|Home | About | Journals | Submit | Contact Us | Français|
Background: Thyroid hormone (TH) plays a key role in the developing brain, including the cerebellum. TH deficiency induces organizational changes of the cerebellum, causing cerebellar ataxia. However, the mechanisms causing these abnormalities are poorly understood. Various animal models have been used to study the mechanism. Lacking dual oxidase (DUOX) and its maturation factor (DUOXA) are major inducers of congenital hypothyroidism. Thus, this study examined the organizational changes of the cerebellum using knockout mice of the Duoxa gene (Duoxa–/–).
Methods: The morphological, behavioral, and electrophysiological changes were analyzed in wild type (Wt) and Duoxa-deficient (Duoxa–/–) mice from postnatal day (P) 10 to P30. To detect the changes in the expression levels of presynaptic proteins, Western blot analysis was performed.
Results: The proliferation and migration of granule cells was delayed after P15 in Duoxa–/– mice. However, these changes disappeared by P25. Although the cerebellar structure of Duoxa–/– mice was not significantly different from that of Wt mice at P25, motor coordination was impaired. It was also found that the amplitude of paired-pulse facilitation at parallel fiber–Purkinje cell synapses decreased in Duoxa–/– mice, particularly at P15. There were no differences between expression levels of presynaptic proteins regulating neurotransmitter release at P25.
Conclusions: These results indicate that the anatomical catch-up growth of the cerebellum did not normalize its function because of the disturbance of neuronal circuits by the combined effect of hypothyroidism and functional disruption of the DUOX/DUOXA complex.
Thyroid hormone (TH) is essential for the normal development of the brain (1). TH deficiency during the postnatal period causes congenital hypothyroidism in humans. The typical findings in cretinism are mental retardation, ataxia, and deafness, together with impaired body growth (2,3). TH deficiency in developing rodents induces organizational changes of the cerebellum, causing cerebellar ataxia (4). However, the mechanisms causing this abnormality are poorly understood.
Perinatal hypothyroidism causes various anatomical changes of the rat cerebellum such as reduction of growth and dendritic arborization of Purkinje cells (5,6), reduction of synaptogenesis between Purkinje and granule cells (6,7), delayed myelination (8), and changes in synaptic connections between cerebellar neurons and afferent neuronal fibers (9). These aberrant phenotypes can be rescued only if TH replacement is instituted by postnatal week 2 in rats (5).
Some of the rate-limiting steps of TH synthesis take place at the apical cell membrane of the thyroid follicular cell. To generate hydrogen peroxide (H2O2), dual oxidase (DUOX) 2 requires the maturation factor (DUOXA) (10,11). To iodinate thyroglobulin (TG), thyroid peroxidase (TPO) requires H2O2 as the final electron acceptor in two steps (12–14). The first step is oxidation of iodide, which is then covalently linked to selected tyrosine residues of TG (12–14). The second step is coupling of two iodotyrosine residues to form iodothyronines (12–14). A defect of DUOX2 or DUOXA2 decreases TH synthesis, causing congenital hypothyroidism in both humans and mice (15–17).
To evaluate the effect of dyshormonogenesis caused by the absence of DUOXA in cerebellar development, the study used Duoxa gene knockout mice (Duoxa–/–), in which part of both Duoxa1 and Duoxa2 exons were deleted (10). Duoxa+/– mice were used as dams. Since the TH status of Duoxa+/– mice is euthyroid (10), the involvement of an altered maternal TH status on fetal or neonatal development can be excluded. Duoxa–/– mice showed delayed disappearance of the external granule cell layer (EGL) followed by a catch-up cerebellar growth, whereas no morphological alteration of Purkinje cells was observed. An electrophysiological study revealed that the amplitude of paired-pulse facilitation at parallel fiber–Purkinje cell synapses decreased at postnatal week 2 but was improved at postnatal week 3. On the other hand, the function of climbing fiber–Purkinje cell synapses was not affected in Duoxa–/– mice. Nevertheless, a severe motor coordination defect due to cerebellar ataxia was seen even at postnatal week 3. Daily thyroxine (T4) replacement reversed aberrant growth and motor coordination in Duoxa–/– mice. These data indicate that Duoxa regulates cerebellar development mainly through the regulation of TH synthesis. It should be noted, however, unlike previous drug-induced congenital hypothyroid models, that alterations of Purkinje cell morphology and electrophysiological properties of climbing fiber–Purkinje cell synapses were not seen. Thus, the cerebellar dysfunction in Duoxa–/– mice may be caused by the combined effect of congenital hypothyroidism and functional disruption of the DUOX/DUOXA complex in the brain.
