https://orcid.org/0000-0003-1946-2275
Abstract
Traumatic brain injury (TBI) is associated with disruption of cerebral blood flow leading to localized brain hypoxia. Thyroid hormone (TH) treatment, administered shortly after injury, has been shown to promote neural protection in rodent TBI models. The mechanism of TH protection, however, is not established. We used mouse primary cortical neurons to investigate the effectiveness and possible pathways of T3-promoted cell survival after exposure to hypoxic injury. Cultured primary cortical neurons were exposed to hypoxia (0.2% oxygen) for 7 hours with or without T3 (5 nM). T3 treatment enhanced DNA 5-hydroxymethylcytosine levels and attenuated the hypoxia-induced increase in DNA 5-methylcytosine (5-mc). In the presence of T3, mRNA expression of Tet family genes was increased and DNA methyltransferase (Dnmt) 3a and Dnmt3b were downregulated, compared with conditions in the absence of T3. These T3-induced changes decreased hypoxia-induced DNA de novo methylation, which reduced hypoxia-induced neuronal damage and apoptosis. We used RNA sequencing to characterize T3-regulated genes in cortical neurons under hypoxic conditions and identified 22 genes that were upregulated and 15 genes that were downregulated. Krüppel-like factor 9 (KLF9), a multifunctional transcription factor that plays a key role in central nervous system development, was highly upregulated by T3 treatment in hypoxic conditions. Knockdown of the KLF9 gene resulted in early apoptosis and abolished the beneficial role of T3 in neuronal survival. KLF9 mediates, in part, the neuronal protective role of T3. T3 treatment reduces hypoxic damage, although pathways that reduce DNA methylation and apoptosis remain to be elucidated.
Traumatic brain injury (TBI) produces edema and inflammation, disruption of blood flow, and resulting hypoxia, which leads to neuronal injury and cell death. Even short-term hypoxia may alter the DNA methylation pattern, resulting in changes in the pattern of gene expression. It has been reported that after short-term exposure to hypoxia, hypermethylation of CpG islands is detected globally in neuronal DNA (1), which may lead to alterations in neuronal function. Long-term hypoxia may attenuate synaptic connectivity and firing frequency, triggering mitochondria-mediated apoptosis (2, 3).
DNA methylation is catalyzed by the DNA (cytosine-5) methyltransferase (Dnmt) family. Dnmt1 is a conserved enzyme and responsible for maintaining existing methylation patterns. Dnmt3a and Dnmt3b are highly expressed during development and involved in establishment of de novo DNA methylation patterns. DNA de novo methylation and demethylation control gene expression critical for brain development and neuronal function. In the adult brain, DNA methylation pattern-driven gene transcription regulates neuronal synaptic plasticity, learning, and memory (4, 5). Demethylation of DNA requires ten-eleven translocation (TET) methylcytosine dioxygenase. There are three TET members (TET1, TET2, and TET3) with overlapping function in catalyzing the oxidization of 5-methylcytosine (5-mc) to 5-hydroxymethylcytosine (5-hmc), which is the first step of demethylation. 5-hmc is found in the CpG-rich region of promoters, near the transcription start site, and is associated with active gene transcription (6). Among mammalian tissues, the brain has the highest level of 5-hmc expression and lowest DNA methylation, resulting in higher levels of gene expression (7). TET-mediated DNA demethylation is correlated with activation of DNA repair in cells. Hypoxia exposure decreases 5-hmc expression, increases 5-mc expression, and results in DNA hypermethylation (8).
