Behavioral Inhibition and Impaired Spatial Learning and Memory in Hypothyroid Mice Lacking Thyroid Hormone Receptor α

Abstract

Thyroid hormone insufficiency leads to impaired neurogenesis, behavioral alterations and cognitive deficits. Thyroid hormone receptors, expressed in brain regions involved in these behaviors, mediate the effects of thyroid hormone deficiency or excess. To determine the contribution of thyroid hormone receptor alpha (TRα) in these behaviors, we examined the behavior of euthyroid as well as hypo- and hyperthyroid mice lacking all isoforms of the TRα (TRα^o/o). The hypothyroxinemic TRα^o/o mice demonstrated behavioral inhibition, manifested in decreased activity and increased anxiety/fear in the open field test (OFT) and increased immobility in the forced swim test (FST) compared to C57BL/6J mice. TRα^o/o mice also showed learning and recall impairments in the Morris water maze (MWM), which were exaggerated by hypothyroidism in TRα^o/o mice. These impairments were concurrent with increased thigmotaxis, suggesting an increased anxiety-like state of the TRα^o/o mice in the MWM. Expression of genes, known to be involved in processes modulating learning and memory, such as glucocorticoid receptor (GR), growth-associated protein 43 (GAP-43) and neurogranin (RC3), were significantly decreased in the hippocampus of TRα^o/o mice. GR expression was also decreased in the frontal cortex and amygdala of TRα^o/o mice, indicating that expression of GR is regulated, probably developmentally, by one or more isoforms of TRα in the mouse brain. Taken together these data demonstrate behavioral alterations in the TRα^o/o mice, indicating the functional role of TRα, and a delicate interaction between TRα and TRβ-regulated genes in these behaviors. Thyroid hormone-regulated genes potentially responsible for the learning deficit found in TRα^o/o mice include GR, RC3 and GAP-43.

Introduction

In all vertebrates, thyroid hormones are essential for normal brain development [11, 12, 27, 61]. Experimentally induced moderate and transient deficiency of maternal thyroid hormones (TH) have profound effects on fetal neurogenesis [5] and brain growth; even mild hypothyroidism of the mother decreases maternal thyroxine (T4) transport to the fetus [57]. This mild maternal hypothyroidism can negatively affect the child’s neuropsychological development [20, 42], since the fetal thyroid is unable to produce any T4 before 12–14 weeks of gestation [42].

Research suggests that thyroid function abnormalities in adulthood can also have profound behavioral consequences. Clinical hypothyroidism is often associated with frank neuropsychological and affective alterations, and considered to be one of the causes of reversible dementia. Even mild hypothyroidism predicts lower levels of memory performance [55]. Additionally, moderate to severe depression and mania are associated with reduced TSH response to TRH [44].

In animals, hyperthyroidism in adulthood leads to increased activity in the open field test [46, 50], and both experimental hypo- [25] and hyperthyroidism [46] have been reported to increase immobility behavior in the forced swim test. Spatial learning and memory deficits have been found in the genetically hypothyroid hyt/hyt mouse [3], and in congenitally hypothyroid animals [1, 2, 32].

Behavioral consequences of developmental or adult thyroid dysfunction are mediated by actions of thyroid hormones on thyroid hormone receptors (TR), predominantly by positive or negative transcriptional regulation of thyroid hormone responsive genes [39]. The ligand-activated nuclear TR isoforms, TRα1, TRβ1 and TRβ2, are produced by alternative splicing of the primary transcripts of the TRα and TRβ genes, respectively [52, 58]. Another spliced product of the TRα gene, TRα2, does not bind TH [29]. The functions of the TR α1 and TR α2 isoforms, which are also expressed in the brain, are still unknown [31]. Recently, mice, lacking both TRα1 and TRα2 and the shorter transcripts initiated from the internal promoter located in intron 7, [15] have been generated offering the opportunity to characterize the role of these receptors in various behaviors. These TRα^o/o mice have decreased T4, but normal TSH levels [16, 31]. As such, the TRα^o/o mice can model the mild hypothyroxemia, which leads to behavioral consequences, when it occurs during development and also in adulthood.

