Type 3 Deiodinase Deficiency Causes Spatial and Temporal Alterations in Brain T₃ Signaling that Are Dissociated from Serum Thyroid Hormone Levels

Abstract

The type 3 deiodinase (D3) is an enzyme that inactivates thyroid hormones (TH) and is highly expressed during development and in the central nervous system. D3-deficient (D3KO) mice develop markedly elevated serum T₃ level in the perinatal period. In adulthood, circulating T₄ and T₃ levels are reduced due to functional deficits in the thyroid axis and peripheral tissues (i.e. liver) show evidence of decreased TH action. Given the importance of TH for brain development, we aimed to assess TH action in the brain of D3KO mice at different developmental stages and determine to what extent it correlates with serum TH parameters. We used a transgenic mouse model (FINDT3) that expresses the reporter gene β-galactosidase (β-gal) in the central nervous system as a readout of local TH availability. Together with experiments determining expression levels of TH-regulated genes, our results show that after a state of thyrotoxicosis in early development, most regions of the D3KO brain show evidence of decreased TH action at weaning age. However, later in adulthood and in old age, the brain again manifests a thyrotoxic state, despite reduced serum TH levels. These region-specific changes in brain TH status during the life span of the animal provide novel insight into the important role of the D3 in the developing and adult brain. Our results suggest that, even if serum concentrations of TH are normal or low, impaired D3 activity may result in excessive TH action in multiple brain regions, with potential consequences of altered neural function that may be of clinical relevance to neurological and neuroendocrine disorders.

This work highlights the important role of the type 3 deiodinase in modulating local availability of thyroid hormones to different brain regions throughout the lifespan.

The importance of thyroid hormones (THs) for the development of the mammalian central nervous system (CNS) is widely documented (1,2,3,4). In this tissue, THs regulate the developmental expression of specific genes that are involved in a variety of cellular processes (4). Thus, a developmental deficiency in TH leads to alterations in the proliferation, differentiation and migration of various brain cell types (4) and results in impaired brain function later in life (1).

Of the two hormones secreted by the thyroid gland, T₃ and tetraiodothyronine (T₄), T₃ is considered the more active hormone because it binds to its nuclear receptor with high affinity and regulates gene expression at the transcriptional level (5). Factors affecting T₃ availability to the brain play an important role in modulating TH action. The type 2 and type 3 deiodinases (D2 and D3, respectively), which are highly expressed in the developing and mature CNS (6), are two of these factors. Whereas D2 can activate the prohormone T₄ and convert it to T₃ (6), D3 is an inactivating enzyme, in that it converts both T₄ and T₃ into inactive metabolites (7). Thus, whereas D2 contributes to T₃ availability under euthyroid conditions and helps maintain the T₃ concentration in hypothyroidism, the D3 provides an additional regulatory mechanism to control brain T₄ and T₃ and can mitigate exposure to excessive amounts of TH. This latter role of D3 is evidenced by its marked developmental pattern of expression (7). Indeed, high levels of D3 activity are present in the pregnant uterus and placenta (8,9) and most fetal and newborn tissues (7). Because early development is characterized by serum TH concentrations much lower than those in the adult (10), the action of D3 likely delays or prevents the untimely occurrence of developmental processes that are dependent on TH.

Using a genetic mouse model of D3 deficiency (D3KO mouse), we have recently shown that this is indeed the case. Due to impaired T₃ clearance, perinatal D3KO mice manifest elevated serum T₃ level and enhanced T₃-dependent gene expression in the developing brain (11). This overexposure to TH during development results in a number of abnormalities in brain function including alterations in the hypothalamic-pituitary-thyroid axis (12) and impaired hearing (13) and vision (14).

In the adult, D3 expression is much lower and limited to certain tissues (7). However, high D3 activity occurs in the adult brain, as during development (15). This suggests that D3 activity is important to modulate TH action throughout the life span of the organism.

