting before the spike in D2 expression, and indeed D3KO mice also have impaired hearing (46). Thus, either premature exposure of the developing cochlea to T₃ (due to an absence of D3) or failure to achieve the needed high local concentration of T₃ later in development (due to an absence of D2) results in uncoordinated development and impaired hearing. This demonstrates that, as in the tadpole, coordinated expression of the D2 and D3 is required for proper development of selected mammalian processes.

The importance of local regulation of TH levels by the deiodinases is not confined to the developmental period. The D2KO mouse in adulthood has elevated levels of both serum T₄ and TSH (with a normal serum T₃), suggesting that there is resistance to the feedback effects of T₄ at the level of the hypothalamus and/or anterior pituitary (47). This thesis has been confirmed by demonstrating impaired suppression of the TSH level by the administration of T₄, but not T₃, to the mutant mice (47). A similar pattern of serum TH levels and resistance to T₄ has recently been described in the only report to date of what is likely a D2 deficiency in humans, in this case resulting from a deficiency of the selenocysteine insertion sequence-binding protein 2, which results in impaired synthesis of D2 and other selenoproteins (48).

D2 has also been demonstrated to be essential as a local source of T₃ in brown adipose tissue (BAT) to facilitate adaptive thermogenesis in the face of cold exposure (49,50). Thus, despite a normal circulating T₃ level in the adult D2KO mouse, cold exposure results in hypothermia due to impaired BAT thermogenesis, and the animals survive only by compensatory shivering with consequent acute weight loss (51). The importance of the D2 in energetic pathways has recently been demonstrated further by the report that in mice, the induction of D2 in BAT by bile acids, working through a unique G protein-coupled receptor (TGR5), protects against insulin resistance and diet-induced obesity, a finding with potentially important implications with regard to the treatment of human obesity (52).

A remarkable feature of the deiodinases is that their activity levels are highly regulated by thyroid status via both pre- and posttranslational mechanisms (Table 1) (2,53,54). For the D2 and D3, the effects of hypo- and hyperthyroidism result in alterations in enzyme activities that appear designed to maintain local tissue T₃ content as normal as possible in the face of altered serum hormone levels. Thus, under conditions of iodine deficiency and in hypothyroidism, D2 activity is markedly up-regulated (55,56), and this has been demonstrated to increase the proportion of T₃ formed locally in D2-expressing tissues and to mitigate decreases in tissue T₃ content (57,58). D3 activity is decreased in hypothyroidism presumably to lessen degradation of T₄ and T₃, although this has not been directly demonstrated (56). Opposite changes in D2 and D3 activity occur in hyperthyroidism (56). In another apparent adaptive response, TSH stimulation of the thyroid gland in primary hypothyroidism results in increased D1 activity in this organ, which may serve to increase T₄ to T₃ conversion and hence T₃ secretion under these circumstances (59). Interestingly, in the setting of Graves’ disease, stimulation of D1 activity by thyroid-stimulating Ig may serve to exacerbate the hyperthyroid state (60,61).

Overexpression of deiodinases can cause human disease

Finally, the potential for overexpression of the D2 or D3 to cause human disease has clearly been demonstrated. Thus, a high level of expression of the D3 in certain vascular tumors can lead to a state of consumptive hypothyroidism wherein the metabolism and inactivation of T₄ and T₃ by this enzyme outstrip the ability of the thyroid axis to secrete TH and necessitate the administration of high doses of these hormones in order to maintain a euthyroid state (62,63). Alternatively, the overexpression of the D2 in rare cases of metastatic follicular thyroid carcinoma can render a patient thyrotoxic due to enhanced T₃ production in malignant cells while taking standard doses of T₄ for replacement or suppressive therapy (64,65). Increased expression of the D1 and/or D2 in the thyroid gland in functioning thyroid nodules, Graves’ disease (61,66), and McCune-Albright syndrome (67) also likely results in an enhanced rate of intrathyroidal T₄ to T₃ conversion and contributes to the increased secretion of T₃ observed in these conditions.

What We Think We Know: Concepts Surmised

The success in demonstrating conclusively the roles of the D2 and D3 in the above physiological and pathological situations results from the availability of appropriate model systems that allow the activity of the deiodinases to be selectively altered to assess directly the effects of such manipulations on clearly defined actions of TH (e.g. retinal proliferation in the tadpole, hearing in the mouse, and suppression of TSH). In contrast, many current concepts about the roles of these enzymes remain tentative or incomplete, in part because it has not been possible to bring such stringent experimental criteria to bear on these problems. Examples of these follow.