All experiments were performed in accordance with the guidelines and protocols approved by the Animal Care and Experimentation Committee, Gunma University. All efforts were made to minimize the suffering and number of animals used in this study.
Duoxa–/– mice were generated as described previously (10). Animals were bred in the Animal Facility of Gunma University Graduate School of Medicine. All mice were housed with food and water provided ad libitum under controlled temperature (25±5°C), humidity, and illumination (12:12 light–dark cycle; lights on at 7:00 a.m.) conditions. Adult heterozygotes (Duoxa+/–) were mated for a suitable reproduction period. Between one and three pups per litter were used in each genotype. Pups from at least three dams were used in each experiment. Some mice received daily subcutaneous injection of 20μg/kg body weight (BW) levothyroxine (LT4 sodium salt pentahydrate; Sigma, St, Louis, MO) starting at postnatal day (P) 3. This amount is sufficient to correct the hypothyroidism of Duoxa–/– mice (18). Pups were weighed and killed at P10, P15, and P25 by decapitation after induction of anesthesia with ketamine/xylazine (22.5mg/mL ketamine and 1mg/mL xylazine in 0.9% NaCl, intraperitoneally; 5mL/kg BW). Their tails were cut to extract DNA to determine their genotype by polymerase chain reaction (PCR) in accordance with the protocol described previously (10). The vermis of the cerebellum was dissected out, rapidly frozen in liquid nitrogen, and stored at −80°C for Western blot analysis. For histological analysis, some brains were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) before dissection and cryoprotected in 10%, 20%, and 30% sucrose in PBS (pH 7.4) at 4°C. Cerebella were embedded in the O.C.T. compound (Sakura, Tokyo, Japan), cut at a thickness of 10–20μm using a cryostat (Microm HM 525; Thermo Scientific, Walthman, MA), and placed on slides for staining with cresyl violet or immunohistochemistry (see below). Tissues were analyzed under a BZ-9000 microscope (Keyence, Osaka, Japan).
Total RNA was extracted at P25 from the cerebellar vermis, cortex, liver, and thyroid using RNeasy (Qiagen, Hilden, Germany). A total of 2μg of total RNA was used for cDNA synthesis using a High Capacity RNA-to-cDNA kit (Applied Biosystems, Foster City, CA). Specific primers for Duox1, Duox2, Duoxa1, Duoxa2, brain-derived neurotrophic factor (Bdnf), hairless (Hr), Krüppel-like factor 9 (Klf9), deiodinase iodothyronine type I (Dio1), nuclear receptor subfamily 1, group D, member 1 (Nr1d1), and glyceraldehyde 3-phosphate dehydrogenase (Gapdh) as internal controls were used (Duox1: Mm01328685_m1; Duox2: Mm01326247_m1; Duoxa1: Mm01269312_m1; Duoxa2: Mm01275445_m1; Bdnf: Mm04230607_s1; Hr: Mm00498963_m1; Klf9: Mm00495172_m1; Dio1: Mm00839358_m1; Nr1d1: Mm00520708_m1; and Gapdh: Mm99999915_g1; Applied Biosystems). The context sequences for all assays are shown in Supplementary Table S1 (Supplementary Data are available online at www.liebertpub.com/thy). Real-time reverse transcription PCR (RT-PCR) was carried out as described in the instruction manual of the TaqMan Fast Advanced Master mix kit (Applied Biosystems) and the StepOne Real-Time PCR System (Applied Biosystems). The real-time RT-PCR protocol for all genes was 95°C for 20sec, followed by amplification using 95°C for 1sec and 60°C for 20sec (40–80 cycles). Levels of mRNA shown above were normalized by Gapdh mRNA level.