In the present work, we have used a transgenic mouse carrying a reporter gene that expresses β-galactosidase (β-gal) as a readout of local T₃ availability (16) to analyze the level and spatiotemporal evolution of T₃ signaling in the D3KO brain as it transitions from neonatal life into adulthood and old age. We show that different brain regions exhibit different ontogeny of T₃ action and that the absence of D3 leads to excessive T₃ action in the adult brain, even in the face of diminished serum TH levels.

Materials and Methods

Animals and tissue sampling

Female mice heterozygous for the D3 mutation on an inbred 129/Sv background (11) were bred with FINDT3A and FINDT3B male mice, two transgenic mouse lines recently generated carrying a T₃-regulated β-gal transgene (16). Initially, both FINDT3 strains (termed A and B) were in a mixed OF1/B6D2 genetic background. Only very minor differences in the brain pattern of β-gal expression exist between the two strains (16), suggesting that the integration site of the transgene has no significant effect on expression patterns. Animals from the FINDT3B strain were used and are referred simply as FINDT3. However, some key findings were reproduced in mice from the FINDT3A strain. These mouse colonies were maintained by breeding animals from each subsequent generation that were heterozygous for the D3 mutation. These matings were used to produce wild-type (WT) and D3KO experimental animals. Only D3KO and WT littermates generated from heterozygous parents were used in the experiments. All experimental animals used were kept hemizygous for the β-gal transgene, except for those used as a source of samples to be used as negative controls in staining experiments. Genotyping was performed as described (11,16) by PCR of genomic DNA isolated by standard procedures from tail snips. Animals were kept under a 12-h light cycle and provided food and water ad libitum. Male mice were used in these studies, but key findings were also observed in female mice.

Animals were euthanized by asphyxiation with CO₂ (adults and 15 d of age) or by decapitation (neonates). In the adults and older neonates, blood was taken from the inferior vena cava, whereas trunk blood was collected from younger neonates. Serum was obtained by centrifugation and stored at −20 C until used. For Northern analysis, RIA, and β-gal enzymatic activity, tissues were dissected, immediately frozen on dry ice, and stored at −70 C. All animal procedures were approved by the Dartmouth College Institutional Animal Care and Use Committee.

β-Gal activities

β-Gal activities were determined in tissue homogenates using the β-gal enzyme assay kit from Promega (Madison, WI), according to the manufacturer’s instructions. For these determinations, the brain was divided into five major areas that included the cerebellum; hypothalamus (which included also the preoptic area); the midbrain (including midbrain, pons, and medulla and superior and inferior colliculus); cerebrum (including most cortical areas, striatum, and hippocampus); and the thalamus. Anatomic brain regions are designated according to Paxinos and Franklin (17). Tissues were homogenized in 0.4–2 ml of the kit’s lysis buffer and centrifuged at 4 C and 800 × g for 3 min to eliminate tissue debris. Depending on the expected activity, a volume of 20–100 μl of homogenate was used in the assay. Units are expressed as specific OD per hour and milligram of protein.

RNA preparation, Northern blot analysis, and real-time RT-PCR

Total RNA was isolated from mouse whole brain or brain areas (as described above) using the RNeasy kit from QIAGEN (Valencia, CA). Total RNA samples were electrophoresed in a denaturing 1% agarose gel containing formaldehyde and blotted onto a Nytran membrane (Schleicher and Schuell, Keene, NH). Blots were hybridized at 42 C in buffer containing 50% formamide, washed with 0.1× saline sodium citrate/0.1% sodium dodecyl sulfate at 65 C and autoradiographed for 1–7 d. Probes were labeled with radioactive ³²P-dCTP (MP Biochemicals, Solon, OH) using the oligo labeling kit (Amersham, Indianapolis, IN) and were purified through G-50 columns (Edge Biosystems, Gaithersburg, MD). Quantification of mRNA bands was performed by computer-assisted densitometry (Molecular Dynamics, Sunnyvale, CA). The cDNA probes used were hairless, a 3-kb BamH1 fragment comprising most of the coding region, and RC3, the complete 1.3-kb cDNA.

