kathi

Perspectives in the Study of Thyroid Hormone Action on Brain Development
and Function
Posted 12/12/2003
Juan Bernal; Ana Guadaño-Ferraz; Beatriz Morte
Abstract and Introduction
Abstract

The purpose of this review is to provide an up-to-date report on the
molecular and physiologic processes involved in the role of thyroid
hormone as an epigenetic factor in brain maturation. We summarize the
available data on the control of brain gene expression by thyroid
hormone, the correlation between gene expression and physiologic
effects, and the likely mechanisms of action of thyroid hormone on brain
gene expression. In addition we propose a role for unliganded thyroid
hormone receptors in the pathogenesis of hypothyroidism. Finally, we
review recent data indicating that thyroid hormone receptors have an
impact on behavior.
Introduction

Brain development proceeds through precisely coordinated events in time
and space. Induction of neural tissue, differentiation and migration of
specific cell types, regional specialization or synapse formation are
major milestones in a myriad of intricately associated events. Most of
these events are largely determined by genetic factors.[1] However,
epigenetic factors, such as the thyroid hormones, are also important
because they act to control the timing and coordination of
mechanistically unrelated processes. For many years thyroid hormone has
been considered essential for brain maturation in humans and other
mammals. Deficiency of thyroid hormones during critical periods of
development leads to profound and potentially irreversible defects of
brain maturation,[2,3] and clinical syndromes arising from thyroid
hormone deficiency during fetal and postnatal periods have been
extensively reviewed elsewhere.[3,4]

The Brain as a Target of Thyroid Hormone

Most of the available evidence supports the view that thyroid hormone
does not influence the major early developmental processes such as
neural induction, neurulation, and establishment of polarity and
segmentation. Thyroid hormone is involved in regulation of later events,
such as cell migration and the formation of cortical layers, and in
neuronal and glial cell differentiation. Thus, as a consequence of
thyroid hormone deficiency during specific fetal stages, altered cell
migration in the neocortex results in less defined cortical layering and
altered distribution of callosal connections.[5] Deficiencies of cell
migration are also observed in the hippocampus, resulting in lower
number of granule cells in the dentate gyrus. A defining feature of
neonatal hypothyroidism is the delayed migration of granule cells in the
cerebellum, which results in abnormal persistence of the external
germinal layer.

Thyroid hormone also controls differentiation of not only neurons and
oligodendrocytes,[6] but also astrocytes, and microglia.[7,8] Specific
neuronal cell types are characteristically affected by hypothyroidism,
such as pyramidal cells of the neocortex and Purkinje cells of the
cerebellum. The latter show a strong dependence of thyroid hormone,
because in its absence they do not fully develop their characteristic
highly elaborated dendritic tree. On the other hand, lack of proper
differentiation of oligodendrocytes results in myelination deficits.
Deiodinases and the Control of Triiodothyronine Concentrations in the
Brain

The main active thyroid hormone is triiodothyronine (T3), which in the
brain derives in large part from 5´ deiodination of thyroxine (T4). This
pathway is closely regulated by developmental and physiologic
factors.[9] Concentrations of T3 in the brain are the result of transfer
to the brain through the blood-brain barrier, by a poorly defined
mechanism, and the local activities of deiodinases 2 (D2) and 3 (D3). D2
generates T3 from T4 by phenolic ring deiodination, whereas D3
inactivates T4 and T3 by tyrosil ring deiodination with generation of
the inactive products reverse triiodothyronine (rT3) and 3,3´
triiodothyronine (T2), respectively. Expression of D2 and D3 in the\0
brain is developmentally regulated. In the rat, D2 activity is low
during the fetal period, and increases postnatally, whereas D3 follows
the opposite pattern.[10] D3 activity is high in placenta and in fetal
tissues and high activity has been detected in the uterine implantation
site.[11] This pattern of expression suggests a protective role during
development. In agreement with this, recent data have shown that
inactivation of the D3 gene leads to partial lethality at birth and
severe growth impairment.[12] It may also be that D3 expression and the
regulation of T3 generation and its subsequent availability to target
cells is a way to control the timing of thyroid hormone signaling in a
fashion similar to the control of amphibian metamorphosis.[13] In adult
animals D3 is diffusely expressed in neurons throughout the brain.[14]
In contrast, during the late fetal and early postnatal period, D3 mRNA
is highly concentrated in discrete nuclei such as the bed nucleus of
stria terminalis, central amygdala, and the preoptic area.[15] These
nuclei are involved in ***ual differentiation of the brain during the
early postnatal period, but the functional significance for the
restricted expression of D3 in these nuclei is not known.