Thyroid hormone (TH) is essential for normal brain development and function in adults. TH is transported to the brain through specific transporters, including monocarboxylate transporter 8 (MCT8) in humans and both MCT8 and organic anion-transporting polypeptide 1 (OATP1) in rodents. Most T3, ∼80%, is produced locally through conversion of T4 by the type 2 deiodinase in astrocytes and tanycytes and transported to neurons and oligodendrocytes via MCT8 and other transporters (9). TH deficiency in the brain resulting from MCT8 mutation in humans, or knocking out both the MCT8 and OATP1 transporters in mice, manifests severe brain abnormalities, including motor and cognitive defects in humans and defective oligodendrocyte maturation and myelination in mouse models (9, 10). HIF1α, which increases after hypoxia, stimulates type 3 deiodinase activity, which converts T4 to the inactive reverse T3 and reduces TH signaling in neurons (11, 12). T4 and T3 levels were significantly reduced after TBI in a controlled cortical injury mouse model. T4 administration normalized TH levels and stimulated genes in the cerebral cortex, which promote recovery (13, 14). Several studies have shown that TH plays a role in promoting post-TBI recovery. TH treatment 1 hour afterTBI reduced brain edema, improved motor and cognitive function, and reduced anxiety (15, 16). Additionally, TH treatment in animal models stimulated expression of the genes involved in neural cell survival (13, 15, 16). TH treatment of hypoxic neuroblastoma cells resulted in early peak induction of VEGF and c-Jun and reduced TGFβ, compared with the conditions without T3 treatment (13), which promoted early adaptation to hypoxia and improved neuronal survival. We postulated that TH produced beneficial effects after injury primarily through TH-mediated gene expression in the brain. TH regulates a wide range of genes in the rodent brain, including in the cerebellum, cerebral cortex, hippocampus, and striatum (17). The Krüppel-like factor 9 (KLF9) gene is highly induced by T3 throughout the brain regions during development, and KLF9 protein interacts with >100 factors during brain development (18, 19). KLF9 is responsible for late-phase neuronal maturation in hippocampal neurogenesis (20), as well as regulating oligodendrocyte differentiation and myelin regeneration (21).
We investigated the pathways of TH protection in neural injury using hypoxic mouse primary cortical neurons as an injury model. Mouse primary cortical neurons were exposed to 0.2% hypoxia for 7 hours, either in the presence or absence of T3 (5 nM). After treatment, we analyzed DNA 5-hmc and 5-mc levels by ELISA assay and the mRNA level of DNA methylation enzymes (Dnmts and Tets). We compared the gene expression pattern, determined by RNA sequencing (RNA-seq), with or without T3 treatment after hypoxic exposure, and identified T3-regulated genes that promote neuronal survival. We used a gene-silencing approach to determine the role that KLF9 plays in T3-promoted neuronal survival.
Materials and Methods
Primary mouse cortical neurons and rat primary neural stem cells
Mouse primary cortical neurons (Thermo Fisher Scientific) were cultured according to the manufacturer’s instructions. In brief, the dish and chamber slides were precoated with poly-d-lysine at 4.5 μg/m2. Cells were plated at a density of 0.3 million cells per 10-cm dish or to a four-chamber slide at 1 × 105 cells per chamber. Cells were grown in Neurobasal® medium containing 200 mM GluMAX™-1 (final concentration 0.5 mM) and B27 supplement (final concentration 2%) with 50% medium change every 3 days. Rat fetal neural stem cells (NSCs) (Thermo Fisher Scientific) were grown and passaged three times, according to the instructions. All media and supplements used were from the fetal NSC kit (Thermo Fisher Scientific). For differentiation studies, fetal NSCs were plated at 5 × 104 cells/cm2 on a CellStar-coated 10-cm dish, or glass chamber slide, and grown for 2 days. Cells attach to the plate and differentiate spontaneously. After 4 days of differentiation, cells were used for experiments.
Hypoxia exposure and T3 treatment
Cell medium was preequilibrated in a hypoxic chamber (HypOxystation H45) for at least 2 hours. The hypoxic chamber was supplied with 94.8% nitrogen, 5% CO2, and 0.2% oxygen at 37°C. After the cells were transferred to the hypoxic chamber, the culture medium was replaced immediately with preequilibrated medium and T3 was added to the medium at a final concentration 0, 5, or 10 nM. Cells were incubated in the hypoxic chamber for variable time periods, as shown. At the end of incubation, cells were lysed in the chamber to extract for either protein or RNA analysis, to prevent reoxygenation-induced cell damage that can occur when exposed to oxygen. Each treatment was performed with replicates from four chambers for imaging and three 10-cm dishes for biochemical analysis.
Diiodothyropropionic acid and 3,5,3′-triiodothyroacetic acid treatment of primary neurons in hypoxia
Cell maintenance and hypoxic exposure conditions were the same as described above for the experiments with T3. The hormone concentrations used, diiodothyropropionic acid (DITPA; 20 μM) or 3,5,3′-triiodothyroacetic acid (Triac; 40 nM), were adapted from previously published cell culture studies (22–24) to achieve stimulation equipotent to the T3 concentrations used. The duration of hypoxic exposure was 7 hours. Each treatment was performed with replicates from four chambers.