The cerebellar sections were washed with PBS and incubated with 5% normal horse serum in 0.3% Triton X-100 in PBS for 30min at room temperature. Then, the sections were incubated with a mouse anti-calbindin-D28K antibody (1:3000; Sigma) overnight at 4°C. Next, the sections were incubated with biotinylated horse anti-mouse immunoglobulin G (IgG; 1:200 dilutions) for 45min. After rinsing in PBS, the sections were incubated with avidin-biotin complex for 1h and visualized by 3,3′-diaminobenzidine (0.5mg/mL Tris-HCl containing 0.01% H2O2). Sections were then immersed in 0.5% cresyl violet for 1–3min, dehydrated by graded series of ethanol, cleared in xylene, and placed under coverslips.
Cerebellar areas were measured in cresyl violet-stained midsagittal sections using ImageJ software v1.47 (NIH, Bethesda, MD). All areas of the EGL and molecular layer in one section were measured. Quantitative data were obtained from three sections per mouse, and four mice per genotype and age group were evaluated. The morphology of Purkinje cells was analyzed using midsagittal sections immunostained for calbindin with cresly violet staining. The lengths of the cell body to the first branch point and the sizes of the soma were measured. Quantitative data were obtained from seven to eight cells from one section per mouse, and four mice per genotype were evaluated.
The serum thyrotropin (TSH) concentration was determined by enzyme-linked immunosorbent assay (Rodent Thyroid Stimulating Hormone ELISA test kit ERKR7015; Endocrine Technologies, Inc., Newark, CA). Samples were measured in triplicate. Free T4 (fT4) and free triiodothyronine (fT3) concentrations were measured on a single automated analyzer using the ARCHITECT system (Abbott ARCHITECT i2000; Abbott Laboratories, Maidenhead, United Kingdom). The lowest detectable limits were 0.40ng/dL and 1.00pg/mL for fT4 and fT3, respectively. Serum samples from Duoxa–/– mice for determining fT4 and fT3 concentrations were collected in one tube from four mice, and five tubes were used for measurement. Twenty Duoxa–/– mice from 10 different dams were used for detection of TH status. Sensitivity threshold values were fitted when the concentration was below the sensitivity limits.
The accelerating rotarod test (model LE8500; Panlab, Barcelona, Spain) was performed at P25 to assess motor coordination. Five rotarod trials with increasing speed from 4rpm to 40rpm in 5min were performed. The time on the rotarod in the last three trials was recorded. Average scores were used for analysis.
All mice were tested individually in an open field apparatus at P25. The unit for detecting the motor activity in mice consisted of a 45cm×45cm×20cm frame containing a total of 16×16 crossed infrared beams at intervals of 2.5cm, located on the sides (LE 8811; Panlab, S.L.U., Barcelona, Spain). Locomotor activity was assessed on the basis of the analysis of the position and frequency with which the mice crossed the infrared beams. Data were analyzed using the Acti-Track program (Panlab, S.L.U.). Locomotor activity was monitored for 30min.
Experiments were conducted using Wt and Duoxa–/– mice aged P10–P30. The methods used for the preparation of thin slices and patch-clamp recordings from visually identified Purkinje cells in cerebellar slices were similar to those described previously (19). Briefly, whole-cell patch-clamp recordings of Purkinje cells were performed under an upright microscope (Axioscope; Zeiss, Oberkochen, Germany). The slices were superfused with an external solution containing 120mM of NaCl, 2.5mM of KCl, 2mM of CaCl2, 1mM of MgCl2, 26mM of NaHCO3, 1.25mM of NaH2PO4, and 25mM of glucose, bubbled with 95% O2 and 5% CO2. The solution was adjusted to pH 7.4 with HCl and bubbled with 100% O2. The pipette solution contained 150mM of Cs gluconate, 8mM of NaCl, 2mM of MgATP, 10mM of HEPES, and 0.1mM of spermine, and was adjusted to pH 7.2 with gluconic acid. Patch pipettes had resistances of 3–5MΩ when filled with the pipette solution. Series resistance (5–10MΩ) was routinely compensated by 70%. To evoke climbing/parallel fiber–excitatory postsynaptic currents (EPSCs) in Purkinje cells, square pulses (0.1ms, 10–100μA) were applied through a tungsten concentric bipolar electrode. The signals were filtered at 3kHz and digitized at 10kHz for EPSCs. Current responses were recorded with an EPC-8 amplifier (HEKA, Darmstadt, Germany), and a pCLAMP system (v7, Axon Instruments, Foster City, CA) was used for data acquisition and analysis. During ESPC recording, 100μM of picrotoxin was added to the external solution. All experiments were performed at 30–32°C. The changes in the amplitude and decay time constant (19) of EPSCs in climbing/parallel fiber–Purkinje cell synapses were also examined. The amplitudes of paired-pulse facilitation and depression were calculated using the peak amplitude of each EPSC (20). The decay time constant (τw) of ESPCs was calculated as previously reported (19), and 100nM cyclothiazide (CTZ) was applied to prevent the desensitization of the α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA)-type glutamate receptor (21).