Hairless expression was also determined by quantitative real-time PCR using standard procedures. A 7300 system and a Sybr Green master mix from Applied Biosystems (Foster City, CA) was used. One microgram of RNA was reverse transcribed using standard conditions, and the resulting cDNA was amplified by PCR using a 60 C annealing temperature for 1 min. Primers used were: forward, 5′-AGCACTGTGTGGCATGTGTT-3′ and reverse, 5′-AACCCTGCATCCAAGTAGCA-3′.

In situ hybridization

In situ hybridization of brain hairless mRNA was performed as recently described (12) using a specific [³⁵S]-labeled antisense RNA probe. Hairless sense and antisense RNA probes were prepared by in vitro transcription using the T3 and T7 polymerases (Maxi-Script kit; Ambion, Austin, TX). In each case, an appropriately linearized pBluescript plasmid (Stratagene, La Jolla, CA) containing a 1.0-kb BamHI/HindIII restriction fragment from the mouse hairless cDNA was used as a template.

β-Galactosidase staining

Two alternative protocols for tissue fixation and sectioning were used. Typical results using these two protocols are shown in Supplemental Fig. 1, published on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org. In the first protocol (protocol 1), the whole brain was collected and placed in a solution of 4% paraformaldehyde in phosphate buffer for 4 h. Coronal brain sections (300 μm thick) were then made using a Vibrotome. Sections were submerged in β-gal staining solution for a time that varied between 1 and 4 h. The staining solution contained saline, 2 mm MgCl₂, 0.02% Nonidet P-40, 5 mm potassium ferricyanide, 5 mm potassium ferrocyanide, and 1 mg/ml of X-gal (5-bromo-4-chloro-3-indolyl-d-galactopyranoside). After staining, brain sections were then washed with saline, dried, and mounted on slides for analysis. In the second protocol (protocol 2), brains were initially fixed in a solution of 4% paraformaldehyde in phosphate buffer for 4 h and then overnight (16 h) in the same solution but containing 0.25 m sucrose. Tissues were then frozen and kept at −70 C until used. Using a cryostat, 50-μm brain sections were cut, mounted on glass slides, and allowed to dry 15 min at room temperature to facilitate the attachment of the section to the slide. The slides were then immersed overnight (12–16 h) in staining solution. After staining, the slides were washed with saline and dried by 1-min-long successive immersions in water solutions with increasing content of ethanol (50, 70, 90, and 100% ethanol). Slides were then allowed to dry at room temperature, mounted with coverslips, and β-gal staining analyzed.

In some experiments, for better contrast of β-gal staining, brain sections were counterstained with neutral red. Microphotographs of stained areas of the brain were taken using the same optical conditions with a Leica microscope equipped with ×10, ×20, or ×40 objectives and a Leica DC300 digital camera (Québec, Canada). Quantification of β-gal staining was based on both the area and OD of staining and was determined using SigmaScan Pro 5.1 software (Systat Software, Chicago, IL). A color and density threshold was established, typically based on the background staining of brain sections from animals not carrying the FINDT3 transgene but also based on areas of the microphotographs not showing any blue staining. Because staining differences between WT and D3KO mice seemed to vary across brain regions for most developmental ages studied, the presentation of staining results and corresponding quantification are focused on particular brain regions.

Hormone determinations and statistical analysis

Serum total T₄ and T₃ concentrations were determined using the total T₄ and T₃ Coat-a-Count RIA kits from Diagnostics Products Corp. (Los Angeles, CA) according to the manufacturer’s instructions.

Statistical analysis was performed using GB-Stat software (Dynamic Microsystems, Silver Spring, MD). Comparison between two groups was performed using the Student’s t test. Comparisons between four groups were performed by ANOVA and Fisher’s least significant differences test.

Results

Growth retardation, neonatal thyrotoxicosis, and adult hypothyroidism in D3KO/FINDT3 mice

D3KO/FINDT3 mice in the mixed OF1/B6D2/129/Sv genetic background demonstrate very similar growth and serum TH phenotypes as that described for D3KO mice in an inbred 129/Sv background (11). Thus, D3KO/FINDT3 mice exhibit significant growth retardation, which is already apparent shortly after birth and becomes more pronounced during the first 2 wk of life (Supplemental Fig. 2). The level of serum T₄ is partially suppressed at postnatal day (P) 2 and remains much lower than normal in P6 and P15 D3KO/FINDT3 animals (Fig. 1A). The level of serum T₃ is markedly elevated at P2 and P6 and becomes significantly lower than normal at P15 (Fig. 1B).

