Minireview: The Neural Regulation of the Hypothalamic-Pituitary-Thyroid Axis

Abstract

Thyroid hormone (TH) signaling plays an important role in development and adult life. Many organisms may have evolved under selective pressure of exogenous TH, suggesting that thyroid hormone signaling is phylogenetically older than the systems that regulate their synthesis. Therefore, the negative feedback system by TH itself was probably the first mechanism of regulation of circulating TH levels. In humans and other vertebrates, it is well known that TH negatively regulates its own production through central actions that modulate the hypothalamic-pituitary-thyroid (HPT) axis. Indeed, primary hypothyroidism leads to the up-regulation of the genes encoding many key players in the HPT axis, such as TRH, type 2 deiodinase (dio2), pyroglutamyl peptidase II (PPII), TRH receptor 1 (TRHR1), and the TSH α- and β-subunits. However, in many physiological circumstances, the activity of the HPT axis is not always a function of circulating TH concentrations. Indeed, circadian changes in the HPT axis activity are not a consequence of oscillation in circulating TH levels. Similarly, during reduced food availability, several components of the HPT axis are down-regulated even in the presence of lower circulating TH levels, suggesting the presence of a regulatory pathway hierarchically higher than the feedback system. This minireview discusses the neural regulation of the HPT axis, focusing on both TH-dependent and -independent pathways and their potential integration.

Neuroendocrine signaling allows for integration of function of distinct tissues in complex organisms, leading to coordinated response to a given challenge and increased fitness for that organism. The hypothalamic-pituitary-thyroid (HPT) axis is a classical example of how a neuroendocrine system regulates distinct functions of an organism both during development (1, 2) and in adult life in response to a variety of challenges, presumably improving its chance of success. For instance, thyroid function and circulating thyroid hormones (TH) levels change in response to some of the most demanding conditions an adult organism may be exposed to, such as reduced food availability, decreased environmental temperature, and illness (3–11).

Interestingly, the presence of TH precedes the thyroid itself, and exogenous TH has major effects even on organisms that lack thyroid-like structures (12, 13). Indeed, it has been hypothesized that some invertebrates may obtain TH from diet (14), suggesting that TH signaling is phylogenetically older than the systems that regulate their synthesis in multicellular organisms. Thus, it is tempting to hypothesize that regulatory mechanisms that control TH synthesis evolved under the selective pressure of TH action. Indeed, it is well known that an excess of TH suppress, whereas the absence stimulates their own synthesis in a variety of organisms, including humans. Thus, it is plausible to assume that a negative feedback system was probably the first mechanism of regulation of TH levels. However, through evolution, new pathways emerged to control TH levels. In the present review, we will address some of these pathways related to the regulation of the HPT axis and discuss the actual relevance of the TH negative feedback to other conditions that regulate TH levels.

Feedback Regulation of the HPT Axis

Hypophysiotropic TRH neurons in the paraventricular nucleus of the hypothalamus (PVN) are believed to represent the regulatory core of the HPT axis (15–17). Early evidence suggested that the loss of those neurons severely impairs the regulation of TSH secretion (18), and it has been shown that TRH has a crucial role in determining TSH bioactivity (19–21). However, it is also known that the pituitary is able to increase TSH synthesis in response to severe hypothyroidism even in the absence of TRH (21, 22). Therefore, TH can affect the HPT axis at different levels, in addition to its actions at hypophysiotropic TRH neurons. Mature TRH is a tripeptide that is derived from pro-TRH by the action of prohormone convertases (23, 24) and, once released in the median eminence, reaches the pituitary and stimulates the synthesis and release of TSH (25–30). Remarkably, both transcription of TRH (31, 32) and its posttranslational processing (33) are suppressed by T₃.