Analysis of DNA 5-hmc) and DNA 5-mc by ELISA
Upon completion of hypoxic/T3 treatment, cells were lysed in the hypoxic chamber and genomic DNA was isolated using a QIAamp DNA mini blood kit (Qiagen). 5-mc and 5-hmc were quantified using a MethylFlash global DNA 5-hmc ELISA kit and a MethylFlash global DNA methylation (5-mc) ELISA kit (Epigentek).
RNA-seq (gene expression dynamics)
RNA was isolated from primary cortical neurons after hypoxic exposure. Control RNAs were isolated from cells cultured in normal oxygen concentration. Each group for RNA-seq included cells pooled from three 10-cm dishes. The RNA-seq was performed by Illumina HiSeq 380, genome reference mm10. RNA-seq and data analysis were performed by the University of California Los Angeles Genomic Core using TMM software. The data were submitted to Gene Expression Omnibus (submission no.19922707). The mRNA data are presented as log2 fold change (log2FC). The expression levels are shown as log2 counts per million (log2CPM).
Quantitative PCR
To evaluate selected RNA-seq results, we analyzed key genes identified by RNA-seq using quantitative PCR (qPCR). Total RNA was isolated from cells, reverse transcribed, and evaluated by qPCR. qPCR was performed in triplicates, and three housekeeping genes were included as controls [glyceraldehyde 3-phosphate dehydrogenase (GAPDH), β-actin, and cyclophilin A]. All primers were from a Qiagen predesigned and tested primer assay kit. qPCR was carried out in a Qiagen Rotor-Gene Q. The data were analyzed by Rotor-Gene Q software.
Western blot
Cells were lysed in RIPA buffer with 1× Halt® protease inhibitors. Total lysate (20 μg protein) was separated in 10% mini-SDS gel and transferred to polyvinylidene difluoride membranes. The membranes were blotted with anti-HIF1α (1:500) (25), anti-HIF2α (1:500) (26), anti-hairless (1:500) (27), anti-KLF9 (1:250) (28), and anti-GAPDH (1:500) (29). Antibodies for HIF1α, HIF2α, and hairless were from Abcam, and those for KLF9 and GAPDH were from Santa Cruz Biotechnology.
Immunofluorescence imaging
Cells were grown in chamber slides and exposed to hypoxia for the indicated time period. Cells were then fixed with 4% paraformaldehyde for 30 minutes and permeabilized with 0.2% Triton X-100 in PBS containing 5% goat serum for 30 minutes. Following washing and blocking with 5% goat serum in PBS for 1 hour, the cells were incubated with primary antibody (1:50 dilution) overnight. The following day, the slides were washed and incubated with secondary antibody, Alexa Fluor, for 1 hour. The slides were mounted with ProLong Gold containing 4′,6-diamidino-2-phenylindole (DAPI). The antibodies used were anti-MAP2 (30), anti-KLF9 (28), and anti-hairless (27). The cells were imaged using Zeiss laser scanning microscopy.
Small interfering RNA knockdown
Rat fetal NSCs were plated in six-well dishes at a density of 3.5 × 105 cells per dish and grown in defined medium (Thermo Fisher Scientific) consisting of KnockOut™ DMEM/F-12 with Stem Pro® neural supplement, fibroblast growth factor, epidermal growth factor, and GlutaMAX-1. Cells were grown to 70% confluence and collected for transfection with small interfering RNA (siRNA) KLF9. The transfections were performed using Viromer Blue (Lipocalyx/OriGene BioTech) following the manufacturer’s instructions. In brief, siRNA was diluted in Buffer Blue to 2.8 μM for a total volume of 15 μL (tube A). In another tube three drops of Virome Blue was added to the wall, and 270 μL of Buffer Blue was added immediately and the tube was vortexed for 3 to 5 seconds (tube B). Solution from tube B (135 μL) was mixed with tube A solution and incubated for 15 minutes at room temperature to allow formation of the complex. The complex was distributed evenly to cells in a single well. The adequacy of knockdown was monitored after 48 hours. After confirming KLF9 knockdown, cells were differentiated for 7 days into neurons in differentiation medium (Thermo Fisher Scientific). Cells were then moved to a hypoxic chamber, with or without 5 nM T3 treatment, for the time periods indicated in the figures or figure legends. For each treatment condition, three replicates plates were used for RNA quantification.