In this study, we aimed to determine the behavioral consequences of deletion of the TRα locus. To achieve this aim we employed TRα^o/o mice and wild type C57BL/6J (B6) mice (the genetic background of the TRα^o/o mice). Since we preferred to assure the normal thyroid milieu of the fetal C57BL/6J mice, we did not choose wild type littermates of the TRα^o/o mice as controls. Instead, we considered the TRα^o/o and C57BL/6J mice as two separate strains that differ genetically at their TRα locus, and developmentally in their thyroid hormonal milieu. To evaluate the role of the adult hypothyroxinemia on behaviors known to be affected by thyroid hormones, we induced hypo- or hyperthyroidism in both strains of adult mice. The behaviors examined included spatial learning and memory in the Morris Water Maze (MWM), activity and anxiety/fear in the open field test (OFT) and immobility in the forced swim test (FST). In addition, we evaluated the expression of GAP-43 and glucocorticoid receptor (GR), both implicated in learning [38, 48, 60], and the thyroid hormone-regulated gene, neurogranin (RC3), in brain regions involved in these behaviors.

Materials and Methods

Animals

Male C57BL/6J mice (Center for Comparative Medicine, Northwestern University, Illinois, USA) and mice with a homozygous null mutation for all forms of thyroid receptor alpha (TRα^o/o) on a C57BL/6J background [31] were housed in a climate-controlled facility with a 12/12 light-dark cycle (lights on 7:00 h). TRα^o/+ mice were crossed 11–13 times with C57BL/6J B6 mice before heterozygotes were intercrossed to result the TRα^o/o homozygotes (Dr. Samuel Refetoff, personal communication). After weaning, male mice were housed by strain (n=4–5/cage), and cages were composed of different litters. At six weeks of age, mice of both strains were placed on one of three diets. As a control diet, mice (n=13 mice/strain) were fed the standard rodent diet from Harlan Teklad (Madison, WI, USA). To elevate thyroid hormone levels, animals (n=13/strain) had unlimited access to water containing 0.0012% thyroxine (T4, Sigma T-2376) as described previously for rats [47]. To deplete thyroid hormone, mice (n=9/strain) were fed an iodine deficient diet containing 0.15% propylthiouracil (PTU, Harlan Teklad, Madison, WI, USA). The experimental diets were maintained throughout the study.

Behavioral tests were administered in the following order to the same animals, always in the morning between 1000h and 1200h. The OFT was administered first at 8 weeks of age, two weeks later the Morris water maze was performed, followed by the forced swim test after two weeks of rest.

Radioimmunoassay

Plasma T4 concentration was evaluated prior to behavioral testing and at the conclusion of the study. Blood was collected by tail nick four days prior to the beginning of the behavioral battery and again by heart puncture at the conclusion of the experiment. Total plasma T4 concentration was measured in a single T4 RIA kit from ICN Pharmaceuticals (Costa Mesa, CA, USA). The assay sensitivity was 0.3 μg/dl, and the intraassay and interassay coefficients of variation were 4.9% and 5.4%, respectively.

Open Field Test

Following two weeks of treatment with T4 or PTU-containing diet, mice were exposed to the open field test (OFT). The open field is a brightly lit circular arena with a total diameter of 60 cm and height of 30 cm. The arena is divided into inner and outer circles, where the diameter of the inner circle is 30 cm. Mice were placed in the center of the field, and their activity was recorded for 5 minutes. The test was recorded with a digital camera using the computer software Limelight from Actimetrix (Evanston, IL, USA). Limelight records the animal’s latency to leave the center (a region in the center of the inner area with a 10 cm diameter), the time spent in the inner circle, the distance traveled in the inner circle, and the total distance traveled during the test.