D2 activity increases during the postnatal period and is sensitive to
thyroid hormone concentrations. In hypothyroidism, its increased
activity tends to normalize T3 concentrations even with greatly reduced
T4 concentrations. In contrast with D3, which is expressed in neurons,
D2 is expressed in two types of glial cells.[16,17] The highest
expression is found in the tanycytes, a specialized type of glial cells
that line the lower third of the walls of the third ventricle. These
cells send processes to the adjacent hypothalamus where they frequently
end in blood vessels, and to the median eminence, ending in portal
vessels. Thus, D2 in these cells could be involved in providing T3 to
the cerebrospinal fluid (CSF) from which it would reach nearby
structures by diffusion, and/or the portal blood, influencing pituitary
function. Stimulation of D2 in the tanycytes by cytokines with
subsequent local production of T3 and inhibition of thyrotropin (TSH)
secretion has been recently suggested to play a role in non-thyroidal
illness.[18] D2 is also expressed throughout the brain in astrocytes and
in some interneurons.[19] Astrocytes appear therefore to have an active
role not only in the uptake of T4 from the blood through the blood-brain
barrier, but also in generating T3 and delivering it to nearby neurons.
Molecular Basis of Thyroid Hormone Action on Brain Development

Many of the effects of thyroid hormone on developmental processes in the
brain can be correlated with the controlled expression of specific
molecules. The most obvious is myelination, which is secondary to the
effects of thyroid hormone on oligodendrocyte differentiation and the
expression of oligodendrocyte specific genes.[20] Lack of thyroid
hormone during the postnatal period in rats, at the time of onset of
myelination, strongly delays the expression of many oligodendrocyte
genes, specifically the genes encoding proteins of myelin such as myelin
basic protein (MBP), proteolipid protein (PLP), and myelin-associated
glycoprotein (MAG).[21] As a result, the number of myelinated axons in
hypothyroid rats is strongly reduced.[22]

How thyroid hormone influences differentiation of neural cells is poorly
understood. It has been suggested that T3 is an instructive factor in
the early steps of oligodendrocyte generation from stem cells, and that
it controls the timing of oligodendrocyte precursor cell
differentiation.[23,24] It has also recently been shown that T3 promotes
neuronal differentiation of embryonic stem cells in culture.[25] The
intimate molecular mechanisms by which thyroid hormone promotes
differentiation is unknown. Molecular suspects include proteins
regulating the cell cycle, such as E2F-1 (see Billon et al.[24]), p53,
cyclins and cyclin-dependent kinase inhibitors.[26] Thyroid hormone
control of neurotrophin expression, especially NGF, BDN, and NT-3, may
provide also an explanation for its effects on differentiation of
particular cells, such as cholinergic neurons and Purkinje cells.[27] In
many cases the effects of thyroid hormone on differentiation are subtle
but with potentially important functional consequences. For example, in
hypothyroidism there is a decreased number of dendritic spines in
pyramidal cells of the neocortex and hippocampus, which is reversible
and also observed after adult-onset hypothyroidism.[28] Although the
gross morphology of the neurons is not appreciably modified, changes in
spine density have strong consequences on synaptic plasticity. The
mechanisms by which dendritic spines are affected by thyroid hormone are
unknown. RC3/neurogranin,[29] a dendritic spine protein is under thyroid
hormone control, both in developing and in adult animals,[30] but it is
not known whether expression of this protein is related to spine
formation.