The dissected cerebella were homogenized in lysis buffer containing 10mM of Tris-HCl (pH 7.8), 150mM of NaCl, 1mM of EDTA, 1% Nonidet P40, and protease inhibitors. The expression levels of postsynaptic membrane proteins (mGluR4, Munc18-1, calmodulin, syntaxin1a, VAMP2, and SNAP25) were extracted from the vermis using a ProteoExtract native membrane protein extraction kit (Calbiochem, San Diego, CA) in accordance with the manufacturer's instruction. The expression levels of proteins were determined by Western blot analysis, as described previously (22,23). Briefly, after boiling for 5min, protein samples (30μg) were subjected to 15% SDS-polyacrylamide gel electrophoresis, and separated products were transferred to nitrocellulose membranes. The membranes were blocked with 5% nonfat dry milk in Tris-buffered saline containing 0.1% Tween 20, followed by overnight incubation with the appropriate diluted primary antibodies for mGluR4 (1:1000; Millipore, Billerica, MA), Munc18-1 (1:1000; Cell Signaling, Danvers, MA), calmodulin (1:1000; Cell Signaling), syntaxin1a (1:10000; Abcam, Cambridge, United Kingdom), VAMP2 (1:1000; Cell Signaling), SNAP25 (1:1000; Cell Signaling) and β-actin (1:1000; Cell Signaling). After washing with Tris-buffered saline containing 0.1% Tween 20, the membranes were incubated with a horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody for 1h at room temperature and detected using an ECL detection system. β-actin was used as a loading control.
Statistical comparisons were performed by Student's t-test or analyzed by one-, two-, or three-way analysis of variance (ANOVA), followed by the Tukey's test (after one-way ANOVA) or the Bonferroni test (after two- and three-way ANOVA) for post hoc analysis. Differences were considered significant at p<0.05. All values are presented as mean±standard error of the mean. GraphPad Prism v6.0 for Windows (GraphPad Software, San Diego, CA) or SPSS Statistics for Windows v22 (IBM Corp., Armonk, NY) were used for statistical analysis and graphing data.
The body weight of Duoxa–/– mice during postnatal weeks 2–3 tended to be lower compared with that of Wt mice (Table 1). A two-way ANOVA showed significant effects of genotype (male: F(2, 69)=39.9, p<0.001; female: F(2, 80)=38.1, p<0.001) and age (male: F(2, 69)=71.0, p<0.001; female: F(2, 80)=51.3; p<0.001). Post hoc analysis by Bonferroni test showed that the body weights of male and female Duoxa–/– mice were significantly lower compared with those of Wt mice at P25 (male: t20=11.5, p<0.001; female: t27=11.2, p<0.001). The increase in body weight of T4-replaced Duoxa–/– mice was similar to that of Wt mice. These findings are consistent with those in a previous report (10).
As shown in Table 2, a one-way ANOVA showed no significant differences between three groups for the whole brain weight. However, a one-way ANOVA showed significant differences for the cerebellar weight (male: F(2, 12)=21.4, p<0.001; female: F(2, 15)=3.18; p<0.05). Post hoc analysis by Tukey's test indicated that cerebellar weights were significantly lower in Duoxa–/– male mice compared with that of Wt male mice (p<0.01). Thus, the cerebellum may be more severely affected than other brain regions. Interestingly, T4 replacement could not fully rescue the decrease in cerebellar weight in Duoxa–/– mice.