Conversion of T₄ to T₃ by the 5′-deiodinases in peripheral tissues and the thyroid gland is an important contributor to systemic T₃ production and serum T₃

A generally accepted concept is that the D1 and D2 contribute significantly to whole-body T₃ production and the serum T₃ level in humans and animals (68). Indeed, common experience indicates that the administration of T₄ to athyreotic humans results in the production of sufficient quantities of T₃ to maintain a clinical state of euthyroidism with normal serum T₃ levels. In vivo kinetic studies, as well as direct measurements of the rates of thyroidal T₄ and T₃ secretion, have led to estimates that in the euthyroid state, as much as 80% of the daily T₃ production in humans and 50% in rodents results from 5′-deiodination processes in extrathyroidal tissues (68,69,70). Both the D1 and the D2 appear to contribute to this T₃ production with estimates of the contribution by the D2 in humans and rodent ranging from 23–44% as determined by early reports using noncompartmental analysis of in vivo kinetic data (71,72,73). More recent analyses have made use of other techniques. Thus, using direct whole-body steady-state measurements of T₃ production, Nguyen et al. (74) have suggested that the D1 and D2 contribute equally to extrathyroidal T₃ production in rats. An even greater role for the D2 in humans has recently been suggested by Maia et al. (75). Based on extrapolating deiodination rates observed in a cell culture system to the intact organism, these investigators have suggested that approximately two thirds of the T₃ production in humans is derived from the D2 with higher and lower proportions obtained in hypothyroidism and hyperthyroidism, respectively, due to the changes in D1 and D2 activities observed under these conditions.

In addition to the importance of the D1 and D2 in converting T₄ to T₃ in peripheral tissues, these enzymes are also highly expressed in the thyroid gland (human D1 and D2 and rodent and dog D1 only) (76,77,78). There is evidence that T₄ to T₃ conversion occurs within this organ, as based on references cited above and more definitively on perfusion studies that demonstrate that thyroidal secretions in dog and human contain a higher ratio of T₃ to T₄ than that found in thyroglobulin molecules, and that this preferential secretion of T₃ can be inhibited by propylthiouracil, which impairs the activity of the D1 (79,80).

Taken together, this substantial body of evidence would lead one to predict that the inability to carry out 5′-deiodination in the thyroid and in extrathyroidal tissues would severely compromise T₃ production and result in a decreased serum T₃ level. Such is not the case, at least not in the mouse under euthyroid conditions. Thus, the D1KO and the D2KO mouse both manifest a normal serum T₃ level (47,81) as, surprisingly, does the combined D1/D2KO animal (82). In this animal, there is no compensatory decrease in D3 activity, nor is the clearance of T₃ significantly reduced. The serum TSH levels in the D2-deficient mouse models (D2KO and D1/D2KO) are approximately twice normal but dramatically less than would be encountered in a hypothyroid animal (47). The only source of T₃ in the D1/D2KO mouse is that derived directly from the hydrolysis of thyroglobulin, thus demonstrating that under selected circumstances, the thyroid gland itself, under the direction of the hypothalamus and pituitary, is able to supply sufficient T₃ to maintain the serum T₃ concentration as normal. This ability of the thyroid axis to compensate for the lack of 5′-deiodinase processes is a remarkable display of its adaptive ability. Indeed, within the warm confines of our vivarium, the D1/D2KO mouse maintains surprisingly normal thyroid homeostasis despite the complete inability to convert T₄ to T₃ in any tissues including the thyroid gland.

Whether such a significant degree of compensation would be observed in humans completely lacking 5′-deiodinase activity is uncertain, because to date, no inactivating mutations in the deiodinase genes in humans have been described. However, Dumitrescu et al. (48) have described two families with partial inactivating mutations in the selenocysteine insertion sequence-binding protein 2, which is essential for the synthesis of selenoproteins (83). Notably, the serum hormone profile in affected members of these families bear a striking resemblance to that of the D1/D2KO mouse: an elevated serum T₄ level, a serum TSH level at the upper range of normal or slightly elevated, and a serum T₃ level that is in the low normal range or minimally reduced. These findings suggest that the phenotype of the D1/D2KO animal has relevance to the human condition and suggests that there can be life without the ability to convert T₄ to T₃.