Figure 1.

Serum levels of T₄ (A) and T₃ (B) in WT and D3KO mice at different postnatal ages. Bars, mean ± se of the number of determinations indicated in parentheses. *, P < 0.01 D3KO vs. WT.

The smaller size and hypothyroidism present at weaning persist in adult D3KO/FINDT3 animals. Body weight in young (P90) and middle-aged (P240 d or older) D3KO/FINDT3 mice is significantly lower than in WT/FINDT3 mice (Supplemental Fig. 2). The serum T₄ level remains similarly low in young and middle-aged D3KO/FINDT3 mice (Fig. 1A). The serum level of T₃ in P90 D3KO/FINDT3 mice is not significantly different from WT/FINDT3 mice, but in middle-aged and older animals (i.e. >40 d), it was markedly lower than that in WT mice (Fig. 1B).

Although careful analysis of these results indicates that the growth and TH phenotype of D3KO/FINDT3 mice is slightly milder that that observed in an inbred strain (11), the data demonstrate that the basic physiological consequences of D3 deficiency apply to multiple genetic backgrounds.

Increased T₃ signaling in the D3KO/FINDT3 mouse brain in early postnatal life

We analyzed T₃ signaling in the brain of neonatal D3KO/FINDT3 mice by examining the intensity of β-gal staining in brain coronal sections. Consistent with our previous observations showing enhanced T₃ content as determined by RIA and T₃-dependent gene expression in the brain of D3KO neonates (11), we observed an increase in β-gal staining in the brain of P2 through P6 D3KO/FINDT3 mice (i.e. data from P5 brains are shown in Supplemental Fig. 3). This is illustrated in Fig. 2. Comparison of stained P6 brain sections from WT/FINDT3 and D3KO/FINDT3 mice shows that the latter exhibit a significant increase in β-gal staining in the septum and several cortical areas, including the motor, sensory, insular, and piriform cortices (Fig. 2, B and E, vs. A and D). No staining was detectable in animals negative for the FINDT3 transgene (Fig 2, C and F). We have previously shown that in the absence of T₃, β-gal expression is undetectable in FINDT3 mice (16), thus demonstrating that β-gal staining is not due to artifactual expression of the transgene but to the direct result of T₃ action.

Figure 2.

T₃ signaling as determined by β-gal staining in septal and cortical areas of P5 WT/FINDT3 (A and D), D3KO/FINDT3 (B and E), and WT (C and F) neonates. Protocol 2 was used to detect β-gal expression (see Materials and Methods). No staining was detected in the brain of animals not carrying the transgene (C and F). Staining quantification was performed in seven consecutive sections spanning the areas depicted from three different mice (21 sections per group in total). S, Septum; PI, piriform cortex; MCx, motor cortex; SCx, sensory cortex.

Thyroid status and T₃-dependent gene expression in the brain of D3KO/FINDT3 mice at P17 and P21

D3KO mice transition from a state of serum T₃ excess during early postnatal life to a state of low serum T₃ and T₄ later in neonatal life and in adulthood. These low levels of serum T₃ and T₄ translate into reduced TH action in peripheral tissues, as determined by reduced hepatic expression of TH-dependent genes (11). Because the brain still expresses a high level of D3 in later developmental stages and adulthood, especially the cerebrum (15), we examined TH action in the D3KO brain as this transition occurs.

