T4 in neuronal tissue, Decreased leads to BDNF, Reduced

The importance of thyroid hormones (TH) in brain development has been recognised and investigated for many decades (Bernal, 2011). Several human studies have shown that low levels of circulating maternal TH, even in the modest degree, can lead to neurophysiological deficits in the offspring, including learning and memory deficits, or even cretinism in most severe cases (Zoeller and Rovet, 2004; Henrichs et al., 2010). The levels of serum TH at birth are not always informative, as most of the neurological deficits are present despite the normal thyroid status of the newborn. That means that the cause of these impairments is rooted in the early stages of the neuronal development during the gestational period. The nature and the temporal occurrence of these defects suggest that TH may exert their effects through the neurotrophins, as they are the main regulators of neuronal system development (Lu and Figurov, 1997). Among them, BDNF represents the prime candidate because of its critical role in CNS development and its ability to regulate synaptic transmission, dendritic structure and synaptic plasticity in adulthood (Binder and Scharfman, 2004). Additionally, hippocampus and neocortex are two of the regions characterized by the highest BDNF expression (Kawamoto et al., 1996), and are also key brain areas for learning and memory functions. Indeed, it has been shown that the thyroid insufficiency (lower TH levels) results in reduction of BDNF levels (mRNA or protein) in the developing brain, and the most likely affected brain regions are the hippocampus and cortex (Koromilas et al., 2010, Shafiee et al., 2016).  The hippocampus direct and indirect interactions with the THs provide crucial information on the neurobiological basis of the hypothyroidism-induced mental retardation and neurobehavioral dysfunction. TH deficiency during the foetal and/or the neonatal period produces deleterious effects for neural growth and development (such as reduced synaptic connectivity, delayed myelination, disturbed neuronal migration, deranged axonal projections, decreased synaptogenesis and alterations in neurotransmitters' levels), possibly through decreased BDNF levels (Koromilas et al., 2010; Shafiee et al., 2016).

The empirical support for this direct KER is moderate. There are a limited number of studies investigating TH concentrations in the brain and BDNF levels. This is due to the fact that TH is difficult to measure in the brain and TH-induced changes in BDNF expression (mRNA or protein levels) can be subtle.

For the current AOP the temporal nature of this KER is described in the context of different developmental stages (Seed et al., 2005). The impact of brain TH concentrations on regulation of TH receptor (TR) regulated genes is age-dependent for a number of genes critical for normal hippocampal development. It is widely accepted that different genes, including BDNF, are downregulated under conditions of TH insufficiency (Pathak et al, 2011). TH supplementation has been shown to reverse some of the effects on gene expression (Liu et al., 2010; Pathak et al., 2011), including also BDNF expression (Wang et al., 2012).

Many in vivo studies have focused on the determination of the relationship between TH-mediated effects and BDNF expression in the brain. The majority of the work has been performed by evaluating the effects of TH insufficiency on BDNF developmental expression profile. The results, despite some differences, are showing a trend toward BDNF down-regulation.

Reductions in BDNF mRNA and protein were observed in hypothyroid rat models, created by exposing the animals to the TPO inhibitors methimazole (MMI) (Sinha et al., 2009) or propylthiouracil (PTU) (Neveu and Arenas, 1996; Lasley and Gilbert, 2011), and perchlorate (NIS inhibitor) (Koibuchi et al., 1999; 2001). These studies supported direct associations between lower level of TH and lower BDNF expression in the developmental cerebellum, hippocampus and cortex. The dose-response relationship could not be evaluated in these studies, as they were conducted in conditions of severe maternal hypothyroidism, namely after exposure to very high doses of the chemicals.

Additionally, in more complex models of maternal hypothyroidism in rats a reduction of hippocampal and cortex BDNF expression was observed (Wang et al., 2012; Liu et al., 2010). In these latter cases, hypothyroidism was developed via thyroidectomies, and T4 supplementation was performed at specific stages during gestation.