Proper migration of neurons depends on the interaction of cell surface
receptors with proteins present in the extracellular matrix. One of
these proteins, reelin,[31] is regulated by thyroid hormone during the
late prenatal and early post-natal life in the rat.[32] Reelin
expression is essential for the orderly migration of neurons to their
specific destinations in the cerebral and cerebellar cortices, and
determines the normal pattern of cortical layers. This protein is
produced by Cajal-Retzius cells, which express another thyroid
hormone-regulated gene, prostaglandin D2 synthase.[33] Other molecules
reported to be involved in cell migration, such as laminin, tenascin C,
and L1, are also under thyroid hormone control.[34,35]

Most of the actions of thyroid hormone are exerted via nuclear
receptors, which are ligand-modulated transcription factors. The
physiologic ligand is T3, and therefore it is considered to be the
active form of thyroid hormone. Because T4 has low affinity for the
nuclear receptors, it might be considered as a prohormone, whose role
would be to deliver T3 intracellularly through 5´ deodination. There
are, however, data suggesting that T4 may have actions of its own. It
regulates D2 activity in astrocytes and has a direct action on F-actin
polymerization.[36,37] Also in astrocytes, T4 has been shown to
influence integrin-laminin interactions,[38] an important process in
cell migration.

Whether these nongenomic actions of T4 actually contribute in vivo to
the effects of thyroid hormone on neural cell migration is however
unknown. Work by Davis et al.[39] also suggests that T4 has biologic
activity per se. In this case, the action of T4 is mediated via the
microtubule-associated protein (MAP) kinase pathway and phosphorylation
of the T3 receptor and other targets.[39] So far it is not known how
this activity of T4 may contribute to the overall physiologic activities
of the thyroid hormones. With respect to the nongenomic actions, mice
deficient in all forms of nuclear T3 receptors have no signs of
hyperthyroidism in the face of highly elevated circulating T4 and T3,
suggesting that most effects of thyroid hormone are mediated through the
nuclear receptors.[40] Therefore, the contribution of nongenomic actions
to the overall effects of thyroid hormone is not yet clear. Davis and
Davis[41] have suggested that at least for the heart, nongenomic actions
of thyroid hormone determine the basal activity of transporters and ion
channels. If this concept is extrapolated to the brain, the nongenomic
actions of thyroid hormone could be important in neurotransmission.

Another long-standing controversial topic is the role of the
mitochondria in the overall effects of thyroid hormone in brain and
other tissues.[4,42] It is clear that thyroid hormone influences
mitochondrial function indirectly through the control of nuclear-encoded
mitochondrial genes. In addition, the expression of
mitochondrial-encoded genes are also influenced by the thyroidal status,
and T3 has direct actions on isolated mitochondria in vitro.[43]
Therefore, direct effects of T3 on the mitochondria in vivo appear
likely. Truncated forms of TR?1 and RXR? have been described in the
mitochondria,[44] and it is proposed that T3 may affect mitochondrial
transcription in a way similar to the action on the nucleus. These
topics have been recently reviewed in detail.[45] Patterns and
Mechanisms of Gene Regulation by Thyroid Hormone in the Brain

During the past 10 years, we and others have identified a number of
genes under thyroid hormone control in the brain (for a review, see
Bernal[4]). This list of genes is likely to be greatly enlarged when
global ****ysis of gene expression using cDNA arrays technologies are
used to identify thyroid hormone target genes. Most of the genes
identified so far are expressed and regulated by thyroid hormone during
the postnatal period (Fig. 1), and the role of thyroid hormone is to
accelerate the normal physiologic process of upregulation or
downregulation that these genes experience after birth. A good example
is provided by the myelin genes, which are induced a few days after
birth, in parallel with the timing of oligodendrocyte differentiation
and the myelination wave. In the absence of thyroid hormone,
ac***ulation of myelin gene products, mRNA, and protein proceeds at a
slower rate, and final normal concentrations are attained but later in
development than in normal animals. Other genes show a region-specific
dependence of T3. One example is RC3. This gene is expressed in subsets
of neurons of the cerebrum, and thyroid hormone is needed to achieve
normal expression in discrete regions such as layer VI of the neocortex
and retrosplenial region, caudate nucleus, and dentate gyrus. The gene
is expressed in other regions such as the upper layers of neocortex, and
pyramidal cells of hippocampus, but it is not sensitive to thyroid
hormone in these locations despite the presence of T3 receptors. Because
RC3 is regulated by T3 directly at the transcriptional level,[46,47] the
most likely explanation forsuch a region-specific control is that
regulation of this gene is based on a combinatorial distribution of
transcription factors, including T3 receptors. Therefore, in sensitive
regions, T3 receptors might complement the pool of transcription factors
needed for target gene expression.