Importantly, the deiodinases ultimately regulate intracellular T₃ availability (34), and many areas in the central nervous system are dependent on local T₃ production (35, 36). Because type 2 deiodinase (dio2) deficiency results in reduced hypothalamic T₃ content despite normal circulating levels (37), it is likely that T₃ acting in the hypothalamus derives, at least in part, from the local conversion of T₄. It has been proposed that this conversion occurs in tanycytes, which are ependymal cells of glial origin that line the floor and ventrolateral wall of the third ventricle and potentially act as selective two-way transporters between the bloodstream and the cerebrospinal fluid (38). These cells are seen in close association with axon terminals of the median eminence and express dio2. The importance of the role of dio2 in the regulation of hypophysiotrophic TRH neurons is underscored by the fact that its expression is inhibited by T₃, but whether those neurons rely mostly on locally generated T₃ is a matter of debate because dio2-knockout mice have normal hypothalamic TRH expression (39–41). In addition to dio2, the tanycytes also express the monocarboxylate transporter 8 (MCT8) a specific transporter of TH that allows for TH action in the nervous system (42–46). Strikingly, disruption of MCT8 expression in mice results in reduced T₃ concentrations in the brain and increased TRH expression (47–49). It is possible that the MCT8 also plays a direct role in T₃ uptake by hypophysiotrophic TRH neurons.

Although it has been proposed that tanycytes play an important role in the regulation of TRH neuron function, astrocytes located in the median eminence and the arcuate nucleus also express dio2 and are potentially an important source of T₃ to hypothalamic neurons (39, 40). In addition, cells in the pituitary also express dio2, and therefore, this enzyme may contribute directly to the regulation of TSH synthesis (50). Indeed, it has been demonstrated that pharmacological inhibition of dio2 increases TSH levels independent of changes in TRH expression (41, 51). Although dio2 generates T₃, neurons are also known to express the type 3 deiodinase (dio3), which inactivates both T₄ and T₃ (52). Within the central nervous system, dio3 expression is widely distributed, and its expression is stimulated by T₃, suggesting the presence of a local regulatory mechanism that works to avoid extreme variations in T₃ availability (53). However, it is not clear whether dio3 has a major role in determining T₃ action in the hypophysiotropic TRH neurons during adult life (54).

Once T₃ enters TRH neurons, it acts via TH receptors (TR) that act as ligand-dependent transcriptional factors. Two genes encode the known TRs, and due to alternative splicing or transcription initiation site, different isoforms are generated. TRα1 and -β1 distribution is widespread, whereas TRβ2 expression is restricted to some cells types including TRH neurons in the PVN (55) and thyrotrophs in the pituitary (56, 57), which suggest a role for this isoform on the feedback regulation of the HPT axis. Indeed, it is known that humans with mutation in the TRβ locus develop central resistance to TH, demonstrating the role of this isoform in humans (58). In addition, mouse genetic models corroborate that, showing that disruption of TRβ2 expression impairs the regulation of both TRH and TSH expression, resulting in a hyperactive HPT axis (59, 60).

Although TRs may have nongenomic actions (61), DNA binding is a requirement for the negative regulation of TRH and both TSH subunit genes (62). The genomic actions of TH also require the association of TRs with different coregulators. On the promoter of positively regulated genes, the presence of T₃ stimulates the recruitment of coactivators by the TR, whereas its absence favors the recruitment of nuclear receptor corepressors, leading to either increased or decreased transcription, respectively. However, it is still unclear how TRs control the expression of negatively regulated genes, such as TRH and the TSH subunits. In vivo models demonstrate that the coactivator steroid receptor coactivator 1 (SRC-1) is required for T₃-induced TSH suppression, because mice lacking this coactivator have increased circulating levels of both T₄ and TSH (63, 64). Similarly, mice that express a ΤRβ isoform that cannot recruit SRC-1 also have a similar phenotype (65). Both of these models support the hypothesis that coactivators have a paradoxical role in the transcription of negatively regulated genes. Conversely, the disruption of the interaction between TR and the nuclear receptor corepressor 1 (NCoR1), has the opposite effect (66, 67), suggesting that NCoR1 is required for the activation of HPT axis. Thus, the balance between coactivator and corepressor binding to TR can potentially change the set point of the HPT axis.

In addition to the control of TRH gene expression, T₃ indirectly regulates TSH expression and release by modulating the transcription of genes that will determine TRH availability and action at thyrotrophs. Tanycytes have been recently shown to express pyroglutamyl peptidase II (PPII), which degrades TRH released in the median eminence (68, 69) and furthermore, PPII mRNA expression is up-regulated by T₃ treatment, which presumably limits even further the TRH concentration in portal blood in hyperthyroid states (69, 70). To stimulate thyrotrophs, TRH binds to a specific cell membrane receptor, TRH receptor 1 (TRHR1), and its absence results in central hypothyroidism (71–73). Interestingly, T₃ also negatively regulates TRHR1 expression in thyrotrophs, reducing their sensitivity to TRH (74–76). Therefore, T₃ directly regulates TRH mRNA expression and directly controls its action by regulating both production and degradation of the tripeptide as well its downstream effects on its cognate receptor.