Morris Water Maze

At 10 weeks of age, the mice were exposed to the MWM. The pool (d=108 cm) was filled with water (24–25^oC) and made opaque by non-toxic white tempera paint (Palmer Paint Products, Inc, Troy, MI, USA). The test consists of two days of 12 one-minute training trials, during which the mice attempted to locate a hidden platform submerged 1–2 cm below the surface of the water using visual cues located on the walls. The platform (d=10 cm) was placed in the middle of one of the quadrants of the pool. The training trials lasted one minute with a 5–6 minute rest period between trials. The trials began at one of four starting locations in bins of three. Thus, the first three training trials began from the same starting location; the next three trials began from a different starting location, and so on for the third and fourth sets of trials. The order of the starting locations was chosen at random and differed between the two days. When the mouse found the hidden platform, it remained on it for thirty seconds and then was removed from the pool. If the mouse was unable to locate the platform within sixty seconds, the animal was placed on the platform and remained there for thirty seconds. Following the final training trial on day two and a five-minute rest period, the mice underwent a one-minute probe trial in which the platform was removed from the pool, and the mice began from a unique starting location directly opposite the platform. During the probe trial, mice remained in the pool for the entire sixty seconds. Training ended at this point due to the plateau in the latency learning curve. All trials were recorded with a digital camera using the computer software WaterMaze from Actimetrix (Evanston, IL, USA). WaterMaze recorded the amount of thigmotaxis (% time spent < 5cm from the edge) and latency to platform of all trials. Additionally, WaterMaze recorded total distance traveled and the number of platform crossings during the probe trial.

Forced Swim Test

Between 13 and 15 weeks of age, mice were exposed to the forced swim test (FST). Mice were placed into water (23–25^oC) for 6 minutes. The FST tank was a beaker (d=13 cm) filled with water to a height of 12.5–13 cm. The test was videotaped and blindly scored using a method for rats described previously [53]. The animal was observed every five seconds beginning with the animal's entrance into the water and scored as either floating or swimming throughout the test. Floating was defined as the lack of movement of any limbs.

Brain Dissections

One week after the FST, mouse brains were rapidly dissected on ice and immediately placed on dry ice. Dissections used Paxinos coordinates [41]: amygdala (AP –0.58 to –2.18, ML 1.5 to 4.5, DV 4 to 5.75), frontal cortex (AP 3.2 to 2.34, ML 1.25 to 1.25, DV 0.6 to 2.25) and hippocampus (AP –0.58 to –2.18, ML 3.1 to 3.1, DV 1.1 to 2.75 and AP –2.3 to –3.16, ML 1.25 to 3.75, DV 1 to 5). Tissues were stored at −80°C.

RNA Isolation and Northern Analysis

Extraction of total RNA was performed using Trizol reagent (Life Technologies, Grand Island, NY, USA). The quality and quantity of RNA were analyzed by gel electrophoresis and spectrophotometry. For northern analysis, 8–10 μg of RNA from each sample was electrophoresed on a 1% agarose-formaldehyde gel, blotted onto a nitrocellulose filter, and fixed by UV-crosslinking. Probes were labeled with [α-32P]dCTP by random primer labeling as described previously [45] using the Random Primers DNA Labeling System kit (Life Technologies, Grand Island, NY). The GR, RC3 and GAP-43 probes were generated by PCR using primers specific for the glucocorticoid receptor cDNA (5′-TGCAGCAGTGAAATGGGCAA-3’; 5′-GGGAATTCAATACTCATGGTC-3’), RC3 cDNA (5′-AGCCAAGGACCCTCAACAC-3’; 5′-GAGGGGCTCACAAACACAGT-3’), and GAP-43 cDNA (5′-ACAAGGAAAAAGCTCAAAG-3’; 5′-GACGGGGAGTTATCAGTGG-3’). The generated GR, RC3 and GAP-43 probes were 533 bp, 356 bp and 300 bp, respectively. The plasmid containing the β-actin cDNA probe was kindly provided by Dr. Michael Prystowsky, Albert Einstein University. Filters were hybridized and washed as previously described [45]. Specific mRNA levels were normalized to the β-actin mRNA level of each sample.