To confirm the presence of hypothyroidism in Duoxa–/– mice, serum TSH, fT3, and fT4 concentrations were determined. A one-way ANOVA showed significant differences in TSH and fT4 (TSH: F(2, 27)=21.4, p<0.001; fT4: F(2, 15)=6.59, p<0.01) but not in fT3 (F(2, 15)=3.59, p=0.06). Post hoc analysis by Tukey's test showed that Duoxa–/– mice had significantly elevated TSH concentrations (p<0.001), indicating that the mice were hypothyroid at P25 (Table 3). The serum fT4 concentration in two of five Duoxa–/– groups and the fT3 concentration in all Duoxa–/– mice were below the detectable range. On the other hand, no significant differences in hormone levels between Wt and T4-replaced mice were observed.
The expression levels of the Duox/Duoxa family members vary among tissues (10). The expression levels of these mRNAs in the cerebellum have not yet been studied. Low expression levels of the Duox/Duoxa family mRNAs were found in the Wt mice cerebellum compared with those in the thyroid (Supplementary Fig. S1). On the other hand, the expression level of Duox1 was upregulated in Duoxa–/– mice cerebellum, as determined by Student's t-test (p<0.001; Supplementary Fig. S1). Thus, the expression pattern of Duox1 and Duox2 mRNAs in Duoxa–/– mice was different compared with other extrathyroidal tissues such as the colon (10).
Several TH-responsive genes are expressed in the brain. Among such genes, Bdnf plays an important role in cerebellar development (18). Hr is also well known as a direct target of TH in brain (24). Both Bdnf and Hr mRNA expressions were lower in both the cerebellum and the cortex of Duoxa–/– mice, as determined by Student's t-test (cerebellum: Bdnf p<0.05, Hr p<0.001; cortex: Bdnf p<0.05, Hr p<0.001; Supplementary Fig. S2). Dio1, which converts T4 to T3, is highly expressed in the liver, and its expression is regulated by TH (25,26). The Dio1 mRNA level in the liver was extremely low in Duoxa–/– mice, as determined by Student's t-test (p<0.05; Supplementary Fig. S2). On the other hand, the expression levels of Klf9 and Nr1d1, which are also known as a TH-responsive gene, were not significantly changed when evaluated by Student's t-test (25,27). These results indicate that Duoxa–/– mice are generally hypothyroid in both the brain and the liver. Klf9 and Nr1d1 might be regulated by not only TH but also other factors in Duoxa–/– mice.
The rotarod test was performed at P25 to examine motor coordination. As shown in Figure 1A, a one-way ANOVA showed significant effects on the duration of remaining on the rotarod (male: F(2, 19)=27.9, p<0.001; female: F(2, 23)=34.8, p<0.001). Post hoc analysis by Tukey's test indicated that Duoxa–/– mice had a significantly shorter remaining time than Wt and T4-replaced Duoxa–/– mice (p<0.001; Fig. 1A). The open field test was also performed for 30min to measure the locomotor activity in a novel environment (Fig. 1B). The activity was not significantly different between three groups, as determined by a one-way ANOVA (male: F(2, 11)=1.37, p=0.29; female: F(2, 17)=1.25, p=0.31).
In normal mice, the EGL usually disappears by P15. As shown in Figure 2, the EGL persisted at P15 in Duoxa–/– mice, indicating delayed proliferation and migration of granule cells. The EGL disappeared by P25 (Fig. 2A and B). The branching of dendrites and the sizes of the somas of Purkinje cells at P15 were not significantly different in Duoxa–/– compared to those of Wt mice (Fig. 3B; Student's t-test, p=0.52). As reported previously, the length of dendrites from the soma to the first branch is shorter in drug-induced hypothyroid mice (18). However, in the present study, the length was not statistically different between Wt and Duoxa–/– mice (Fig. 3B; Student's t-test, p=0.32).
To evaluate the synaptic function in the cerebellum, an electrophysiological study was performed on parallel fiber–Purkinje cell synapses and climbing fiber–Purkinje cell synapses. When a pair of pulses was delivered in close succession, EPSCs evoked by the second pulse tended to increase compared with EPSCs evoked by the first pulse in parallel fiber–Purkinje cell synapses. Such a phenomenon is called paired-pulse facilitation, which reflects presynaptic plasticity (20). As shown in Figure 4B, the paired-pulse facilitation was recorded in both Wt and Duoxa–/– mice. The ratio of the second response over the first response was lower in Duoxa–/– mice than that of Wt mice after P14 (Fig. 4D; two-way ANOVA: group×interval between two pulses, F(5, 252)=7.99; p<0.001). Interestingly, significant differences were detected at a short interval between two pulses of 20msec (Bonferroni test, t43=6.69; p<0.001), 50msec (t43=6.79; p<0.001), and 100msec (t42=8.48, p<0.001). These findings indicate that the presynaptic plasticity at parallel fiber–Purkinje cell synapses was altered in Duoxa–/– mice.