Click to zoom Figure 1. (click image to zoom) Thyroid hormone-regulated
genes in rat brain. The dependency of gene expression on thyroid hormone
supply is represented schematically along a time line from conception
(C) through embryonic (E) and postnatal (P) development. Birth is
indicated by P0. Most genes are affected by the thyroidal status only
during a limited period of development, which is indicated by the
horizontal bars. NGF, TrkA, and RC3 are also thyroid hormone-dependent
during adult life. A few genes, such as those encoding NSP-A, Oct-1, and
RC3 have also been shown to be regulated during embryonic development.
The inset delimited by the discontinuous lines illustrates the regional
and temporal pattern of thyroid hormone regulation of myelin genes which
follows the myelination wave, from the caudal to rostral regions. BS,
brain stem; CB, cerebellum; MB, midbrain; Hipp, hippocampus. At the
bottom of the figure, the approximate time of first detection of TR? and
of TR?, the appearance of the thyroid gland, and the peak generation of
neurons, astrocytes and oligodendrocytes. Modified from Bernal[42] with
permission from Elsevier.

Another example of region-specific control is ZAKI-4. This gene encodes
a calcineurin inhibitor expressed throughout the brain,[48] but
sensitive to T3 only in layer VI of neocortex. Regulation of this gene
by T3 is indirect because it requires new protein synthesis. In this
case, it is likely that in layer VI, T3 regulates expression of a
protein required for ZAKI-4 expression. Recent evidence suggests that
the mechanism of ZAKI-4 induction by thyroid hormone involves activation
of phosphoinositide-3 kinase.[49]

The temporal patterns of thyroid hormone-dependent gene expression in
the brain suggest that the critical period of thyroid hormone
sensitivity is limited to the first 2-3 postnatal weeks in the rat. In
humans the sensitive period would correspondingly start after
midpregnancy. However, there may be a bias in this concept, derived from
the fact that most searches for thyroid hormone-dependent genes in the
brain have been made during the postnatal period, at the peak of T3
receptor expression and occupancy. From the data illustrated in Figure
1, it can be seen that the period of thyroid hormone sensitivity for
some genes extends to the adult period (RC3, TrkA, and NGF) whereas for
others it starts during the fetal period (tenascin C, L1, reelin).
Earlier ages were not ****yzed. In other studies[50,51] it was found
that maternal hormones influenced fetal brain expression of RC3, NSP-A,
and Oct-1. Maternal hormones have recently been demonstrated to play an
important role in cell migration in the fetal neocortex.[52] The
application of global ****ysis of gene expression using suitable models
of fetal hypothyroidism may help to identify T3-regulated genes during
fetal brain development.

The primary action of thyroid hormone on gene expression is mediated
through interaction of the T3 receptors with responsive elements located
in gene regulatory regions.[53] Some of the genes known to be responsive
to thyroid hormone in the brain contain triiodothyronine response
elements (TREs) and in some cases the action of T3 has been shown to be
at the transcriptional level in vitro. Genes containing TREs in their
promoter or intronic regions include those encoding myeline basic
protein,[54] the Purkinje cell-specific gene (PCP2),[55] which encodes a
G protein nucleotide exchange factor,[56] the calmodulin binding and
protein kinase C (PKC) substrate RC3,[47] prostaglandin D2
synthetase[57,58] the transcription factor hairless,[59] the neuronal
cell adhesion molecule (NCAM),[60] and the early response gene
NGFI-A.[61] Expression of other genes are regulated at the levels of
mRNA stability (acetyl cholinester-ase), protein translation (MAP2), or
mRNA splicing (tau). Regulation of splicing might be indirect, and
subsequent to a primary action on the transcription of splicing
regulators.[62] Role of Thyroid Hormone Receptors

In mammals, T3 receptors are the products of two genes known as TR? and
TR? that encode nine protein products that arise by alternative splicing
and differential promoter usage. The TR? gene encodes five protein
products (TR?1, TR?2, TR?3, and the truncated products DTR?1 and DTR?2)
from which only TR?1 binds T3. The TR? gene encodes four T3 binding
proteins, of which TR?1, TR?2, and TR?3 also bind to responsive elements
in DNA. In addition, a truncated protein, DTR?3 binds T3 but not DNA. It
may be said, therefore, that there are two types of receptors, ? and ?,
and four different receptor isoforms. The physiologic role of the
nonreceptor proteins is at present still unclear.