Statistics

Data were analyzed by a two-way analysis of variance (ANOVA, factors: strain, treatment) with the exception of MWM, which was analyzed by 3-way ANOVA with repeated measures. The Bonferroni Multiple Comparisons test, with a p < 0.05 adjusted significance level, was used when appropriate to identify significant differences between groups. In some situations, hypothesis testing using a Student’s t-test was applied to determine the existence of significant differences between specific groups in the absence of significant strain-treatment interactions. All data are presented as mean + SEM.

Results

Hypo- or hyperthyroidism was confirmed in our animals by measuring plasma T4 levels at the beginning of the study (Table 1). Both treatments changed T4 levels significantly (F[2,56]=102.7; p<0.001) from untreated controls, but no significant strain differences were observed. To confirm the presence of the previously characterized hypothyroxinemia in the TRα^o/o mice, we carried out a planned comparison of T4 levels between untreated B6 and TRα^o/o mice. As expected, untreated TRα^o/o mice had significantly lower plasma T4 levels than untreated B6 mice (Table 1; F[1,21]=28.3; p<0.001).

Table 1.

Plasma Thyroxine (T4) levels at the beginning of behavioral experiments.

Group	T4 (μg/dl)
B6	1.78+/−0.15
B6+T4	9.04+/−1.49*
B6+PTU	0.49+/−0.08*
TRαo/o	0.79+/−0.07#
TRαo/o+T4	10.09+/−0.7*
TRαo/o+PTU	0.32+/−0.02*

Values are means +/− SEM of n=9–13 mice per group.

^*p<0.05 from euthyroid mice of the same strain,

^#p<0.05 strain difference.

TRα^o/o mice are hypoactive in the OFT and FST

The open field test is a well-established behavioral assay that can be used as a test of fear and anxiety (or emotionality) and exploratory behavior under stressful conditions. Two measures of anxiety were taken: latency to leave the center of the arena and time spent in the center vs the periphery. An anxious mouse may hesitate to go to the periphery and freeze for a short while, thus registering a higher latency to leave the center, and will spend less time moving in the center area of the arena. The total distance traveled by the animal is a measure of locomotor activity under the stress of novel environment.

In the OFT, there was a general strain difference with TRα^o/o mice showing a hypoactivity/increased anxiety profile compared to B6 mice. The increased anxiety profile did not seem to be related to the adult hypothyroxinemia of the TRα^o/o mice, since even frank hypothyroid state did not alter the general behavioral pattern of B6 mice. Specifically, TRα^o/o mice took significantly longer to leave the center of the open field (latency from center) compared to B6 mice regardless of treatment (Figure 1a; Strain: F[1,67] = 8.90, p< 0.01), a measure often identified with increased anxiety/depression-like behavior. In agreement with this finding, TRα^o/o mice were less active in the inner circle of the open field compared to B6 mice regardless of thyroid status (inner distance; Figure 1c; Strain: F[1,67] = 1, p< 0.001). TRα^o/o mice also showed decreased locomotion, and hypothyroidism, induced by PTU diet, reduced total distance traveled in both strains (total distance; Figure 1b; Strain: F[1,67] = 15.22, p< 0.001; Treatment: F[2,67] = 3.51, p< 0.05).

Figure 1.

Open Field Test Behavior. At 8 weeks of age, euthyroid (control), hyperthyroid (T4) and hypothyroid (PTU) C57BL/6J, the wild type (B6) and TRα^o/o mice (n=9–13/group) were exposed to a novel open field. (a) After being placed in the center of the arena, a subject’s latency to leave the center was measured in seconds (Latency from Center), (b) The total activity of the mice was measured by recording the total distance traveled in cm (Total Distance), (c) The activity of the mice in the inner region was measured by recording the distance traveled in the inner region in cm (Inner Distance). Values are means ± SEM. Asterisks indicate significant differences within a strain from the euthyroid control group (* = p< 0.05). Bars with a significance level above indicate significant strain differences.