The plasticity of climbing fiber–Purkinje cell synapses was also examined, which usually show depression by two pulses delivered in close succession (paired-pulse depression). As shown in Figure 5B, traces of EPSCs were not significantly different between Wt and Duoxa–/– mice. The short-term plasticity (Fig. 5C; 50msec interval) and decay time of EPSCs (Fig. 5D) were also not significantly different between the two groups. CTZ, a positive allosteric modulator of AMPA-type glutamate receptors (21), was also applied to reduce the effect of desensitization of these receptors on the decay time constant of EPSCs from climbing fiber–Purkinje cell synapses (21,28). However, the decay time constant was also not significantly different between the two groups (Fig. 5D). These findings indicate that the function of climbing fiber–Purkinje cell synapses was not affected in Duoxa–/– mice.
Duoxa–/– mice showed attenuated paired-pulse facilitation at parallel fiber–Purkinje cell synapses, indicating alteration of presynaptic function. Thus, the expression levels of presynaptic proteins were examined by Western blot analysis. A large number of proteins are involved in presynaptic function. The presynaptic group III metabotropic glutamate receptor (mGluR) 4 is expressed in the presynaptic active zone and reduces synaptic vesicle release. Munc18-1 and calmodulin are proteins essential for neurotransmission, and they promote soluble NSF attachment protein receptor (SNARE)-mediated vesicle fusion; for a review, see Mochida (29). The expression levels of these representative proteins were examined.
The expression levels of presynaptic proteins were analyzed by a three-way ANOVA (with age, genotype, and sex as factors). A three-way ANOVA of the expression levels of mGluR4, Munc18-1, VAMP2, and SNAP25 showed significant effects of age (mGluR4: F(2, 31)=3.61, p<0.05; Munc18-1: F(2, 31)=3.58, p<0.05; VAMP2: F(2, 31)=4.96, p<0.05; SNAP25: F(2, 31)=5.65, p<0.01). However, there were no differences in calmodulin and Syntaxin1a (calmodulin: F(2, 31)=0.92, p=0.41; Syntaxin1a: F(2, 31)=1.83, p=0.18). There were no significant differences by Bonferroni test post hoc analysis.
A three-way ANOVA did not show significant effects of sex for the studied proteins, with the exception of SNAP25 (mGluR4: F(1, 31)=2.77, p=0.11; Munc18-1: F(1, 31)=3.00, p=0.09; calmodulin: F(1, 31)=0.98, p=0.33; Syntaxin1a: F(1, 31)=0.58, p=0.45; VAMP2: F(1, 31)=2.79, p=0.11; SNAP25: F(1, 31)=15.0; p<0.01). However, post hoc analysis by a Bonferroni test showed no significant differences.
A three-way ANOVA of the expression levels of presynaptic proteins did not show differences between the genotypes (mGluR4: F(1, 31)=1.59, p=0.22; Munc18-1: F(1, 31)=1.58, p=0.22; calmodulin: F(1, 31)=0.10, p=0.75; Syntaxin1a: F(1, 31)=0.07, p=0.80; VAMP2: F(1, 31)=1.95, p=0.17; SNAP25: F(1, 31), p=0.93).
Even though protein expression levels at the presynaptic region of the cerebellum in Duoxa–/– mice before P15 was not consistent with the alterations of electrical physiological properties (Figs. 4 and and6),6), it is concluded that the presynaptic function in the developing cerebellum is impaired in Duoxa–/– mice.
Duoxa–/– mice showed various morphological changes (e.g., retardation of granule cell migration; Fig. 2). Motor coordination function was also disturbed (Fig. 1) and parallel fiber–Purkinje cell synaptic function (Fig. 4) was altered. Interestingly, the abnormal phenotype in this mouse is not entirely consistent with the findings in other congenital hypothyroid animal models (e.g., Purkinje cell morphology, brain size; Fig. 3 and Table 2). Thus, Duoxa–/– mice may be a new animal model to study the combined effect of congenital hypothyroidism and DUOXA functional disruption. However, the effect of DUOX/DUOXA on cerebellar development may be smaller than that of TH because of limited expression of Duox/Duoxa in the brain (Supplementary Fig. S1).