Concerning whether different receptor isoforms subserve different
physiologic functions by selectively regulating specific genes, the
current view is that the receptor isoforms are mostly equivalent in
their biologic activity in vivo, including binding affinity for T3,[63]
and that the different physiologic roles of each receptor depends on
their particular patterns of expression. Thus, TR? is involved in
regulation of pituitary, liver, and cochlear function, whereas TR?
regulates cardiac function, body temperature, gut maturation, and
lymphocyte development. These specific effects of each receptor type are
mainly due to their tissue distribution. In the cerebellum, TR? is
expressed in the granular cells, whereas TR? is expressed in the
Purkinje cells. Therefore, it is not surprising that the effects of T3
on migration if granular cells is mediated by TR?, whereas those on
differentiation of Purkinje cells are mediated by TR?.[64] Another
recent example (see below) is the finding of quantitative differences in
TR?1 and TR?1 expression among subsets of GABAergic interneurons of the
cerebral cortex and hippocampus, which is correlated with effects on
behavior.[65] A different case has been made on the control of
thermogenesis in brown adipose tissue, where the use of the
TR?-selective ligand GC-1 has allowed to define specific functions for
TR?1 and TR?1.[66]

A prominent role of TR?1 in brain development and function may be
deduced from its relative expression in cerebrum and cerebellum,
accounting for approximately 70%- 80% of total T3 receptor binding.[67]
Given the widespread distribution of T3 receptors in the brain it is
noteworthy that mutant mice lacking TR?1, TR?, or both, do not display
obvious signs of developmental abnormalities. One possible explanation
for this paradox is that in the absence of ligand, transcriptional
repression or abnormal regulation of transcription by the unliganded
receptor is responsible for the effects of profound hypothyroidism. This
is supported by two lines of evidence. One was provided by Samarut et
al.,[68] which showed that congenitally hypothyroid, Pax 8-deficient
mice, which die during the first weeks of life, can be rescued by TR?1
gene deletion. Other evidence was provided by our group in studies on
cerebellar development in TR?1-deficient mice.[64] In these studies we
showed that contrary to what happens in wild-type mice, the mutant mice
do not show delayed granular cell migration and arrested Purkinje cell
differentiation after induction of neonatal hypothyroidism. It appears,
therefore, that these cerebellar alterations which constitute one of the
hallmarks of neonatal hypothyroidism in rodents are the result of
negative influences of the unliganded TR?1 cerebellar expression of a
mutated form of TR?1 with dominant negative activity strongly arrested
Purkinje cell differentiation.[69] Role of Thyroid Hormone Receptors in
Behavior

As pointed out above, absence of thyroid hormone receptors is not
associated with obvious developmental alterations. The possibility
exists, however, that absence of receptor during development may lead to
subtle alterations derived from disturbances in correct maturation of
specific neurons. A role of TR?1 in early neuronal differentiation has
been suggested from studies using stem cells in culture.[25,70] It is
likely that receptor deficiency is therefore associated with alterations
in differentiation or migration of selected neuronal groups leading to
subtle anomalies of brain function, not easily related to receptor
function. Behavioral ****ysis of receptor-deficient mice may disclose
alterations not previously expected on the basis of the known classical
effects of thyroid hormone. TR?-deficient mice did not present any
alterations in several tests including the open field test, the Morris
water maze test, or contextual fear conditioning tests.[71] However,
female mice with deletions of either TR?1 or TR?1 show opposite response
to estrogens on *** behavior:[72] absence of TR?1 led to an increase
***ual behavior, whereas absence of TR?1 reduced ***ual behavior. In
other terms, ligand-activated TR?1 would be an inhibitor of ***ual
response, whereas liganded TR?1 would have the opposite function. The
data for TR?1 fit with results from Pfaff and coworkers[73] that showed
that thyroid hormone may interfere with the actions of estrogens on
female *** responses. At the molecular level it was shown that TR may
interfere with estrogen induction of responsive genes by interactions at
the level of the estrogen responsive elements of target genes.[73] The
fact that TR?1 behaved differently than TR?1 in these behavioral
paradigms may be because of additional roles of TR?1 in other forms of
behavior.