Indirect evidences supporting this KER:

- Koibuchi et al., 2001: in this in vivo study, newborn mice were rendered hypothyroid by administering MMI (TPO inhibitor) and perchlorate (NIS inhibitor) in drinking water to their mothers. Neurotrophin-3 (NT-3) and BDNF gene expression was depressed in the perinatal hypothyroid cerebellum. Since TH levels in the brain were not measured, the same study was also included in the indirect KER "TH synthesis decrease leads to BDFN reduction".

- Morte et al., 2010: in this in vivo study, maternal and fetal hypothyroidism was induced by maternal thyroidectomy (Tx) followed by the antithyroid drug MMI (TPO inhibitor) treatment. Tx dams treated with MMI showed increased TSH (~10 fold increase) and decreased serum T4 (~95%) and serum T3 (~90%). Whilst fetuses from Tx rats had normal serum TSH and cerebral cortex T4 and T3, pups born from dams treated with MMI showed increase of TSH (~8 fold), decreased cortex T4 (~75%) and cortex T3 (~95%). Additionally, analysis of gene expression in the fetal cerebral cortex showed a reduction of Camk4 (Ca(2+) and calmodulin-activated kinase) gene expression (~70%) induced by maternal and fetal hypothyroidism. As Camk4 during development is known to induce BDNF expression (Shieh et al., 1998), a reduction of cortical TH levels (leading to Camk4 mRNA downregulation) may lead to reduction of BDNF expression and/or signaling; however, BDNF expression was not measured in Morte et al. 2010.

- da Conceição et al., 2016: in this in vivo study, thyroidectomized (i.e,. hypothyroid) adult Wistar rats showed significant increase of serum TSH (~ 750% increase vs control rats), decrease of T4 (~ 80% decrease vs control) and T3 serum levels (~ 45% decrease vs control), together with a reduced hippocampal expression of MCT8 (~ 83% decrease vs control rats), TH receptor alfa (TRα1) (~ 77 % decrease vs control), deiodinase type 2 (DIO2) (~ 90% decrease vs control), and BDNF mRNA expression in hippocampus (~ 75% decrease vs control). The reduced gene expression of TRα1 (expressed in nearly all neurons in the brain (Schwartz et al., 1992)), MCT8 and DIO2 (both essential for the regulation of glia-neuron cell interaction in TH metabolism in the brain (Remaud et al., 2014)), indicate the presence of low TH levels in the hippocampus (hippocampal hypothyroidism) in thyroidectomized rats. However, direct measures of TH (T3 or T4) brain levels were not assessed in this study.

 - Pathak et al, 2011: in this in vivo study, the effect of maternal TH deficiency on neocortical development was investigated. Rat dams were exposed to MMI (TPO inhibitor) from GD6 until sacrifice. Decreased number and length of radial glia, loss of neuronal bipolarity, and impaired neuronal migration were recovered with early TH replacement (at E13-15). BDNF mRNA resulted downregulated (80% decrease at E14) while trkB expression was increased (2-fold) in hypothyroid fetuses at E14 stage. However, TH levels in the brain were not measured in this study.

Sui et al. 2010: in this in vivo study, T3, rT3 or vehicle were administered to young adult male rats either via systemic injection (i.e., IP administration of a single dose of 30 μg of T3/100 g body weight or the same dose of rT3 or the same volume of vehicle solution), or local brain infusion (i.e., intrahippocampal bilateral infusion (in the dorsal region) with 50 pmol T3, 50 pmol rT3 or vehicle (0.05% ethanol and saline solution), in 1 μl). Data showed that T3 administration increased reelin (∼ 3-fold increase relative to the vehicle group at 24 h following T3 IP injection, and ∼ 7-fold increase relative to the vehicle group after 1 h following T3 intrahippocampal injection), total BDNF (∼ 10-fold increase relative to the vehicle group at 24 h after IP injection, and ∼ 6-fold increase relative to the vehicle group at 12 h after intrahippocampal injection), and exon-specific BDNF mRNA expression in the hippocampus (after T3 IP injection: BDNF exon I and II transcripts was ∼ 50-fold higher compared to the vehicle levels, whereas exon IV and VI transcripts were ∼ 10-fold and ∼ 30-fold higher respectively; after T3 intrahippocampal injection: exon I and II: 3.5 to 4-fold; exon IV: ∼ 7-fold; exon VI: ∼ 12-fold relative to the vehicles, respectively at 2 h to 24 h (for exon I), 1 h and 6 h (for exon II),  1 h (for exon IV), and 12 h (for exon VI) after T3 infusion).