The morphogenesis of the rodent cerebellar cortex is completed in the early postnatal period after termination of external granule cell migration, neuronal and glial growth, and synaptogenesis. Multiple factors regulating this process are directly or indirectly regulated by TH. The Purkinje cells receive excitatory inputs from parallel fibers of granule cells and climbing fibers from the inferior olive. Parallel fiber–Purkinje cell synaptic connection is established around P12 during granule cell migration. The climbing fiber–Purkinje cell synaptic connection is established much earlier (around P3) (30). Because of such periodic differences, TH may affect them through different pathways, causing different outcomes. The parallel fibers convey the conditioned stimuli through multi synapses, while the climbing fibers convey the unconditioned stimuli through a single synapse (31). The parallel fiber–Purkinje cell synapse may substantially contribute to cerebellar motor learning (32,33). Thus, the disruption of parallel fiber–Purkinje cell synapses (retardation of granule cell maturation) impairs the motor coordination function in Duoxa–/– mice.
To study the effect of TH on cerebellar development, several rodent models have been generated. Mice or rats can be rendered hypothyroid by administering anti-thyroid drugs such as propylthiouracil (PTU) or methylmercaptoimidazole through food or drinking water (5,18). These drugs block TH synthesis by inhibiting TPO, which oxidizes iodide and facilitates the iodination of tyrosine residues on TG (18). When drugs are administered during pregnancy or lactation, the fetus and newborn are rendered hypothyroid. However, such treatment induces hypothyroidism in both pups and dams. Maternal hypothyroidism may alter nursing behavior, which may affect the development of the pups. On the other hand, because Duoxa+/– mice are not hypothyroid (10), the effect of altered maternal behavior can be eliminated. In this regard, these mice can be a good model for studying the effects of neonatal hypothyroidism.
Using drug-induced hypothyroid rats, Nicholson and Altman showed a reduced cerebellar weight, a prolonged cell proliferation in the EGL and retarded EGL disappearance, a retarded cell differentiation in the molecular and internal granule cell layers, terminal increases in the numbers of granule cells and astrocytes, and a decrease in those of basket cells (34). Legrand (5) and Clos et al. (35) showed a decreased dendritic arborization of Purkinje cells with twofold longer primary dendrites, reduction of growth and branching of dendrites, and shorter parallel fibers with fewer synaptic contacts with Purkinje cells. In Duoxa–/– mice, a retardation of the EGL disappearance was also found in this study (Fig. 2). However, the cerebellar catch-up growth motor dysfunction was still detected on the rotarod at P25 when the EGL no longer exists (Fig. 1A), indicating a functional defect of the cerebellum, although an influence of the body weight difference between the control and Duoxa–/– mice cannot be excluded.
In the open field test, Duoxa–/– mice showed similar travel distances as Wt mice (Fig. 1B). In previous studies, using drug-induced hypothyroid or mutant rats, animals tended to be hyperactive in the open field test (36,37). The exact reason causing such differences is not clear. While these differences may be caused by species differences, there is a possibility that the phenotype of Duoxa–/– mice may be partly unique compared with those of other hypothyroid animals.
Parallel fiber–Purkinje cell synapse impairment causes motor coordination and motor learning dysfunctions, which have been studied by electrophysiology (38,39). However, although a motor coordination impairment has also been reported in hypothyroid animals (36,40), electrophysiological analyses have rarely been conducted. Only one study has shown a delay of the evolution of excitations between parallel fibers and Purkinje cells, especially at P10–P20, and a slight disturbance of the climbing fiber responses of Purkinje cells after P9, with recovery by P30 (41). In the present study, the short-term plasticity of parallel fiber–Purkinje cell synapses was affected in Duoxa–/– mice (Fig. 4). On the other hand, the short-term plasticity of climbing fiber–Purkinje cell synapses was not affected (Fig. 5). These findings indicate that the release of neurotransmitters at parallel fiber–Purkinje cell synapses is disrupted, whereas the effects on climbing fiber–Purkinje cell synapses are negligible in Duoxa–/– mice. This is a unique feature of Duoxa–/– mice compared with other rodent hypothyroid models. Although the exact reason for this difference is not clear, Duoxa–/– may have a different phenotype compared with other mutants, since no aberrant Purkinje cell morphology was observed (Fig. 3). Further study is required to clarify these points.
