Maternal Thyroid Hormone is Required for ...

Accepted Article

PROFESSOR JENS MITTAG (Orcid ID : 0000-0001-7778-5158)

Article type

: Original Article

Maternal Thyroid Hormone is Required for Parvalbumin Neuron Development in the Anterior Hypothalamic Area Lisbeth Harder1, Susi Dudazy-Gralla2, Helge Müller-Fielitz3, Jens Hjerling Leffler4, Björn Vennström2, Heike Heuer5, Jens Mittag1 1

University of Lübeck, Center of Brain, Behavior and Metabolism CBBM / Medizinische Klinik I, 23562 Lübeck, Germany 2 Department of Cell and Molecular Biology, Karolinska Institutet, 171 77 Stockholm, Sweden 3 University of Lübeck, Center of Brain, Behavior and Metabolism CBBM / Institut für Pharmakologie und Toxikologie, 23562 Lübeck, Germany 4 Department of Medical Biochemistry and Biophysics, Karolinska Institutet, 171 77 Stockholm, Sweden 5 IUF – Leibniz Research Institute for Environmental Medicine, 40225 Düsseldorf, Germany Corresponding author: Prof. Dr. Jens Mittag University of Lübeck Center of Brain, Behavior and Metabolism CBBM/Medizinische Klinik I Ratzeburger Allee 160, 23562 Lübeck Tel: +49 (0) 451 3101 7826 Fax: +49 (0) 451 500 3339 E-mail: [email protected] Short title: Thyroid Hormone Dependency of Parvalbumin Neurons Key Words: Thyroid hormone, hypothalamus, parvalbumin, thyroid hormone receptor α1, development Conflict of Interest The authors declare no competing financial interests. Funding Information This work is supported by grants from the Deutsche Forschungsgemeinschaft (Mi1242/2-1 and 3-1 to J.M.; GRK1957 “Adipocyte-Brain Crosstalk” to L.H. and J.M.; HE3418/7-1 to HH and MI1242/6-1 to JM in the framework of SPP1629 “Thyroid TransAct”).

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/jne.12573 This article is protected by copyright. All rights reserved.

Accepted Article

Abstract (278 words) Thyroid hormone (TH) is crucial for brain development and function. This becomes most evident in untreated congenital hypothyroidism, leading to irreversible mental retardation. Likewise, maternal hypothyroxinemia, a lack of TH during pregnancy, is associated with neurological dysfunction in the offspring such as autism and reduced intellectual capacity. In the brain, TH acts mainly through TH receptor alpha 1 (TRα1). Consequently, mice heterozygous for a dominant-negative mutation in TRα1 display profound neuroanatomical abnormalities including deranged development of parvalbumin neurons. However, the exact timing and orchestration of TH signaling during parvalbumin neuron development remains elusive. Here we dissect the development of parvalbumin neurons in the anterior hypothalamic area (AHA) in male mice using different mouse models with impaired pre- and postnatal TH signaling in combination with bromodeoxyuridine birth dating and immunohistochemistry. Our data reveal that hypothalamic parvalbumin neurons are born at embryonic day 12, are first detected in the AHA at postnatal day 8 and reach their full population number at P13. Interestingly, they do not require TH postnatally, as their development is not impaired in mice with impaired TH signaling after birth. In contrast, however, these neurons crucially depend on TH through TRα1 signaling in the second half of pregnancy - a period where the hormone is almost exclusively provided by the mother. Our findings therefore for the first time directly link a maternal hormone to a neuroanatomical substrate in the fetal brain, and underline the importance of proper TH signaling during pregnancy for offspring mental health. Given the role of hypothalamic parvalbumin neurons in the central control of blood pressure, the study advocates the inclusion of cardiovascular parameters in the current discussion on possible TH substitution in maternal hypothyroxinemia.

Introduction Disturbances in thyroid hormone (TH) economy resulting in altered 3,3’,5-triiodothyronine (T3) or its precursor thyroxine (T4) can have profound effects on neuronal development and mental health. In congenital hypothyroidism, characterized by reduced TH levels in newborns, developmental delay

This article is protected by copyright. All rights reserved.