We used Northern blot analysis of two TH-regulated genes (RC3 and hairless) to assess the TH action in the brain. At P17, whole-brain hairless mRNA expression was similar in WT/FINDT3 and D3KO/FINDT3 mice, but the mRNA level of RC3 was significantly lower in D3KO/FINDT3 mice than in WT/FINDT3 animals (Fig. 3A). At P21, low expression of hairless mRNA was observed in the brain of D3KO/FINDT3 mice by in situ hybridization (Fig. 3C). At P21, Northern blot analysis also revealed that the mRNA expression of both RC3 and hairless was decreased in D3KO/FINDT3 mice when compared with WT/FINDT3 animals (Fig. 3B). Decreased β-gal staining is also observed in several areas of the P21 D3KO brain (Fig. 3D). These results indicate that at P15 the brain of FINDT3/D3KO mice is starting to exhibit manifestations of deficient T₃ availability and action, and this situation appears to affect most brain areas at P21.

Figure 3.

Brain expression of TH-regulated genes in late neonatal life. A and B, Northern analysis of RC3 and hairless expression in P17 (A) and P21 (B) WT and D3KO brains. Data are expressed as the ratio of the TH-responsive gene mRNA vs. cyclophilin mRNA cy, Cyclophilin. The number of samples quantified is shown in the bars, which represent the mean ± se. *, P < 0.01 D3KO vs. WT. C, In situ hybridization of hairless mRNA in P21 sections of WT and D3KO brains. Hss, Hairless. (Note: these sections are anatomically equivalent based on analysis of serial sections, with apparent differences secondary to morphological abnormalities of the D3KO brain such as enlargement of the lateral ventricles.) D, β-Gal staining in P21 brain regions of WT and D3KO mice. Staining protocol 1 (see Materials and Methods) was used. MSCx, Motor and sensory cortex; PCx, piriformcotex; CCx, cingulate cortex; SCo, layers of the superior colliculus; St, striatum; Sp, septum; Pag, periaqueductal gray.

Region-specific changes in T₃ signaling in the brain of P15 D3KO/FINDT3 mice

Because these results suggest a transition in the thyroid status of the brain from one of increased to decreased levels of TH action, we determined β-gal activity in homogenates of major areas of the P15 brain. No significant difference in β-gal activity between WT/FINDT3 and D3KO/FINDT3 mice was observed in the cerebral cortex, cerebellum, or midbrain (Supplemental Fig. 4). However, the hypothalamus and the thalamus of D3KO/FINDT3 mice exhibited significantly lower β-gal activity than those of WT/FINDT3 animals, suggesting decreased T₃ action in these regions.

At P15, no significant difference in RC3 mRNA level was observed in the cortex and hypothalamus of D3KO/FINDT3 mice (Supplemental Fig. 5, top panels) when compared with that in WT/FINDT3 animals. However, RC3 expression was increased in the thalamus, midbrain, and cerebellum of D3KO/FINDT3 mice when compared with corresponding brain regions of WT animals (Supplemental Fig. 5, bottom panels), suggesting increased T₃ action in these regions at this stage.

To determine more precisely the spatial distribution of the changes in T₃ action in the P15 D3KO brain, we analyzed the β-gal staining of brain sections from WT/FINDT3 and D3KO/FINDT3 mice. A significant decrease (25–40%) in β-gal staining was observed in areas of the prefrontal cortex (Fig. 4, B vs. A), septal region (Fig. 4, D vs. C), pyramidal cell layer of the anterior hippocampus (Fig. 4, F vs. E), and the anterior hypothalamus (Fig. 4, H vs. G). In contrast, a 2- to 3-fold increase in β-gal staining was observed in thalamic nuclei and the pyramidal cell layer of the posterior hippocampus (Fig. 4, J vs. I), in areas of the posterior hypothalamus (Fig. 4, L vs. K) and in discrete regions of the corpus colliculus and midbrain (Fig. 4, N vs. M). These results indicate that the transition from the high T₃ state to the low T₃ and T₄ state as observed in the serum of D3-deficient animals also occurs in the brain. However, not all brain regions undergo this transition at the same time.

Figure 4.