The mouse forced swimming test (FST) [43] is a useful model to predict antidepressant activity [9]. Genetic factors, such as strain, contribute to the behavioral performance of mice in this model of depression. TRα^o/o mice tended to float more (greater immobility) than B6 mice, although PTU treatment reversed this trend: hypothyroidism increased floating (immobility) in the B6, but decreased it in the TRα^o/o mice (Figure 2; Strain: F[1,65]= 1.8, p= 0.069; Strain x treatment; F[2,65] = 6.2, p< 0.01).

Figure 2.

Forced Swim Test (FST). Between 13 and 15 weeks of age, mice (n=8–12/group) described in Figure 1 were subjected to the FST for 6 minutes. Behavior was recorded in five second bins using a time-sampling method. Values are means ± SEM. Asterisks indicate significant differences within a strain (*** = p< 0.001). Bars with a significance level above indicate significant differences between strains with the same treatment.

Impaired learning and increased thigmotaxis in TRα^o/o mice

Since we were interested in spatial learning and memory in the TRα^o/o mice under conditions that are similar to those employed in the previous two tests, namely novel environment, we only employed the hidden platform version of this task. In the novel environment of the Morris water maze, TRα^o/o and B6 mice did not differ in their latency to locate the submerged platform on day 1 indicating no motor or visual impairments between these strains regardless of treatment. However, on day 2, TRα^o/o mice took significantly longer to find the hidden platform of the MWM than B6 mice, and PTU treatment further increased this learning deficit (Figure 3a; Strain: F[1,67] = 8.93, p< 0.01; Treatment: F[2,67] = 8.28, p< 0.001). Since the strain x treatment interaction only approached significance (F[1,67]=2.7, p=0.076), we also used a planned comparison of the hypothyroid mice of the different strains. When compared using a Student t-test, the TRα^o/o+PTU animals were found to take significantly more time to locate the platform than the TRα^o/o mice on day 2 (Bins 5–8), while PTU treatment had no effect on the learning behavior of the B6 mice (Figure 3a).

Figure 3.

Animals (10–12 weeks old, n=9–16/group) underwent training in the MWM to locate a hidden platform. Training consisted of 24 trials over two days. Trials occurred in bins: three one-minute trials conducted from the same starting location before moving the animal to the next starting location. (a) Latency for the animal to reach the platform was recorded. (b) Percentage of thigmotaxis (time spent near the edge of the pool) was chronicled during training. (c) Number of platform crossings in the quadrant of the hidden platform during the probe trial. (d) Percentage of thigmotaxis during the probe trial as measured. Values are means ± SEM. Asterisk indicates a significant (p<0.05) difference from groups with the same treatment but of different strains (a and b) or p<0.01 differences due to treatment within the strain (c and d). Bars with a significance level above indicate significant differences between strains with the same treatment (b and c).

Similar to the latency to platform measure, there were no significant differences across groups in thigmotaxis on day 1. On day 2, thigmotaxis (Figure 3b) was significantly greater in the TRα^o/o compared to the B6 mice regardless of treatment (day 2, Strain: F[1,67]= 15.09, p< 0.001), suggesting that their inability to locate the platform was not related to decreased activity, but rather to increased anxiety. Again, similar to the latency to platform learning measure, only the PTU-treated TRα^o/o mice showed increased thigmotaxis during the second day of training, while PTU treatment did not significantly affect performance of the B6 mice (Strain x treatment; F[2,67]=5.77, p<0.01). There were no significant differences in swim speed between groups.

In the probe trial, there were no significant strain differences in the number of platform crossings (Figure 3c). However, PTU treatment decreased their recall ability, particularly in the TRα^o/o mice as indicated by their decreased number of platform crossings (Strain x treatment; F[2,67]=3.75, p<0.05) compared to TRα^o/o controls. Thigmotaxis during the probe trial (Figure 3d) resembled the thigmotaxis during training: the hypothyroid TRα^o/o animals spent more time near the edge of the water maze than their euthyroid counterparts or the hypothyroid B6 mice (Strain x Treatment: F[2,67] = 10.90, p< 0.001).