To clarify the mechanism underlying the change in electrophysiological properties further, the expression of presynaptic proteins was measured. Proteins measured in the present study play crucial roles in the depression of synaptic transmission (mGluR4) (42–46) and the exocytosis of neurotransmitters (Munc18-1, calmodulin, syntaxin1a, VAMP2, and SNAP25) (29,47–49). The findings indicate that the change in electrophysiological properties is not be caused by an altered expression of these presynaptic proteins, which could cause the disruption of neurotransmitter release. In rats with drug-induced hypothyroidism, on the other hand, the numbers of granule cells and parallel fiber–Purkinje cell synapses are decreased (50–52). Taken together, the decrease in the neural transmission efficiency in Duoxa–/– mice is not caused by a change in synaptic property, but by alterations of neuronal circuits (synaptogenesis) between parallel fibers and Purkinje cells.
In the present and one previous study (10), Duoxa–/– mice showed severe hypothyroidism. However, although the phenotype of cerebellar dysfunction in Duoxa–/– was similar to those of other congenital hypothyroid animal models, several differences could be observed. For example, the rdw/rdw rat, a thyroid dyshormonogenesis animal caused by a mutation in the Tg gene, shows not only a severe motor coordination impairment and retarded migration of granule cells, but also small somata and poor dendritic arborization of Purkinje cells (36). In contrast, Duoxa–/– mice did not show aberrant Purkinje cell morphology. Furthermore, the short-term plasticity of climbing fiber–Purkinje cell synapses was not affected. Nevertheless, the mice showed a similar motor coordination defect on the rotarod, as did the rdw/rdw rats. Such a difference may not only be caused by neonatal hypothyroidism. Rather, DUOX/DUOXA may be directly involved in brain development. DUOXA1/2 are members of the NADPH oxidase (NOX) protein families and play a critical role in the production of reactive oxygen species (ROS) (53). Coyoy et al. identified specific patterns of the expression levels of other members of the family (such as NOX1, NOX2, and NOX4) and ROS generation during various stages of rat cerebellar development (54). Damiano et al. reported that DUOX1/2 protein levels were increased by platelet-derived growth factor through production of ROS (55). Based on these findings, one may hypothesize that Duox/Duoxa may help the formation and maintenance of synapses in the neuron. However, the physiological functions of the NOX/DUOX family are still unclear (56). In the present study, since we cannot fully explain the mechanism inducing the cerebellar phenotype in Duoxa-/- mice, there is a possibility that some phenotypic features may be caused by a direct effect of the functional disruption of the DUOX/DUOXA complex. Further studies are required to clarify the role of NOX/DUOX in brain development.
In conclusion, Duoxa deficiency induces developmental hypothyroidism and cerebellar ataxia. Interestingly, although the morphological changes in Duoxa–/– mice were negligible compared with mice with PTU-induced hypothyroidism, severe cerebellar dysfunction did persist. These results suggest that the cerebellar dysfunction in Duoxa–/– mice may be caused by the combined effect of congenital hypothyroidism and the functional disruption of the DUOX/DUOXA complex in the brain. Thus, Duoxa–/– mice may be a unique model to study the role of Duoxa on brain development.
Serum free T3 and T4 levels were kindly determined by Prof. Masami Murakami, Department of Clinical Laboratory Medicine, Gunma University Graduate School of Medicine. We thank Prof. Hirokazu Hirai, Department of Neurophysiology and Neural Repair, Gunma University Graduate School of Medicine, for permission to use the electrophysiological setup. We thank Mr. Michifumi Kokubo and Ms. Miski Aguhnia Khairinisa, Department of Integrative Physiology, Gunma University Graduate School of Medicine, for technical assistance.
This study was supported by Grant-in-Aid for Scientific Research (B) (25281024) from the Japan Society for the Scientific of Sciences (JSPS) to N.K., by Cosmic Innovative Research Grant from Japan Thyroid Associtation to T.I., and by grant DK 15070 from the National Institutes of Health to S.R.
The authors declare that there is no conflict of interest.