The deiodinases are important determinants of the alterations in systemic TH levels observed in illness and nutritional deficiency

Fasting and severe illness induce profound changes in TH economy marked by decreases in serum T₄ and T₃ levels without a compensatory rise in the serum TSH level in both humans and animal models (84,85,86). This generalized suppression of the thyroid axis is associated with a significant decline in basal metabolic rate and decreased protein and fat catabolism (87,88), and it has been postulated that this has adaptive benefit to the organism, although there is contradictory evidence (89,90). The pathophysiological mechanisms responsible for these systemic changes are complex and remain poorly understood (84,91). However, alterations in deiodinase activities have been postulated as playing important roles (Fig. 2).

Figure 2.

Proposed roles for the deiodinases in the systemic response of the thyroid axis to illness or nutritional deprivation. Recent studies have called these concepts into question (see text).

Observed decreases in hepatic D1 activity (resulting in a possible decrease in T₃ production) and increases in hepatic and skeletal muscle D3 activity (resulting in a possible increase in T₃ and T₄ degradation) in states of fasting and/or illness have been implicated as causative factors in the declines in serum T₃ and T₄ levels (84,91). In addition, increased D2 activity in the hypothalamus of ill or fasted rodents has been shown to result in an increased T₃ level in this tissue, and this is postulated to result in enhanced feedback suppression on TRH and TSH production despite lowered serum levels of T₃ and T₄ (92,93). [Note that D2 activity in human skeletal muscle, a potential modulator of systemic T₃ levels during acute illness, is unchanged in this condition (94).]

Such concepts, however, are based on indirect evidence whereby deiodinase activities have been correlated with serum and/or tissue TH levels, and it is possible and perhaps likely that some of the observed changes in deiodinase activity (e.g. the decrease in hepatic D1 activity) are a consequence of, rather than a cause of, the decrease in the serum T₃ level (95,96). Indeed, in two studies in which the in vivo rate of T₄ to T₃ conversion was measured directly in fasting animals by injecting [¹²⁵I]T₄ and quantifying [¹²⁵I]T₃ appearance in whole-animal homogenates, the fraction of T₄ converted to T₃ was approximately doubled and total T₃ production was unchanged (97,98). These latter reports call into question the interpretation of kinetic studies from human and animal experiments, which are based solely on data derived from analysis of serum and where the decreased plasma appearance of T₃ is assumed to be attributed to a decrease in tissue 5′-deiodinase (98,99). Thus, much of the T₃ produced during fasting or illness may remain in tissues and not exchanged with the plasma T₃ pool but, rather, metabolized via other degradative pathways or excreted via stool or urine. This concept is supported by recent studies in which the administration of supraphysiological amounts of T₄ and/or T₃ to rabbits with systemic illness was required to correct the serum levels of these hormones, suggesting that enhanced hormonal degradation and/or excretion, rather than diminished thyroidal secretion or decreased T₄ to T₃ conversion was playing a dominant role in the response to nonthyroidal illness (100).

Studies in deiodinase-deficient mice raise further questions regarding the roles of these enzymes in the systemic response to fasting and illness. Thus, D3KO mice subjected to a systemic bacterial infection develop marked reductions in their serum T₄ and T₃ levels comparable to those in infected wild-type animals, demonstrating that at least in this model, D3 is not required for development of a nonthyroidal illness-type syndrome (101). Furthermore, in preliminary studies, we have observed that fasting or the induction of systemic illness by the administration of lipopolysaccharide induces alterations in serum TH levels in D1/D2KO mice that are the same as those observed in fasted or ill wild-type animals, calling into question the importance of changes in the D1 and D2 as mediators of the systemic response to nutritional deprivation (Galton, V. A., and D. L. St. Germain, unpublished observations). Thus, the extent to which changes in deiodinase activities are necessary or responsible for systemic alterations in the thyroid axis in response to illness and nutritional deprivation remains uncertain.