Thus, TRα^o/o mice showed spatial learning and recall deficits in the MWM, which may be due to their increased anxiety-like behavior. The hypothyroid TRα^o/o, but not the B6, mice showed an exaggerated deficit.

Differential Decreased Expression of GR, GAP-43 and RC3 in response to developmental and experimental hypothyroidism in TRα^o/o mice

Both GR and GAP-43 have been extensively studied and implicated in learning, and RC3 is positively regulated by thyroid hormones in the rodent brain [23, 36]. Therefore, we determined the expression of GR, GAP-43 and RC3 in the amygdala, frontal cortex and hippocampus of our euthyroid, hypothyroid and hyperthyroid B6 and TRα^o/o mice. These three brain regions are involved in the behaviors tested in this study [7, 8, 26, 30].

Alterations in GR, GAP-43 and RC3 mRNA levels were found in the amygdala, frontal cortex and hippocampus of TRα^o/o mice (Table 2). GR, RC3 and GAP-43 expressions were significantly decreased in the hippocampus of TRα^o/o mice compared to B6 mice (Strain: GR: F[1,30]=7.8; p<0.05; RC3: F[1,30]=14.5; p<0.01; GAP-43: F[1,30]=7.26; p<0.01). In addition to the hippocampus, GAP-43 and GR expression were also significantly decreased in the amygdala of TRα^o/o mice compared to B6 regardless of treatment (Strain: GAP-43: F[1,30]=14.2; p<0.001; GR: F[1,30]=52.8, p<0.001). Moreover, GR mRNA levels were significantly lower in the frontal cortex of TRα^o/o mice compared to B6 (Strain: F[1,30]=47.9, p<0.001).

Table 2.

Messenger RNA levels of GR, GAP-43 and RC3 in amygdala, frontal cortex and hippocampus.

Group	GR	GAP-43	RC3
Amygdala
B6	1.40 +/− .16	2.23+/− .23	1.44 +/− .20
B6+T4	0.93 +/− .27	2.11 +/− .32	1.45 +/− .23
B6+PTU	0.76 +/− .29	1.40 +/− .53	1.23 +/− .11
TRαo/o	0.60 +/− .17#	1.57 +/− .12#	1.07 +/− .20
TRαo/o+T4	0.87 +/− .38	1.20 +/− .18	1.40 +/− .18
TRαo/o+PTU	0.74 +/− .26	0.84 +/− .15	1.11 +/− .21
Frontal
Cortex
B6	1.50 +/− .15	1.46 +/− .31	0.72 +/− .20
B6+T4	1.03 +/− .14	1.47 +/− .35	0.65 +/− .11
B6+PTU	0.64 +/− .18	1.79 +/− .33	0.56 +/− .13
TRαo/o	0.29 +/− .15#	2.20 +/− .27	1.02 +/− .14
TRαo/o+T4	0.42 +/− .20	1.68 +/− .34	0.70 +/− .13
TRαo/o+PTU	0.39 +/− .16	1.67 +/− .36	0.51 +/− .08*
Hippocampus
B6	2.83 +/− .50	1.45 +/− .24	2.13 +/− .36
B6+T4	2.29 +/− .41	1.13 +/− .21	2.34 +/− .32
B6+PTU	1.48 +/− .56	0.74 +/− .52	1.77 +/− .50
TRαo/o	0.58 +/− .43#	0.47 +/− .22#	0.31 +/− .40#
TRαo/o+T4	1.76 +/− .72	1.10 +/− .30	0.98 +/− .77
TRαo/o+PTU	1.32 +/− .68	1.49 +/− .23*	1.77 +/− .30*

Specific mRNA levels were normalized to the β-actin mRNA level of each sample. Values are means ± SEM.

^*p<0.05 from euthyroid mice of the same strain,

^#p<0.05 strain difference.