T₃ availability as determined by β-gal staining in areas of the P15 WT/FINDT3 and D3KO/FINDT3 brains. A and B, Orbital (OCx) and prefrontal cortices. C and D, Septum (S). E and F, Pyramidal cell layer of hippocampus (Py). G and H, nuclei in anterior hypothalamus. 3v, Third ventricle; Pvn, paraventricular nucleus. I and J, Posterior hippocampus and thalamic nuclei. Th, Thalamus. K and L, Nuclei in posterior hypothalamus. M and N, Gray layers of the superior colliculus (SCo) and midbrain nuclei. Staining protocol 2 (see Materials and Methods) was used. The data are representative of one of three different experiments in pairs of P15 WT and D3KO mice. Quantification was performed in five to 13 microphotographs corresponding to the same number of consecutive sections comprising the areas shown. For each area, bars represent the mean ± se of pixels stained in those sections.

Increased T₃ signaling in the brain of P90 and older than P240 D3KO/FINDT3 mice

In young adult animals (P90), some areas of the D3KO/FINDT3 brain exhibit increased T₃ action again, as determined by increased T₃ signaling. Thus, β-gal staining is significantly increased in the neocortex of D3KO/FINDT3 mice compared with that of WT/FINDT3 mice. This increase is 2-fold in the motor cortex (Fig. 5, B vs. A), 5-fold in the sensory cortex (Fig. 5, D vs. C), and 4-fold in the insular cortex (Fig. 5, F vs. E). When using a different protocol for the fixation and staining of tissues (see Materials and Methods), very similar increases in β-gal staining were observed in the same areas (Fig. 5, G–L). The increase in T₃ content in the P90 D3KO brain is primarily observed in areas of the neocortex.

Figure 5.

T₃ signaling as determined by β-gal staining in areas of the P90 WT/FINDT3 and D3KO/FINDT3 cortex. Staining protocol 2 was used in A–F, and staining protocol 1 was used in G–L. Areas depicted are the motor cortex (A and B and G and H), the sensory cortex (C and D and I and J), and the insular cortex (E and F and K and L). Quantification of staining (A–F) in each region (number of pixels) is shown in bars on the right, which represent the mean ± se of staining intensity in six consecutive sections from four different animals per group across each region (a total of 24 sections per group). *, P < 0.01 D3KO vs. WT.

As D3KO/FINDT3 mice age (i.e. > P240), the abnormally elevated TH action in the brains becomes more generalized. In middle-aged mice, increases in β-gal staining can be observed in most areas of the brain, including the prefrontal cortex (Supplemental Fig. 6, B vs. A), the motor and sensory cortices (Supplemental Fig. 6, D vs. C), the hippocampus (Supplemental Fig. 6, F vs. E), and the hypothalamus (Supplemental Fig. 6, H vs. G). This expansion of increased T₃ signaling to virtually all areas of the aging D3KO brain occurs in the face of persistently diminished TH levels in the serum and most peripheral tissues. Taken together, these results indicate that the D3-deficienct brain undergoes in adulthood a second transition from decreased to increased TH action. The successive, general changes in TH action that occur in the D3KO brain throughout the lifespan are followed closely by the levels of hairless mRNA expression in the cerebrum (Fig. 6).

Figure 6.

Expression of hairless at different developmental stages. Expression was determined by quantitative RT-PCR in the cerebrum of WT and D3KO mice of different ages. Data represent the mean ± se of determinations in the number of samples indicated in the bars. *, P < 0.01 WT vs. D3KO.

D3 protects the brain from excessive levels of T₃

We have recently shown that D3-deficient mice exhibit impaired serum T₃ clearance when administered T₃ (11). The results presented above also indicate that even in conditions of normal or low serum levels of TH, D3 deficiency results in excessive T₃ action in the brain of the adult animal. We wanted to determine how D3 deficiency affects brain T₃ signaling in the presence of a supraphysiological concentration of serum T₃. For these experiments, we used young adult animals (P60), expecting no major differences in brain β-gal staining. Half of the experimental animals were administered T₃ at a concentration of 0.25 μg/ml in the drinking water for 7 d. As assessed by β-gal staining in the insular cortex, T₃ signaling was just minimally elevated in untreated D3KO/FINDT3 mice compared with untreated WT/FINDT3 animals (Fig. 7, C vs. A, quantifications in Fig. 7E). After T₃ administration, no increase in staining is observed in WT/FINDT3 mice (Fig. 7, B vs. A). In contrast, D3KO/FINDT3 mice exhibit a significant increase in β-gal staining after T₃ administration (Fig. 7, D vs. C). This different response to T₃ treatment between WT and D3KO mice occurs despite the fact that the serum T₃ levels achieved in both experimental groups are comparable (Fig. 7F) and underscores the importance of D3 in protecting the brain from T₃ overexposure. This is further confirmed by hairless mRNA expression in the cerebrum of these mice. In WT mice, T₃ treatment produced a 1.5-fold increase in hairless mRNA expression (Fig. 7G), whereas a 4.5-fold increase is observed in corresponding D3KO mice.