Accepted Article

and irreversible mental retardation occur if the condition is not recognized and treated immediately with T4 (1). The importance of TH for central nervous system (CNS) development becomes also evident in the Allan-Herndon-Dudley syndrome (AHDS), a condition caused by loss of function mutations in the monocarboxylate transporter 8 (Mct8) gene (2), a critical membrane transporter for the blood-brain-barrier (BBB) passage of THs in humans. Patients display severe mental retardation as a consequence of the hypothyroid state of their brain (3). The CNS phenotype has been replicated in the corresponding animal models. Pax8 knockout (ko) mice are born without a functional thyroid gland and constitute an established model of congenital hypothyroidism (4, 5). For AHDS, only Mct8/Oatp1c1 double ko animals (Mct8/Oatp1c1 dko) represent a suitable model including CNSspecific hypothyroidism and abnormal neuronal differentiation (6), as the single Mct8 ko mouse model (7) only reproduces the human peripheral symptoms but shows no major signs of neurological deficits (8) due to the compensating BBB transporter organic anion-transporting polypeptide 1c1 (OATP1C1) in mice. Even before birth, TH is crucial for embryonic CNS development. However, as the fetal thyroid gland only starts folliculogenesis around embryonic day 15 (E15) in mice (9) and does not produce hormone until shortly before birth in mice (corresponding to the beginning of the third trimester in humans), but nuclear TH receptors (TRs) are already present in the developing brain from embryonic day E13.5, the required TH during this critical period is provided by the mothers (10, 11). Consequently, maternal hypothyroxinemia, e.g. due to iodine deficiency of the mother, can severely affect offspring brain development (12). Despite these clear links between TH supply and brain development, little is known about the neuroanatomical targets of TH in the developing brain. T3 action is largely mediated by TRα1 and TRβ (encoded by Thra and Thrb respectively), with TRα1 accounting for 70-80% of the TR in the adult brain (13), whereas TRβ controls selective functions such as the feedback loop of the hypothalamic-pituitary-thyroid axis (14). The major isoform for almost all neurons, however, is TRα1, being expressed in mice in postmitotic neurons from E13.5 onwards throughout neuronal maturation into adulthood (15). TRs are transcription factors, which bind to thyroid hormone response


Accepted Article

elements in the promotor region of target genes independently of ligand availability (14, 16) and activate or suppress gene expression depending on the presence or absence of THs (14). The importance of TRα1 signaling for neuronal development is underlined by findings in mice heterozygous for the mutant TRα1R384C (TRα1+m). In this particular model, the affinity of the mutant TRα1R384C to T3 is ten-fold reduced, leading to a hypothyroid-like phenotype at physiological T3 concentrations for TRα1 targets. However, when T3 concentrations are increased 10fold, e.g. by TH treatment, the receptor can be reactivated in vivo (17). TRα1+m mice show anxiety and memory deficiencies as well as motoric deficits, associated with a reduced number of neurons expressing the Ca2+-binding protein parvalbumin (PV) in the hippocampus (18) and motor cortex respectively (19). In these mice, another TH sensitive population of PV neurons was identified in the anterior hypothalamic area (AHA), which was strongly reduced in TRα1+m mice (20). The ablation of these neurons in wildtype (wt) animals leads to an increase in blood pressure and heart rate (20), suggesting that they play an important role in the regulation of the cardiovascular system. However, the development of these AHA PV neurons had never been studied and their dependency on TH is not well characterized. Here we show that AHA PV neurons are born at E12 and reach their destination in the AHA between P8 and P13. In contrast to cortical PV neurons, however, PV neurons in the AHA do not require TH postnatally, but depend on TRα1 mediated T3 signaling between E12 and birth, a period where the hormone is provided by the mother.