We have ****yzed whether deletion of TR?1 would lead to behavioral
alterations in adult mice using several behavioral tests,[65] the first
of which was the open-field test to assess locomotor behavior in a new
environment. In this test the animals were more prone to display
"freezing," (i.e., to stay motionless) (Fig. 2). This indicated that the
mice had high emotionality. In a second battery of tests, the animals
learn to associate an innocuous stimulus (the conditional stimulus),
which might be either the context (as in contextual fear conditioning)
or a cue, for example a sound (as is cued fear conditioning), with an
aversive stimulus (unconditional stimulus) that is usually an electric
shock delivered through metal rods in the cage floor. As shown in Figure
2, there were no differences in behavior at 24 hours of training.
However, the extinction of the response was surprisingly delayed in the
context fear conditioning test. In other words, TR?1-deficient mice have
a reduced capacity to forget negative memories. The results of these
experiments, combined with data on the microstructure of the hippocampus
and cerebral cortex, as well as the cellular distribution of thyroid
hormone receptors have led us to hypothesize that TR?1 is involved in
the specification of hippocampal neuronal circuits involved in
behavior.[65]

Click to zoom Figure 2. (click image to zoom) Behavior of adult
TR?1-deficient mice (solid boxes) in the open field (A) and the
contextual (B) and cued (C) fear conditioning tests. The behavior of
control wild-type mice is represented in open boxes. In the open field
test the animals were placed in the center of a circular open field of
140 cm diameter and 32 cm height, and the behavior and locomotor
activity were videotaped. A significant number of TR?1-deficient mice
stayed motionless (freezing), displayed less rearing, and spent less
time in the center area of the field than the wild-type mice. The fear
conditioning tests took place in Pavlovian conditioning cages. In this
context, fear conditioning, which depends on the integrity of both
amygdala and hippocampus, the mice learn to associate the context with
an aversive stimulus (i.e., an electric shock). Extinction of the
response took much more time for the TR?1 knockout mice. In cued-fear
conditioning, a test that depends on the integrity of the amygdala, the
mice learn to associate the aversive stimulus with a cue, usually a
sound. In this test, both learning (24-hour test) and the extinction of
the response (1 week) were similar in knockout and wild-type mice. The
results suggested a lesion at the level of the hippocampus, and were
correlated with a decreased number of GABAergic terminals in the CA1
field. Reproduced from Guadaño-Ferraz et al.[65] with permission.

Conclusions

From the data ac***ulated during the past 10 years, we are starting to
understand the general features of thyroid hormone action in the brain
in molecular terms. Knowledge in this field has, however, had a very
strong limitation because of the complexity of the target organ under
study, which requires the convergence of several disciplines. In
addition to a more precise understanding of the classic effects of
thyroid hormone deficiency or excess, the ****ysis of mice with
genetically altered thyroid hormone receptors are providing surprising
and unexpected results. In particular, the notion that some features of
the hypothyroid phenotype may be the result of altered transcriptional
regulation by the unliganded receptors more than to the lack of hormone
per se. In addition, the indication that thyroid hormone receptors may
have a role in behavior opens new fields of inquiry and raises the
intriguing possibility that receptor mutations may underlie psychiatric
forms of behavior in humans.

Funding Information

Supported by grants BF12002-00489 of Ministerio de Ciencia y Tecnología,
and 08.5/0044.1/2000 from Comunidad de Madrid. Reprint Address

Dr. Juan Bernal, Instito de Investigaciones Biomedicas, Arturo Duperier
4, 28029 Madrid, Spain. E-mail: jbernal@iib.uam.es


Juan Bernal, Ana Guadaño-Ferraz, and Beatriz Morte, Instituto de
Investigaciones Biomédicas Alberto Sols, Consejo Superior de
Investigaciones Científicas y Universidad Autónoma de Madrid, Madrid,
Spain
Thyroid 13(11):1005-1012, 2003. © 2003 Mary Ann Liebert, Inc.

http://www.medscape.com/viewarticle/465419?mpid=22427