Thyroxine administration had no effect on the expression of these genes, but PTU treatment had strain-dependent effects (Table 2). In the TRα^o/o mice, hypothyroidism increased hippocampal GAP-43 mRNA and RC3 mRNA levels (GAP-43: Strain x treatment; F[1,30]=13.6; p<0.001; RC3: Strain x treatment; F[1,30]=3.92, p<0.05) and decreased frontal cortex RC3 expression (Strain x Treatment: F[1,30]=5.1, p<0.05;).

Discussion

The present study demonstrated that deletion of all TRα isoforms was coupled with increased behavioral inhibition and spatial learning and memory deficits. Most of these effects were exaggerated in hypothyroid TRα^o/o mice, suggesting that thyroid hormones can enhance behavior and increase spatial learning via TRβ receptor-mediated events.

TRα^o/o mice are hypothyroxinemic [31], and although there is no information on the thyroid hormonal profile of pregnant TRα^+/o mice, they might also show decreased T4 levels similar to TRα2 heterozygotes [51]. Since maternal hypothyroxinemia increases the risk of neurodevelopmental deficits in the fetus [28, 61], we chose to use C57BL/6J (B6) mice as controls instead of the wild type littermates of TRα^o/o mice, as customary. Thus, the potential cause of behavioral differences between the TRα^o/o and the B6 mice might include the residual original 129sv x B6 background of the TRα^o/o mice [16]. Alternatively, they may be caused by maternal hypothyroxinemia of the TRα^o/o mice and finally, but most importantly, by the lack of the TR alpha locus. Although the influence of 129sv background flanking genes, in addition to the targeted TRα ablation, cannot be excluded, the residual 129sv background is a less likely cause given that the TRα^o/o mice was backcrossed to B6 11–13 times by the time we received them (Dr. Samuel Refetoff, personal communication). Thus, the potential of the TRα^o/+ mother being hypothyroxinemic outweighed the possibility of the residual 129sv background on the offspring.

A similar consideration could be made in other studies as well, where the heterozygote dam can affect the development of their wild type and homozygote fetuses differentially, depending on their genotypes. To our knowledge, this gene-prenatal environment interaction has not been considered in practice, quite the opposite; the value of identical maternal genotype has been purported regardless of the effects of fetal genotype-prenatal environment interactions.

Finally, the behavioral strain differences found could be due to the lack of all TRα isoforms leaving only the TRβ isoforms to mediate thyroid hormone effects. Earlier studies indicate no compensatory differences in the expression of TRβ in the periphery of TRα^o/o mice [31, 49] but a region-specific compensatory over-expression of TRβ in the brain of TRα^o/o mice cannot be excluded. Regardless of the expression levels of TRβ, the behavioral consequences of TRα ablation could be due to lack of stimulation or repression of TRα-regulated genes or the activity of TRβ. The contribution of liganded TRβ on these behaviors could be ascertained by the effect of PTU treatment in TRα^o/o mice.

Behavioral consequences of the ablation of TRα, which are not exaggerated by hypothyroid milieu, may be due to the lack of stimulation or repression of TRα-regulated genes. These behaviors include the increased latency to leave the center and the decreased activity in the inner circle of the OFT, both suggesting increased fear and anxiety. Since previous studies found that TRα1^−/− mice show increased anxiety/fear-related behaviors in the OFT [19] and elevated plus maze [56], the lack of TR α1 might be responsible for the increased anxiety/fear shown by the TRα^o/o mice. On the other hand, overall locomotor activity is normal in TRα1^−/− mice [59], but moderately decreased in our TRα^o/o mice, indicating a role of TRα2 in locomotion.

Altered behaviors in the “euthyroid” TRα^o/o mice, which are worsened by their hypothyroidism, or are found only in the hypothyroid TRα^o/o mice, could be due to thyroid hormone effects mediated by TRβ. These include deficits in spatial learning and memory and increased immobility in the FST. Interestingly, impaired hippocampal-dependent learning is observed in the TRα1^−/− mice [19], but mice with TRβ deletion do not show deficits in the MWM [13, 14]. This latter finding argues against the contribution of liganded TRβ to these behaviors. However, it is important to note we found increased anxiety/fear in both the learning and memory components in TRα^o/o mice, while no such differences were reported with TRβ deletion. TRα^o/o mice demonstrated increased thigmotaxis, which is a manifestation of increased anxiety and fear. As the PTU-induced hypothyroid state further increased thigmotaxis in TRα^o/o mice, liganded TRβ may regulate anxiety/fear in the MWM. Thus, our findings suggest that absence of TRα may affect transcription of TRβ-regulated genes involved in anxiety/fear, and in turn, affect TRα–regulated hippocampal functions, fundamental to spatial learning.