Conversely, administration of rT3 (inactive T3 isoform) through IP injection or intrahippocampal infusion did not significantly alter the hippocampal reelin or BDNF mRNA levels (two pathways critical for learning and memory processes regulation). Reelin protein levels resulted increased upon both IP injection (2-fold increase 24 h after injection) and intrahippocampal infusion of T3 (3-fold increase 4 h after infusion). Likewise, BDNF protein levels were upregulated after IP injection (~ 2.7-fold increase 24 h after injection) and intrahippocampal infusion (~ 2.5-fold increase 12 h after infusion) of T3. Analysis of transcriptional coactivators binding (e.g., cAMP response element binding protein-binding protein (CBP), and thyroid hormone receptor associated protein 220 (TRAP 220)) and RNA polymerase II (RNA Pol II)) revealed specific patterns of associations between such transcription factors and reelin or BDNF upon T3 administration. This study suggests that hippocampal BDNF mRNA and protein expression is under T3 regulation.

- Blanco et al. 2013: in this in vivo study rat dams were exposed to 0, 1 and 2 mg/kg/day of BDE-99 from GD 6 to PND 21. Data showed that transmission of maternal accumulated BDE-99 through placenta and breast milk caused a decrease of serum levels of T3 (by 13 ± 9% in the 2 mg/kg/day group), T4 (by 25 ± 13% in the 2 mg/kg/day group) and free-T4 (by up to 17 ± 9% in the 2 mg/kg/day group), causing downregulation of BDNF gene expression in the hippocampus of pups (by 32 ± 14% in the 2 mg/kg/day group). On the contrary, the expression of other TRs isoforms did not change in both cortex and hippocampus. Moreover, BDE-99 produced a delay in the spatial learning task in the water maze test (i.e., longer latency in reaching the platform at the highest BDE-99 dose vs control group), and a dose-response anxiolytic effect as revealed by the open-field test.

- Abedelhaffez and Hassan, 2013: in this in vivo study rat dams were exposed to MMI (TPO inhibitor) to induce hypothyroidism. Pups showed a decrease of plasma free T3, free T4, and growth hormone (GH), whilst plasma TSH was significantly increased. BDNF level was significantly decreased in both the hippocampus and cerebellum of rat pups.

- Shafiee et al., 2016: in this in vivo study rat dams were exposed to PTU (100 mg/L in drinking water) from embryonic day 6 to their PND 21. For 14 days (from PND 31 to 44), the rat pups were trained with either the mild treadmill exercise (TE) or the voluntary wheel exercise (VE). On PND 45-52, a water maze was used for testing their learning and memory ability. Hippocampal BDNF levels were assessed one day later. Data showed that pups exposed to PTU underwent a reduction of T4 levels (~ 70%) and an increase of TSH levels (~ 80%) at PND21. A reduction of hippocampal BDNF levels (~ 7-8% reduction comparing sedentary hypothyroid vs sedentary control pups) was observed in the treatment group. The conclusion of this quantitative study is that hypothyroidism during the foetal period and the early postnatal period is associated with the impairment of spatial learning and memory (e.g., ~ 55-60% increase of platform location latency in both sedentary hypothyroid male and female rats), and reduced hippocampal BDNF levels in both male and female rat offspring. Importantly, physical exercises (both VE and TE) significantly increased BDNF levels in both male and female hypothyroid animals (by ~ 2-3 percentage points) and improved learning and memory skills. Authors concluded that: "These findings suggest that the increase in BDNF levels following a period of physical activity in hypothyroid rat pups is an important mechanism by which exercise alleviates the learning and memory deficits induced by hypothyroidism".