Materials and Methods Animal husbandry Mice were housed (21°C, 12h light/12h dark cycle) with ad libitum access to standard diet and water. Animal care procedures were in accordance with the guidelines set by the European Community Council Directives (86/609/EEC) and were approved by Stockholm’s Norra Djurförsöksetiska Nämnd, Sweden, or the MELUR Schleswig-Holstein, Germany. Tissue from the following mouse strains were used: wt C57BL/6J or NCr (The Jackson Laboratory, USA); Pax8 ko (Pax8-/-, no


Accepted Article

detectable serum TH) and control (Pax8+/-) mice at 3 weeks of age (4); Mct8 ko mice (Mct8-/y, (7)) and Mct8/Oatp1c1 (M/O) dko mice (Mct8-/y/Oatp1c1-/-, no uptake of TH into the brain after bloodbrain-barrier closes), (6)) as well as Mct10 ko (Mct10-/-, (21, 22)) and Mct10 control (Mct10+/-) mice as adults; TRα1+m mutant mice heterozygous for the dominant-negative R384C mutation in TRα1 (normal level of serum TH, but 10-fold reduced TRα1 signaling (17)) as well as the combination with TRβ ko (TRβ

-/-

, elevated levels of serum TH due to impairments in the hypothalamus-pituitary-

thyroid axis due to lack of TRβ signaling (23, 24)) as adult mice. Moreover, offspring of Nkx2-1Cre:RFP mice (25) crossed to 5HT3aR-BACEGFP mice (26) were analyzed.

Bromodeoxyuridine (BrdU) birth dating Wt C57BL/6J mice were mated and females were separated after a positive plug (defined as E0.5) and i.p. injected with 50 mg/kg BrdU (Sigma, Germany) in 0.1 M phosphate buffered saline (PBS, 1.4 M NaCl, 0.027 M KCl, 0.018 M KH2PO4, 0.19 M Na2HPO4) on different days after conception between E9.5 and E.13.5. Male offspring were killed by cervical dislocation at 6 weeks of age, brains were removed and fixed in 4% paraformaldehyde (PFA, overnight, 4°C) followed by incubation in 30% sucrose for several days. Brains were deep-frozen, cut in 20µm coronal section using a cryostat (Leica Biosystems, Germany) and stained as described below. Pictures from the AHA and the motor cortex (M1 and M2 area) of 3-6 animals per time point were obtained using a DMI6000B fluorescence and a SP5 confocal microscope (Leica Biosystems, Germany) and the total numbers of PV labeled cells as well as the amount of BrdU/PV double-labeled cells were counted in the respective areas.

Rescue of PV neurons in the AHA of TRα1+m mice by different prenatal T3-treatment approaches. For the attempt to rescue PV neurons in the AHA by high maternal TH levels throughout the entire pregnancy, TRβ ko dams, which display endogenously elevated TH levels due to impaired negative feedback in the hypothalamus-pituitary-thyroid axis, were mated with male TRα1+m or TRα1+m/TRβ dko mice. For the attempt to rescue PV neurons in the AHA by high maternal TH levels during the second half of pregnancy, wt dams were mated with male TRα1+m mice, pregnant