Both TRα and TRβ-regulated genes seem to influence immobility behavior in the FST. Euthyroid and particularly T4-treated TRα^o/o mice demonstrated increased immobility compared to T4-treated B6 mice, while reducing thyroid hormone levels decreased immobility in TRα^o/o mice only. These changes in response to thyroid hormonal milieu suggest that TRα-regulated genes may inhibit, while TRβ-regulated genes increase, immobility behavior.

Glucocorticoid receptor, GAP-43 and neurogranin (RC3) are potential neurochemical substrates of the behavioral changes observed in the TRα^o/o mice. Immobility in the FST [6, 24], and learning and memory [38] are may be influenced by GR-mediated processes in the hippocampus. Thus, the novel finding of significantly decreased GR expression in the hippocampus (as well as amygdala and frontal cortex) of TRα^o/o mice may explain their learning impairment. GAP-43 [60] and RC3 [17] regulate calmodulin availability in axons and dendritic spines, respectively, and are involved in learning and synaptic plasticity. Under neonatal hypothyroid conditions, brain RC3 expression is significantly reduced [23, 36]. Overexpression of GAP-43 has been found to enhance learning [48], while downregulation of GAP-43 expression impairs learning [37]. Similarly, RC3 knockout mice demonstrate impaired spatial learning and memory [33] and increased anxiety/fear in the OFT [45]. Since these same behavioral abnormalities were found in the TRα^o/o mice, decreased GAP-43 and RC3 expression may contribute to their behavioral impairments. Messenger RNA level differences are likely functional, since message and protein levels typically correlate for GR [4], as well as for GAP-43 [37, 40] and RC3 [34].

Despite well-known TR-mediated actions of thyroid hormones in the brain, TR knockout mice surprisingly show no apparent abnormal brain development compared with neonatal hypothyroidism [18, 22]. It has been proposed that the structural alterations found in hypothyroid animals may result from persistent transcriptional repression exerted by the unliganded TR [21, 35]. Specifically, the lack of TRα may not only cause a loss of thyroid hormone responsiveness of target genes but could result in a broad increase in the expression of TRα-regulated genes due to the lack of constitutive silencing activity. In the TRα^o/o mice, abrogation of the constitutive silencing mediated through TRα2, the ligand-unresponsive TR with a weak dominant negative effect [11, 54], may alter the expression of TRα -mediated genes. Although we did not find any significant increases in GR, GAP-43 and RC3 in the TRα^o/o mice in any of the brain regions studied, the possible effect of decreased constitutive silencing (mediated through TRα2) on these or other genes cannot be excluded.

Taken together these data demonstrate behavioral alterations in the TRα^o/o mice, most of which are exaggerated by hypothyroid state. The strain differences in response to hypo- and hyperthyroidism in adulthood indicate the functional role of TRα, and a delicate interaction between TRα and TRβ-regulated genes in these behaviors. However, it is conceivable that the perinatal hypothyroidism of the TRα^o/o mice is responsible for the behavioral abnormalities observed, due to insufficient thyroid hormone during CNS development, rather than the complete lack of the TRα gene. Further studies are required to distinguish the effect of perinatal hypothyroidism from that of TRα deficit on these behavioral measures, as well as on the expression of thyroid hormone-regulated genes, GR, RC3 and GAP-43, potentially responsible for the learning deficit of TRα^o/o mice. Since the behaviors investigated in this study are all stress-responsive, further examination of the potential cross-regulation between stress-induced anxiety/fear, thyroid function and learning and memory is warranted.