Induction of D3 activity in response to tissue injury may influence healing or regenerative processes

Even if the deiodinases do not play a major role in mediating the systemic response of the thyroid axis to illness or altered nutrition, the changes in deiodinase expression observed in these circumstances could influence TH action locally in individual tissues or subsets of cells within an organ. In this regard, considerable interest has been generated by recent observations that D3 is induced in vivo at sites of tissue injury and inflammation and in cell culture systems by conditions of hypoxic or oxidative stress (102,103,104). Thus, D3 induction has been observed in animal models at sites of local inflammation (105,106), in peripheral nerves in response to cryolesion-induced injury (107), and in the heart in the setting of myocardial infarction (108) and cardiac hypertrophy (109). However, caution should be observed in assigning significance to these observations, because in some cases, the observed level of D3 activity have been exceedingly low (e.g. the reported D3 activity level of 0.1 fmol/min·mg protein in the right ventricle of a rat model of cardiac hypertrophy is approximately 1/60th of that observed in the neonatal mouse hypothalamus (35) and 1/5000th of that noted in the rat pregnant uterus shortly after blastocyst implantation (32) and could represent a nonspecific induction of gene expression with little or no functional consequences. Alternatively, there may be a significant level of D3 expressed in a subset of cells under these circumstances that has important effects on local TH action.

If this is the case, then presumably the induction of D3 leads to a localized region of hypothyroidism within the tissue in question, and this could influence healing or regenerative processes. For example, Li et al. (107) have suggested that the induction of D3 in peripheral nerve after injury may serve to reduce temporarily TH action locally and delay the myelination of regenerating fibers, given that the expression of myelin basic protein is a TH-dependent process (110). This elegant concept requires confirmation in an in vivo system in which the level of D3 activity can be manipulated at sites of nerve injury and the effects on regeneration assessed.

As an additional example, a low level of D3 activity (∼0.5 fmol/min·mg protein) is induced in the heart after myocardial infarction (108). Conceivably, D3 expression in this setting could serve to induce or accentuate a hypothyroid state within cardiac tissue after ischemic injury. This might be protective to cardiac cells in the acute stages of injury, without contributing to the adverse effects (e.g. increased peripheral vascular resistance) that accompany the systemic hypothyroid state. Such an outcome might tend to reinforce any long-term beneficial effects of the nonthyroidal illness syndrome. Alternatively, the D3 induction might have detrimental long-term consequences given that T₄, T₃, and T₄ analogs (e.g. diiodothyropropionic acid) have been shown to stimulate angiogenesis in the postinfarct rat heart (111,112,113,114). Such potentially paradoxical effects highlight the complexities in trying to predict the local functional consequences of deiodinase expression and point toward the need for carefully controlled studies to assess the significance of this D3 stress response.

The D2 is important for brain development and function

Neurological development in rodents, including mice, is known to be dependent on the provision of appropriate TH levels in the brain; severe abnormalities in learning, memory, and locomotion are observed in hypothyroid animals (115) or animals in which TH action has been disrupted (116). Much circumstantial evidence has led to the concept that the D2, and as a corollary the prohormone T₄, is critical for brain development and function (117). This concept is derived from studies in rodents where 1) it is estimated that more than 50% of the T₃ in the euthyroid adult rat brain is derived from local T₄ to T₃ conversion (118), 2) the D2 has been demonstrated to be the predominant pathway for extrathyroidal and intracerebral T₃ production in the euthyroid neonatal rat (58,119), and 3) it has been observed that only the provision of T₄, and not T₃, to a hypothyroid pregnant dam provides for restoration of the fetal brain T₃ content (120). Given that D2 is required both in utero and later in development for T₄ to T₃ conversion in the brain, the latter finding implies that normal brain development is dependent largely, if not exclusively, on the D2 to maintain brain TH action.

Observations made on the D2KO mouse call this concept into question; the neurological phenotype of this animal, which manifests a D2 deficiency from the time of conception, is surprisingly mild and manifests essentially none of the severe neurological abnormalities noted when rodents are hypothyroid during the developmental period (121). This is particularly surprising given that the T₃ content of the 15-d-old D2KO mouse is reduced by 50% (thereby verifying that the D2 is an important contributor to brain T₃ content), the same percentage as that noted in concurrently studied hypothyroid wild-type neonatal mice. Consistent with the mild neurobehavioral phenotype of the D2KO animal are molecular studies showing that alterations in expression of TH-dependent genes in several brain regions are significantly milder than in the hypothyroid neonates.