Quantification of cell survival rate was performed using images of the cells in different treatment after KLF9 knockdown. Each treatment had three slides. The number of cells on each slide was counted from 6 consecutive fields, for a total of 18 fields from three slides. The survival rate is presented as the percentage of cells surviving relative to siControl cells grown in 21% oxygen condition.
Statistical analysis
One-way ANOVA was performed for all experimental data, except RNA-seq data. RNA-seq statistical analysis was done using TMM software. P < 0.05 was designated as statistically significant.
Results
T3 treatment protects primary cortical neurons exposed to hypoxia
We performed a T3 dose-response study to determine the optimal T3 concentration to protect primary cortical neurons. Primary cortical neurons were treated with 0, 1, 5, and 10 nM T3 and exposed to hypoxia (0.2% oxygen) for 7 hours. Neurons exposed to hypoxia either in the presence or absence of 1 nM T3 had shortened axons and axon breakage (Fig. 1A). With an increase in the dose of T3 treatment (5 and 10 nM), neuronal morphology of the hypoxic neurons was similar to control cells in normal oxygen concentrations, and axon breakage was not observed (Fig. 1B). These data indicated that T3 at doses of 5 to 10 nM reduced axon damage and prolonged neural cell life after hypoxic exposure. Based on these findings, T3 (5 nM) was used in the subsequent hypoxic neuron injury studies.
Hypoxia-induced damage in cultured primary mouse cortical neurons and response to progressive doses of T3. Primary cortical neural cells were cultured in poly-d-lysine–coated chamber slide dishes for 7 d prior to hypoxic exposure. (A) Control cells were grown in 21% oxygen (left) and the cells exposed to 0.2% oxygen for 7 h in the absence of T3 (−T3; right). (B) Cells were exposed to 0.2% oxygen in the presence of various concentrations of T3 for 7 h (for hypoxia treatment details, see “Materials and Methods”). After hypoxia, cells were fixed in the hypoxic chamber and stained with anti-Map2 antibody conjugated with Alexa Fluor 448 (green). DNA was stained with DAPI to show the nucleus (blue). Cells were imaged in a confocal microscope and assessed for axonal fragmentation and cell damage. −T3, without (0 nM) T3.
We next determined the duration of the T3 treatment required to protect hypoxic neurons. Cells were exposed to hypoxia for periods of 0 to 10 hours in the presence or absence of T3. In non–T3-treated cells, minor axonal breakage was observed after 4 to 6 hours of hypoxic exposure. After 8 hours of hypoxic exposure, axonal fragmentation was clearly seen. After 9 to 10 hours, apoptosis occurred (Fig. 2A and 2B). In T3-treated cells, minor axonal breakage was seen after 10 hours of hypoxia exposure, but signs of apoptosis were not observed. The data suggested that T3 treatment of hypoxic cells improved neural cell survival compared with cells grown in the absence of T3 and exposed to hypoxia.
Time course of hypoxic exposure to compare cell survival (A) without or (B) with T3 treatment. Primary cortical neurons were cultured in conditions as described in Fig. 1. T3 was added to the medium to a final concentration of 5 nM. Cells were exposed to hypoxia for 0 to 10 h. At end of each hour, cells were fixed, incubated with Map2 antibody conjugated with Alexa Fluor 448 (green), and imaged to evaluate axonal fragmentation and cell damage. DNA was stained with DAPI to show the nucleus (blue). The arrows indicate axon breakage. −T3, without T3.