- Shi et al. 2017: in this in vivo study rat dams were randomly treated with decabromdiphenylether (BDE)-209 (100, 300, and 900 mg/kg body weight) or corn oil by gavage on gestational days 6 to 20. Blood was obtained through heart puncture for TH analysis in male rat offspring on PND 60. Data indicated that BDNF protein levels in the hippocampus decreased by 13% and 33% respectively in the 300 mg/kg and 900 mg/kg dose group. Total T4 levels and free T4 levels were significantly decreased in the BDE-209 treated-group (900 mg/kg, 300 mg/kg), and total T3 levels in 300 mg/kg group were also significantly decreased compared to the control group (ctr) (no significant difference was observed in 100 mg/kg group). In this study, decreased BDNF levels are well correlated with decrease of total and free T4 levels occurring upon exposure to BDE-209. 

- Mokhtari et al. 2017: in this in vivo study rats underwent transient middle cerebral artery occlusion (tMCAo) to induce ischemic brain stroke. Rats were randomly divided in four groups: Co (control), Sh (sham), tMCAo and tMCAo + T3 (intracerebroventricular injection of T3 at 25 ug/kg body administered 24 after reperfusion). T3 significantly improved the learning and memory compared with tMCAo group, as shown by Morris water maze test. Step-through latency significantly increased in the T3 group compared with tMCAo group. Moreover, BDNF mRNA and protein levels were decreased in the tMCAo compared with Co and Sh group (~ 15% decrease of protein and ~ 20% decrease of mRNA vs Co or Sh), and addition of T3 increased BDNF mRNA and proteins compared to Co, Sh and tMCAo groups (~ 94% increase of protein and ~ 750% increase of mRNA comparing tMCAo + T3 vs tMCAo). This study points out again that BDNF levels are under the control of T3.

- Sabbaghziarani et al. 2017: similar to previous study from Mokhtari et al. (2017), here cerebral ischemia was induced by MCAo in male Wistar rats; a group of rats was also injected with T3 (25 μg/kg, IV injection) at 24 hours after ischemia. BDNF gene and protein levels (along with nestin and Sox2) were increased upon T3 treatment vs ischemic group. T3 treated rats also showed higher levels of serum T3 and T4, and lower levels of TSH vs ischemic group 4 days post ischemia induction. These data globally indicate that brain increased T3 levels increase BDNF expression and protein levels.

- Kawahori et al. 2018: in this in vivo study rat dams were administered with MMI (0.025% w/v) from 2 weeks prior to conception until delivery, which induced mild maternal hypothyroxinemia during pregnancy, comparing MMI and control offspring at day 28 and day 70 after birth. MMI-exposed pups showed an impaired learning capacity in the behavior tests. Hippocampal steady-state Bdnf exon IV (responsible for neural activity-dependent Bdnf gene expression) expression was lower in MMI group than in Ctr at day 28, while at day 70, hippocampal Bdnf exon IV expression at the basal level was comparable between the two groups. Additionally, persistent DNA hypermethylation was found in the promoter region of Bdnf exon IV in the hippocampus of MMI group vs ctr, which may be responsible for the decrease of Bdnf exon IV expression in the treated group.

Hypothyroidism (i.e., induced by chemicals known to inhibit TPO or NIS, or by thyroidectomy, leading to low TH serum levels) is generally associated with lower levels of BDNF in brain tissues. As described in Conceição et al., 2016, thyroidectomized adult rats, apart from showing reduced TH levels, also presented reduced hippocampal gene expression of MCT8, TRα1, DIO2 and BDNF, which support a link between hippocampal hypothyroidism and reduced BDNF levels.

However, despite the fact that many in vivo studies have shown a correlation between hypothyroidism and BDNF expression in the brain, there are no studies simultaneously measuring the levels of both TH and BDNF in the brain. Therefore, no clear consensus can be reached by the overall evaluation of the existing data. There are numerous conflicting studies showing no significant change in BDNF mRNA or protein levels under hypothyroid conditions (Alvarez-Dolado et al., 1994; Bastian et al., 2010; 2012; Royland et al., 2008; Lasley and Gilbert, 2011).However, the results of these studies cannot exclude the possibility of temporal- or region-specific decreased BDNF effects as a consequence of foetal hypothyroidism. A transient TH-dependent BDNF reduction in early postnatal life can be followed by a period of normal BDNF levels or, on the contrary, normal BDNF expression in the early developmental stages is not predictive of equally normal BDNF expression throughout development. Moreover, significant differences in study design, the assessed brain regions, the age and the method of assessment in the existing studies, further complicate result interpretation.