The reasons for these unexpectedly mild phenotypic features of the D2KO mouse are as yet uncertain but may indicate that T₃ obtained from serum, the level of which is normal in D2KO pups, plays a more important role than previously thought in brain development, or that T₄ in brain can compensate to a significant extent for the deficiency of T₃. (The T₄ levels in both serum and brain are approximately doubled in the D2KO pup.) Indeed, the lowering of serum TH levels in the D2KO mouse to render it hypothyroid results in a severe and lethal phenotype (Galton, V. A., unpublished observations), suggesting that the role of the D2 in brain development is particularly important in states of thyroid insufficiency. Alternatively, or in addition, we may not yet be aware of the appropriate subset of TH actions that would allow us to detect compromised neurobehavioral or neuromolecular development in the D2KO mouse.

The D1 serves principally as a scavenger enzyme

Whereas the D2 and D3 appear to have well defined functions, the physiological role of the D1 is difficult to define and to a considerable extent remains an enigma. Based on the prior concept that the majority of circulating T₃ was derived from T₄ via the catalytic activity of this enzyme, emphasis has been placed on the capability of the D1 for 5′-deiodination of T₄ to T₃. However, the human and rodent D1 is extremely inefficient in carrying out this reaction when compared either with the D2, which has a 700-fold greater catalytic efficiency for 5′-deiodination of T₄ (75), or with its own catalytic capabilities for 5′-deiodinating rT₃, rT₃S, or T₂S or 5-deiodinating T₄S or T₃S (6,122). Thus, the D1 catalyzes the deiodination of these latter substrates with efficiencies that range from approximately 20-fold (T₃S) to over 600-fold (rT₃) greater compared with the 5′-deodination of T₄. Although the large mass of D1 present in the liver does clearly allow for T₄ to T₃ conversion, this deiodinase seems much better positioned to clear from the circulation rT₃ and other inactive, sulfated, and lesser iodothyronines, compounds generally not degraded efficiently by the D2 or the D3 (122). Indeed, studies in the D1KO mouse have demonstrated that iodothyronines other than T₄ are the principal substrates for 5′-deiodination by this enzyme (81), which releases iodine back into the circulation for excretion in the urine or reclamation in the thyroid gland should the iodine supply be deficient. In the absence of the D1, inactive and lesser iodothyronines escape deiodination and are excreted in the feces with the potential loss of the associated iodine. Thus, this scavenging function of the D1 may take on a particularly important role in the setting of iodine deficiency.

This concept of the D1 as a scavenger deiodinase may also hold true regarding its expression in the thyroid gland, where in addition to converting T₄ to T₃, it may also deiodinate rT₃ and lesser iodothyronines released from thyroglobulin. In this role, it might assist the iodotyrosine dehalogenase (DEHAL1), which deiodinates monoiodotyrosine and diiodotyrosine, for recycling iodine within the thyroid gland (123). Evidence for this comes from studies comparing the iodothyronine content of thyroglobulin with the experimentally determined rates of secretion of T₄, T₃, and rT₃ from the human thyroid gland. Notably, the T₃/T₄ ratio in thyroid secretions is considerably higher than that found in thyroglobulin, whereas just the opposite finding is true regarding the rT₃/T₄ ratio (70,80,124). These observations suggest that there is a net formation of T₃ and a net degradation of rT₃ within the thyroid gland after thyroglobulin hydrolysis, presumably by the D1 (and/or D2 in humans).

Current Challenges and Future Opportunities

Much has been learned regarding the structure and function of the deiodinases over the past four decades, yet much remains to be done to achieve the goal of using our current and future knowledge for maintaining health and treating disease. An important aspect of future research will be to move beyond correlative studies, whereby speculation about deiodinase function is based on associated changes in serum or tissue TH levels or some readout of TH action. Although such associations are important for hypothesis generation, they rarely allow for the definitive assignment of functional roles. Among the major challenges in accomplishing this goal are the following.

Relative lack of knowledge regarding the effects of TH in tissues and cells

Given that the function of the deiodinases is to regulate the circulating and prereceptor concentrations of TH, our understanding of their roles in health and disease is dependent on a detailed knowledge of the effects of TH action in individual cells and tissues. Thus, a comprehensive understanding of TH action is needed to allow a less biased and more expansive approach to assessing functional outcomes of altered deiodinase expression. Unfortunately, our knowledge of both the physiological and molecular effects of TH action remains largely incomplete regarding most cells and tissues. For example, it is possible that the mild neurological phenotype of the D2KO mouse results, at least in part, from our ignorance with regard to appropriate molecular and neurobehavioral readouts to assess. Thus, a concerted effort is needed to define in a comprehensive manner the effects of TH at the molecular (i.e. the thyroidome or thyrome) and physiological level in major tissues and cell types to provide additional markers of TH action. This is of particular importance with regard to those cells and tissues where low levels of expression of deiodinases occur. In the case of tissues with heterogeneous cell types, techniques to isolate individual cell types will need to be employed.