Figure 7.

Effect of T₃ treatment on brain T₃ signaling and serum T₃. A–D, T₃ signaling as determined by β-gal staining in the P60 insular cortex in WT/FINDT3 and D3KO/FINDT3 animals treated (B and D) or not (A and C) for 7 d with 0.25 μg/ml of T₃ in the drinking water. Staining protocol 2 was used (A–D). E, Quantification of staining, expressed as the mean ± se of staining intensity in 18 sections from three animals (six consecutive sections from each animal). F, Serum T₃. G, Hairless mRNA expression in the cerebrum as determined by quantitative RT-PCR. Data represent the mean ± se of the number of animals indicated in parentheses in each group. *, P < 0.01 treated vs. nontreated groups.

Abnormal distribution of β-gal-expressing cells in the D3KO brain

In some instances, we have observed a disruption in the spatial distribution of cells in the brain of D3KO mice. Thus, in the young adult D3KO brain, this disruption is evident in the cingulate cortex and diagonal band (Fig. 8, A–D). In the motor cortex, the layer of cells expressing the transgene in the D3KO brain is thicker than in the WT brain (Fig. 8, E and F). These observations suggest alterations in brain cytoarchitecture as a consequence of D3 deficiency.

Figure 8.

Abnormal patterns of β-gal staining in certain regions of the P60 D3KO brain. A, and B, Prefrontal cortex. C and D, Septal diagonal band. E and F, Motor cortex. Arrows point to differences in staining patterns (A–D), and white lines of the same length are depicted in E and F to appreciate the thickness of the cell layer stained.

Discussion

D3 inactivates T₄ and T₃, and contributes significantly to TH clearance. D3 is highly expressed in developing tissues, and we have shown that constitutive D3 deficiency leads to overexposure to TH during perinatal life (11) and to low serum levels of thyroid hormones in adulthood that is the consequence of an altered set point of the thyroid axis (12). Consistent with these published results, D3 deficiency in the mixed genetic background used in the present studies also leads to neonatal thyrotoxicosis and persistent low TH levels in adulthood. It must be noted, however, that this phenotype is slightly milder in this background, as evidenced in early adulthood by a normal serum T₃ level and only a 50% reduction in serum T₄ as opposed to 20 and 70% reductions, respectively, in the serum levels of these hormones previously described in an inbred 129/Sv genetic background (11) at the same age.

In contrast to most tissues, in which the D3 is expressed primarily during development, the brain exhibits high D3 expression throughout life, an observation that suggests a role for D3 in modulating TH action in neural tissue across the life span. Because serum TH levels in the D3KO mouse change with the developmental stage, several questions relevant to brain function arise: what is the TH status of the D3KO brain at any given point in time; does it correlate with serum levels of TH; does it relate to the pattern of D3 expression?

We addressed these questions by analyzing the expression of the FINDT3 transgene as well as that of TH-dependent genes. The brain expression of the FINDT3 is not based on endogenous TH receptors. However, it predominates in neurons and closely correlates with the level of T₃ in brain tissue (16), the latter parameter being a key factor in determining TH signaling.