Endogenous Clock, Feeding, and the Neural Control of the HPT Axis

Although TH is a critical regulator of its own production via the regulation of TRH production and action, it is also clear that negative feedback in the TRH neuron can be overridden by distinct pathways that also control the HPT axis via TRH neurons. Both humans and rodents show a circadian variation in circulating TSH concentrations thought to be related to changes in TRH neuronal activity. This rhythm is characterized by peak secretion of TSH during the night in humans (77, 78), but that is not absolutely true for rodents, which also show high levels of TSH in the first hours of the light phase (79–82). Because TH are modulators of TSH synthesis and secretion, it would be reasonable to think that circadian variation results from a change in circulating TH levels. However, this does not seem to be the case. Recently, it has been shown that TSH determines the circadian variation in free T₃ levels in healthy individuals (83) and that circadian TSH rhythm is preserved in primary hypothyroidism (84). Although the physiological relevance of this circadian oscillation remains unclear, it supports the notion that TRH and TSH levels are not a function of circulating TH concentration exclusively. Because nuclear receptors mediate entrainment of peripheral clocks and thyroidectomy disrupts the diurnal expression pattern of some clock genes, it is possible that the HPT axis contributes to synchronize the circadian rhythm present in different tissues (85).

It is well known that local TH concentration may vary among different tissues, at least in part due to the action of deiodinases (34). Therefore, changes in deiodinase activity in the hypothalamus or the pituitary may account for the circadian oscillation in circulating TSH levels. In fact, this hypothesis was raised previously (86), but the actual role of the deiodinases on the control of HPT circadian rhythm remains unclear. Interestingly, the circadian variation in the HPT axis can be shifted by inverting the photoperiod cycle in rats (87, 88), and dio2 activity can be suppressed by light exposure in rodents (89). However, the increase in hypothalamic dio2 activity during the dark phase does not increase tissue T₃ content and is accompanied by increasing TSH circulating levels (90). Thus, changes in dio2 activity cannot explain the circadian rhythm of the HPT axis, and it is possible that other factors such as TH transport and catabolism play an important role. Indeed, MCT8 and dio3 are photoperiodic-regulated genes in the hypothalamus of seasonal mammals (90), but it remains to be shown whether this has any relevance in the circadian regulation of the HPT axis.

Early studies have shown that the anterior hypothalamus contributes to the circadian variation of circulating TSH (91), suggesting that the suprachiasmatic nucleus of the hypothalamus (SCN) may have a relevant role on the control of HPT axis circadian oscillation. In fact, SCN neurons project to parvocellular neurons in the PVN (92), and SCN ablation affects the circadian variations in TSH and TH levels (80, 93). Because TH does not regulate circadian clock genes expression in the SCN (94), the circadian rhythm of the HPT axis may be under the control of a neural pathway unrelated to the negative feedback mechanism exerted by TH. Nevertheless, we cannot rule out the possibility of nongenomic actions of TH (95) controlling the excitability of neurons composing the hypothalamic circuitry that determines the rhythmicity of TRH release.

Remarkably, destruction of the SCN also affects rhythmic fluctuation in circulating corticosterone, and it is known that glucocorticoids have inhibitory action on TRH expression (96, 97). Thus, it is possible that the HPT axis rhythm is also determined by circulating glucocorticoids. Indeed, exogenous glucocorticoid blunts circadian secretion of TSH (98). Still, adrenalectomized rats retain circadian TSH secretion, as well as patients with Addison's disease infused with glucocorticoid (99, 100). Therefore, although glucocorticoids are able to suppress the HPT axis, it is unlikely that its circadian rhythm is under control of these steroid hormones.