Relative lack of knowledge regarding the trafficking of T₃ and T₄ within and between cells

The recent description of specific transporters that mediate TH entry and exit from certain cells has been a major advance in our understanding of the processes that control TH action. Yet, to understand the function of the deiodinases, we need to learn much more about how TH are trafficked within and between cells. For example, one explanation offered as to why the catalytic action of the D2 contributes to the local T₃ content of a tissue is that the enzyme is located in the endoplasmic reticulum, and thus T₃ derived from this enzyme is presumed to have ready access to the nucleus and the T₃ receptors (125). Yet this hypothesis fails to take into account the observations that in at least some tissues, in particular the central nervous system, T₃ derived from the D2 appears to act in cells other than those in which it is formed. One example of this is in the developing cochlea where D2 is localized in the connective tissue that gives rise to the bony labyrinth, whereas TRβ and TRα1 are mainly expressed in the sensory epithelium and spiral ganglion (44). Similarly, in the hypothalamus and other regions of the brain, the D2 is expressed primarily in glial cells and not in neurons, which are thought to be the principal T₃-responsive cells (126). Thus, a more complete understanding of cellular transport of TH into, within, and between cells is critical to our appreciating how deiodinases function in complex tissues such as the brain.

Current models for studying the effects of deiodination have significant limitations

Current models for defining the roles of deiodinases include amphibian and rodent models and various cell culture systems. Although extremely useful, these have certain shortcomings including the fact that the expression patterns of D2 and D3 appear to differ somewhat between rodents and humans under both basal conditions, where D2 expression is more widespread in the human (127), and during illness, where D3 activation in human liver and skeletal muscle appears to be more rigorous than in the rodent (128). In addition, the availability of pharmacological inhibitors of the deiodinases is limited. Although propylthiouracil is a relatively specific inhibitor of the D1 and iodinated contrast agents such as iopanoic acid inhibit all three deiodinases, both also have direct inhibitory effects on the function of the thyroid gland, which complicates their use in in vivo studies (129,130). To date, no inhibitors specific for the D2 and D3 have been described. The availability of these would provide useful tools for functional studies.

In place of selective pharmacological inhibitors, genetic models of deiodinase deficiency have been developed and proven useful in the hands of several investigators (34). However, complexities arise in the use of these animals in that the developmental abnormalities noted in the D2KO and D3KO mice, which have constitutive deiodinase deficiencies, persist into adulthood and complicate the assessment of the functional roles of these enzymes later in life. Thus, the development of conditional knockout animals that allow for the targeting of deiodinase deficiencies to specific organs or at specific times in the life cycle of the organism would be of significant benefit. An alternative approach in this regard would be the use of small interfering RNA techniques.

Only limited use has been made of transgenic overexpression models in the study of the deiodinases, but when employed, these have been very instructive. The few examples reported include transgenic tadpoles expressing the D3 (131) and mouse models where D2 expression has been targeted to the myocardium (132,133).

Finally, we need to be more creative in the experimental paradigms employed in defining the roles of these enzymes, because it is likely that environmental stresses will exacerbate phenotypic abnormalities in deiodinase-deficient animals relative to when they are housed under controlled conditions in a vivarium.

Use of the deiodinases to manipulate TH action in selected tissues

Ultimately, a better understanding of the importance of the deiodinases in controlling TH action during both development and in adulthood may lead to translational applications. With this knowledge may come treatments for neurological, psychiatric, metabolic, or cardiac disorders influenced by TH. Paradigms of such research are the discovery that selected bile acids can induce D2 activity, and hence thermogenesis, in selected tissues (52) and a recent report of D2 being targeted for overexpression in the heart (133). This latter maneuver altered gene expression patterns in this organ and negated the adverse effects of pressure overload on the left ventricle.

Thus, eventually we may learn to manipulate local deiodinase activity and TH action in clinical situations through the use of pharmacological agents that can target selective tissues and whose design is based on a more complete understanding of the mechanisms responsible for regulating deiodinase expression and catalytic function (134).