- In Alvarez-Dolado et al. 1994, hypothyroid rats showed decreased trk (BDNF receptor) mRNA levels in the striatum on PND 5, PND 15 and in adults, increase of the low affinity neurotrophin receptor p75LNGFR mRNA in hypothyroid cerebellum on PND 5 and PND 15, decrease of nerve growth factor (NGF) mRNA in the cortex, hippocampus, and cerebellum of hypothyroid rats on neonatal hypothyroid rats on PND 15 and also after adult-onset hypothyroidism, whilst the relatively high expression of the two BDNF mRNAs did not change in any brain area.

- Bastian et al. (2010) assessed the effects Cu and Fe deficiencies on circulating and brain TH levels during development in pregnant rat dams rendered Cu deficient (CuD), Fe deficient (FeD), or TH deficient (by PTU treatment) from early gestation through weaning. Serum T4 and T3, and brain T3 levels were subsequently measured in PND 12 pups. Despite the remarkable decrease of serum TH and brain T3 induced by PTU treatment (and also by CuD and FeD), no significant changes of Bdnf IV mRNA levels were found. Authors commented that 'one explanation for this discrepancy is that many of the previous studies were performed using discrete brain regions, whereas this study was performed on whole-brain RNA'. Along the same line, in a follow up study, Bastian et al. (2012) could not find statistically significant reductions of Bdnf IV, Bdnf VI, and total Bdnf mRNA levels in hippocampus or cerebral cortex of Fe and TH deficient pups.

- Royland et al. (2008) assessed the effects of a PTU (TPO inhibitor) administration to pregnant rats from gestational day 6 until sacrifice of pups prior to weaning. However, PTU treatment did not change the expression of Bdnf at the mRNA level.

- In Lasley and Gilbert, 2011 study, different concentrations of PTU were administered to rat pregnant dams from gestational day 6 until weaning of the pups. Pups were sacrificed on PND 14, PND 21 and PND 100, analysis of TH serum levels was performed, along with analysis of hippocampal, cortical, and cerebellar levels of BDNF protein. While PTU caused a strong decrease of TH serum levels, no differences in BDNF protein were detected in the pre-weanling animals as a function of PTU exposure. On the contrary, dose-dependent decrease of BDNF levels emerged in adult males as a consequence of prenatal exposure despite the return to control TH levels. These findings reflect the potential for delayed impact of even modest TH reductions during critical periods of brain development on BDNF, a protein important for normal synaptic formation, as commented by the authors of this study.

It should also be considered that in severe models of TH deficiency, BDNF responsivity to TH is regulated in a promoter-, age-, and brain region-specific fashion (as described by Anderson and Mariash, 2002), and even modest differences of these parameters in study design may explain inconsistencies in study results.

The absence of significant changes in BDNF levels in the above cited studies could be also due to different sensitivity of analythical tools, experimental design and statistical processing of the results.

While PTU (TPO inhibitor) has been shown to decrease serum TH levels and brain BDNF protein levels and mRNA expression in offspring born from PTU-treated rat dams (Shafiee et al. 2016; Chakraborty et al., 2012; Gilbert et al. 2016), in Cortés et al., 2012 study, treatment of adult male Sprague-Dawley rats with PTU induced an increase in the amount of BDNF mRNA in the hippocampus, while the content of TrkB, the receptor for BDNF, resulted reduced at the postsynaptic density (PSD) of the CA3 region compared with controls. Treated rats presented also thinner PSD than control rats, and a reduced content of NMDAr subunits (NR1 and NR2A/B subunits) at the PSD in hypothyroid animals. While these data indicate differential effects elicited by PTU (i.e., TPO inhibition) on BDNF expression/regulation comparing the adult vs foetal brain, downregulation of TrkB receptors still leads to decrease signalling pathways regulated by BDNF.