Accepted Article

females were separated and treated orally with T3 containing drinking water (0.5 μg/mL T3 (Sigma, Germany) and 0.01% bovine serum albumin) from E12.5 until birth. Both treatment regimen (TRβ ko as well as T3) lead to 10-fold elevated levels of T3 which is sufficient to reactivate the mutant TRα1 (17, 19). Brains of male wt, TRα1+m, TRβ ko and TRα1+m/TRβ dko offspring were removed around 3-6 weeks of age, and used for immunohistochemistry as described below. Immunohistochemistry 3,3’-diaminobenzidine (DAB) staining was performed on 4% PFA fixed, free floating 20µm cryostat coronal sections from Pax8 ko and controls, Mct8 ko and M/O dko, Mct10 ko and controls, TRα1+m, TRβ ko and TRα1+m/TRβ dko as well as C57BL/6NCr mice. 1x PBS was used for all solutions and washing steps. Sections were blocked 1 h with 1% H2O2 followed by 5% normal goat serum (NGS, abcam, UK), avidin (avidin/biotin blocking kit, Vector Laboratories, USA) and 0.3% Triton-X 100 for 1 h. Subsequently sections were incubated in 5% NGS, biotin (avidin/biotin blocking kit, Vector Laboratories, USA), 0.3% Triton-X 100 and rabbit anti-PV primary antibody (Suppl. Table 1A) at 4°C overnight. Next, sections were treated with a biotinylated secondary antibody (Suppl. Table 1B) in 0.3% Triton-X 100 followed by ABC-Solution (ABC-kit, Vector Laboratories, USA) for 1 h each and stained in DAB-solution (0.5 mg/mL). Sections were mounted and dried overnight at room temperature (RT), dehydrated and cleared in xylene, and cover slipped using pertex (Medite GmbH, Germany). Cell counts were obtained as described previously (20). Section incubated without first or secondary antibody were used as controls and did not show any stainings.

Immunofluorescence Immunofluorescence staining was performed on 4% PFA fixed, free floating 20µm cryostat coronal sections of wt C57BL/6J animals. All solutions were prepared with 1x PBS. Sections from BrdU treated mice were blocked 1 h with 5% NGS and 0.3% Triton-X 100, followed by incubation in rabbit anti-PV primary antibody (Suppl. Table 1A) in blocking solution at RT overnight. Next, sections were incubated in Alexa Fluor 488 labeled secondary antibody (Suppl. Table 1B) for 1 h followed by antigen retrieval (10 min in 1M HCl on ice, 10 min in 2M HCl at RT, 20 min in 2M HCl at 37°C, 10


Accepted Article

min in 0.1M borate buffer at RT). Subsequently sections were incubated in rat anti-BrdU primary antibody (Suppl. Table 1A) in blocking solution at RT overnight. Finally, sections were incubated 1 h in Alexa Fluor 594 secondary antibody (Suppl. Table 1B). Wt samples were treated with 10 mM NaCitrate at 80°C for 10 min for antigen retrieval and blocked with 5% normal donkey serum (SigmaAldrich, USA) and 0.3% Triton-X 100. Next, section were incubated in goat anti-PV primary antibody in combination with rabbit anti-VGLUT2 (Suppl. Table 1A) as well as rabbit anti-PV primary antibody in combination with mouse anti-GAD67 (Suppl. Table 1A) in blocking solution at 4°C overnight. Subsequently, sections were incubated 1 h using the respective secondary antibodies (Suppl. Table 1B). All slices were mounted on object slides using ProLong Diamond Antifade Mounting Medium with Dapi (Life Technologies, USA) or Hard Set Antifade Mountain Medium with DAPI (Vector Laboratories, USA). Tissue from Nkx2-1-Cre:RFP x 5HT3a-EGFP double reporter mice were processed as described before (25, 26) with a rabbit anti-PV primary antibody (Suppl. Table 1A) and an Alexa Fluor 568 labeled secondary antibody (Suppl. Table 1B).

Experimental Design and Statistical Analysis Statistical analysis was performed using GraphPad Prism 5 or 6 (GraphPad Software, USA) and p as well as t values were obtained by an unpaired 2-tailed Student’s t-test for simple comparison to wt controls or 2-way ANOVAs for comparisons involving genotype and treatment with subsequent Holm-Sidak posthoc tests for multiple comparisons. Significance was defined as p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***). All data are presented as mean +/- s.e.m.