In addition to light, food availability also enables variations in endocrine organ function (101). It has long been known that fasting suppresses thyroid function and reduces circulating levels of TH in both humans and rodents presumably as an energy-sparing adaptive response (3, 4, 102, 103). It is noteworthy that TRH and TSH levels are also reduced by fasting (104, 105), suggesting the existence of a regulatory pathway hierarchically higher than the negative feedback system. Although it is possible that increased hypothalamic dio2 activity accounts for the fasting-induced suppression of TRH expression (106), it remains to be demonstrated that T₃ action in the hypophysiotropic TRH neurons is actually increased during fasting. Also, it is unlikely that increased T₃ action at the pituitary level increases during fasting (107), but rather, the thyrotrophs' sensitivity to TRH is reduced (108).

Evidence of neural circuitry controlling HPT axis activity during energy deficiency emerged more than a decade ago, when a critical role for both leptin and the arcuate nucleus of the hypothalamus (ARC) for the fasting-induced suppression of HPT axis was demonstrated (109, 110). The ARC is an important leptin target because leptin is able to influence the expression of different ARC neuropeptides (111). Importantly, ARC neurons project to the PVN, and leptin administration stimulates the HPT axis (81, 112–114), suggesting that decreased leptin levels during fasting suppress the thyroid axis (109, 115).

Among ARC neurons, those that express both neuropeptide Y (NPY) and agouti-related peptide (AgRP) represent a major subpopulation of neurons that project to the PVN, and most of them express the leptin receptor (112, 116, 117). Interestingly, leptin inhibits these neurons, and both AgRP and NPY expression is increased during fasting (111, 118, 119). Importantly, central administration of both AgRP and NPY inhibits the HPT axis (120–122), and thus it is likely that one of these neuropeptides mediates leptin's effect on TRH neurons and contributes to suppressing TRH expression during starvation. Indeed, we have recently demonstrated that NPY signaling is absolutely required for the fasting-induced suppression of the HPT axis (123, 124). In addition to this indirect pathway via the ARC, it is also likely that leptin exerts a direct stimulatory action on the hypophysiotropic TRH neurons (122, 125, 126).

Although the hypothalamic-pituitary axis plays a crucial role in the response to fasting, circulating TH levels are also regulated by peripheral catabolism. It has been long demonstrated that peripheral metabolism of T₄ is increased in fasted rodents (127), but ex vivo models showed that hepatic 5′-deiodination does not increase in fasted states (128, 129). Therefore, other pathways may be involved in fasting-induced TH degradation. We have shown that both central NPY and melanocortin signaling control the hepatic expression of sulfotransferases and glucuronidating enzymes during fasting, leading to increased metabolism and excretion of TH and contributing to reduce TH action during negative energy balance states (124). Therefore, hypothalamic signaling pathways control the HPT axis both centrally through the TRH neuron and peripherally in catabolic states, supposedly increasing the chances of survival.

Final Remarks

Regulation of the HPT axis is complex, and every year new advances in the area are made. However, we are far from fully understanding its control. Undoubtedly, the negative feedback imposed by TH plays a role in the regulation of the HPT axis, but there are clearly other key pathways that are working to keep TH levels adequate. Indeed, under physiological conditions, feedback regulation seems to play a less relevant role when compared with conditions where primary dysfunction of the thyroid gland is present. It is true that in some situations (e.g. starvation), changes in central action of TH might cause a shift in the set point of the HPT axis. However, the signaling pathways driving these putative set-point-modifying phenomena need to be elucidated. For instance, it is known that the coregulators SRC-1 and NCoR1 control the action of TH also on negatively regulated genes and that changes in their expression/action shift the set point of the HPT axis. However, it remains to be demonstrated how this is orchestrated in physiological conditions and what would be driving these modifications.

Neural circuitries regulate thyroid activity through the control of TRH release in the median eminence, and this seems to be especially relevant in the control of circadian rhythm and in response to both fasting and reduced environmental temperature. Interestingly, during those situations, changes in circulating TH levels do not elicit a counterregulatory response of the hypothalamic-pituitary axis. Therefore, it is tempting to assume the existence of regulatory mechanisms able to override negative feedback regulation. Strikingly, some of these pathways may be controlling distinct responses to a common stressor, such as during restricted food availability. In that situation, NPY signaling plays a crucial role in the control of both food intake and HPT axis activity, suggesting that these pathways may have evolved together as a common energy-replenishing response. Taken together, this suggests that the regulation of the HPT axis occurs at multiple levels and is highly integrated with the internal milieu and the external environment.