Consistent with the marked elevation of serum T₃ level in D3KO animals, we observed an increase in β-gal expression in the early postnatal brain of D3KO/FINDT3 mice, indicative of increased TH availability. However, later in neonatal life, the TH status of the D3KO brain follows the changes in the serum levels of TH and undergoes in most regions a transition from excessive to diminished TH action. This transition occurs, even though the clearance of T₃ in the brain (and presumably in other tissues) is seriously compromised by the lack of D3 activity and probably reflects a reactive suppression of the thyroid axis from the prior thyrotoxic state. Thus, as early as P15, the previously thyrotoxic D3KO brain initiates a transition to low T₃ availability, which applies to most brain regions after weaning. This is probably due to the decreasing levels of T₃ and T₄ in the serum because brain cells use both serum T₃ and serum T₄ (via local conversion to T₃ by the D2) for their T₃ needs. Particularly important is the observation that this transition does not occur simultaneously across the brain. This is best illustrated by the occurrence of elevated and diminished TH action in different areas of the same P15 D3KO brain (Fig. 4) and probably reflects region-specific differences in the complements of factors involved in TH homeostasis.

In this context, the dynamics of the FINDT3 system may be of importance. Specifically, the β-gal protein is relatively stable in neurons (16). Thus, the transition from a state of elevated to decreased T₃ availability may not be immediately reflected in a decrease in β-gal staining. Under these circumstances, the expression levels or endogenous TH-dependent genes may reflect changes in TH action in a more timely manner.

Later the D3KO brain undergoes a second, slower transition back to a state of apparent enhanced TH action, which coincides with an increase in serum TH levels toward normal levels, which occurs after weaning. Thus, in young adult mice (P90), increased β-gal staining is observed in D3KO brains compared with those of WT animals in areas of the neocortex, whereas other brain regions exhibit unchanged or low staining. This increase in T₃ signaling becomes widespread across the brain as the animals age, even though diminished serum TH levels persist. The transition of the adult D3KO cortex to a state of apparent thyrotoxicosis occurs earlier than in other brain regions, an observation that is consistent with the higher level of D3 expression in that area of the normal adult brain (15). Furthermore, under a limited thyrotoxicosis challenge achieved by exogenous administration of T₃, the WT brain appears to remain close to a euthyroid state, in contrast to the overt thyrotoxicosis of the D3KO brain, demonstrating an important protective effect of D3 on brain tissue in this state.

Taken together, our observations highlight the importance of D3 in the TH homeostasis of the CNS, and its specificity in protecting brain regions from thyrotoxicosis across the life span. As we continue to learn about the detrimental consequences of D3 deficiency for brain structure and function, it is important to be mindful of the dynamic changes observed in the TH status of the D3KO brain. Abnormalities in neural function in the D3KO brain may result from: 1) thyrotoxicosis early in development that has developmental consequences; 2) decreased TH action in late neonatal life; 3) thyrotoxicosis in adulthood; or 4) a combination of the above. For instance, the life-long persistence of low TH levels in D3KO mice and the abnormal expression of TRH in the hypothalamus (12) may be due to hypothalamic defects caused by developmental thyrotoxicosis in early life, improper local control of T₃ availability in the adult hypothalamus, or both. Future studies may use these and additional mouse models of conditional D3 deficiency to determine how defects in specific brain functions in the D3KO mouse correlate with altered TH signaling in particular brain regions.

In summary, we have shown that D3 plays an important role throughout the life span in regulating TH availability and action in the brain in a region-specific manner. This is critical to understand the etiology of the brain dysfunction caused by D3 deficiency. In addition, our results demonstrate that when D3 activity is impaired, the CNS may evidence effects of enhanced T₃ action in the face of normal or even low levels of circulating TH. Given that certain chemical compounds and endocrine disruptors might influence D3 activity by either inhibiting the enzyme or altering the epigenetic regulation of the D3 gene, our results raise the possibility that environmental influences could alter brain function in the absence of significant changes of serum TH levels. This concept may have important clinical implications if the brain dysfunction observed in the D3KO mouse apply to humans. Specifically, alterations in D3 activity and TH signaling in certain neurological disorders might occur and contribute to symptoms in the absence of altered serum TH levels. Additional studies are needed to explore how D3 deficiency, in particular regions of the CNS, influences specific brain functions.