Results Development and basic characterization of AHA PV neurons As the development of AHA PV neurons has not been studied to date, we initially aimed to determine their appearance in the AHA postnatally. PV immunoreactivity in the AHA was first detected at postnatal day 8 (P8), and rose gradually until a population of PV neurons equivalent to the number in adult mice was detected at P13 (Figure 1A and B). Next we aimed to define the respective “birthdates” of these neurons, i.e. their transition from proliferation to a postmitotic state in relation to


Accepted Article

cortical PV neurons. To this end, proliferating neurons were labeled with BrdU at different time points after conception during embryonic development, and the amount of BrdU-labeled PV positive neurons in the AHA was determined postnatally at the age of six weeks (Figure 1C with overview and high magnification of the AHA at E11.5 in 1D). We found that the percentage of BrdU labeled PV neurons in the AHA showed maximum labeling at E11.5 and E12.5 and already declined at E13.5 (Figure 1C and Supplementary Figure 1A). This is in contrast to cortical PV neurons that continuously increased from E9.5 onwards, with a maximum on day E13.5 (Supplementary Figure 1B), as expected from previous studies (27, 28), demonstrating that the precursor cells of PV neurons in the AHA exit proliferation earlier than precursors of cortical PV neurons. Using Nkx2-1-Cre:RFP x 5HT3a-EGFP double reporter mouse, in which all cells from the medial ganglionic eminence (MGE) are labelled with a red fluorescent protein (RFP) and all cells from the caudal ganglionic eminence (CGE) express a green fluorescent protein (EGFP) (25, 26, 29), we also revealed that PV neurons in the AHA do not derive from the MGE, where most cortical neurons are born (Supplementary Figure 1C and D) (30). With regard to their neurotransmitter repertoire, however, PV neurons in the AHA were positive for glutamate decarboxylase 67 (GAD67, Supplementary Figure 1E) as marker for gamma-aminobutyric acid (GABA) positive neurons and negative for vesicular glutamate transporter 2 (VGLUT2, Supplementary Figure 1F) as a marker for glutamatergic neurons.

Impaired postnatal TH signaling has no influence on PV neurons in the AHA. To identify the period within the developmental program in which PV neurons depend on TH signaling, we analyzed PV immunoreactivity in the cerebral motor cortex and the AHA from two different animal models for impaired postnatal TH signaling: Pax8 ko mice, which are an established model for congenital hypothyroidism (4) and do not have detectable level of thyroid hormone after birth due to the lack of a thyroid gland, and Mct8/Oatp1c1 dko mice that exhibit a pronounced TH deficiency postnatally in the CNS, as with the tightening of the BBB starting at E16, TH entering to the brain is no longer possible due to the lack of these two transporters (6). In both animals, we observed a reduced PV immunoreactivity in the motor cortex compared to control littermates in agreement with previous studies (6), while the AHA population of PV neurons remained unaffected


Accepted Article

(Figure 2; p=0.50 and t=0.73 for Pax8 ko to control and p=0.43 and t=0.87 for Mct8/Oatp1c1 dko to Mct8 single ko). Mice lacking the monocarboxylate transporter 10 (MCT10) did not show altered numbers in the cortex or AHA population of PV neurons (Supplementary Figure 2A and B; p=0.18 and t=1.50 for Mct10 ko to control), presumably due to the minor role of this transporter for TH transport into the brain. Taken together, these findings demonstrate that postnatal TH signaling is crucial for the development of cortical PV neurons, whereas PV neurons from the AHA do not depend on TH after birth.

The development of PV neurons in the AHA is directly influenced by maternal TH Based on our findings that PV neurons in the AHA are born around E12, but do not seem to require the hormone postnatally, we speculated that their critical window for TH signaling might be prenatally. To test this hypothesis, we took advantage of the TRα1+m strain, which has the R384C mutation in TRα1 that lowers affinity to the ligand TH by a factor of 10, thus rendering it essentially unresponsive at physiological TH concentrations. However, the mutant TRα1 can be reactivated in vivo at any time through an elevation of TH levels. First, we specifically reactivated the embryonic mutant TRα1 throughout the entire pregnancy using TRβ ko females (23) with high endogenous levels of TH as dams (17, 19). When we then analyzed the male TRα1+m and control offspring of these hyperthyroid mothers, we observed a significant interaction between treatment and TRα1 mutation (Figure 3A and C left panel, 2-way ANOVA effect of TRα1 genotype p