T3 provides the optimal protection to neurons compared with Triac and its analog DITPA
T4 is transported to the brain primarily through MCT8. T4 is locally converted to T3 by type 2 deiodinase in glial cells and transported into neurons by MCT8. Brain injury leads to dysfunction of the blood–brain barrier (BBB). T4 transport is interrupted, resulting in inadequate TH in the brain, which may contribute to loss of neuroprotective effects of TH and impair the recovery after injury. DITPA and Triac do not require MCT8 to cross the BBB and may be superior to treatment with T3. To investigate this possibility, primary neural cells were treated with or without DITPA (20 μM) or Triac (40 nM) and exposed to hypoxia for 7 hours and compared with T3 treatment. Cells treated with Triac had shortened axons at 6 hours. After 7 hours, cells had severe axonal fragmentation and apoptosis was seen (Fig. 3). Cells treated with DITPA showed axonal breaking starting after 5 hours of hypoxic exposure and cell death after 7 hours. Cells treated with T3 showed axonal filament thinning after 6 to 7 hours of hypoxic exposure, but no sign of axon breaking or fragmentation. These data indicated that T3 was superior to Triac and DITPA with respect to neural protection.
Comparison of T3 to T3 analogs DITPA and Triac in neuronal protection from hypoxia. The culture and treatment conditions for primary cortical neuron culture, hypoxia exposure, and imaging are same as described in Fig. 2. For treatment, T3 (5 nM), 50 nM Triac (50 nM), and DITPA (20 μM) were added to the media, and the impact on axonal fragmentation and cell damage was assessed. Map2 is shown as green and DNA as blue.
TH treatment attenuated hypoxia-induced BCL2 downregulation
Apoptosis was detected in hypoxia-exposed neurons beginning with axonal breakage after 7 hours of hypoxia (Fig. 2). We analyzed activation of the apoptosis pathway genes after 7 hours of hypoxic exposure and compared levels with the gene expression obtained from control cells grown in normoxic conditions. The expression of antiapoptotic factors includes Bcl2, Bcl2L1, Bcl2L2, myeloid cell leukemia 1 (Mcl1), and Bcl2-associated athanogenes (Bags), and proapoptotic factors Bcl2L11, phorbol myristate acetate–induced protein 1 (Pmaip1), and caspase family genes were analyzed by RNA-seq. T3 treatment significantly increased the expression of Bcl2L2 and attenuated the reduction of Bcl2, compared with the condition without T3 treatment (Fig. 4A). Antiapoptotic factor Bag binds to BcL2 and enhances BcL2 antiapoptotic effects. Bag1 was significantly stimulated in T3-treated cells (Fig. 4A). Other Bag family genes were slightly increased but not significantly changed, compared with non–T3-treated cells. Proapoptotic factor Bcl2L11 (also known as Bim) is a hypoxia-responsive gene reported in several studies (31, 32). In cortical neurons, Bcl2L11 mRNA was induced by hypoxia 1.1 log2FC (abbreviated as fold below) and further increased to 1.8-fold by combined T3/hypoxia, compared with the normoxia condition. Although Bcl2L11 is a proapoptotic factor, studies have shown that its apoptotic activity can be neutralized by phosphorylation or sequestered by proapoptotic factors (33). Expression of Bcl2L1, Bcl2L2, and Mcl1 was increased in T3-treated cells. Bcl2-associated X (Bax) and Pmaip1 promoting activation of caspases were significantly lower in T3-treated cells (Fig. 4B). Other proapoptotic factors were not significantly changed in response to T3 treatment.
Analysis of apoptotic gene expression in neurons exposed to hypoxia with or without T3 treatment (+T3 or −T3, respectively). Primary cortical neurons were exposed to hypoxia for 7 h, RNA was isolated, and RNA-seq analysis was performed. For mRNA levels shown in (A) antiapoptotic genes, (B) proapoptotic genes, and (C) caspase genes, the threshold expression level was log2CPM >0.5. The mRNA change was shown as log2FC compared with control cells (set as 1 log2FC = 0). (D) Casp-3/7 staining of hypoxic neural cells. These cortical neurons were exposed to hypoxia for 9 h. Cells were then fixed and stained with Casp-3/7 using a Casp-3/7 detection kit. *P < 0.05, hypoxia/+T3 vs hypoxia/−T3. (E) Casp-3/7–positive cells were quantified by counting three continuous imaging frames from each slide (total of three individual slides). Data are presented as Casp-3/7–positive cells as a percentage of total cells counted. *P < 0.05, hypoxia/−T3 vs control or hypoxia/+T3.
Members of the caspase (Casp) gene family, Casp1, Casp4, and Casp8, were hardly detectable in primary cortical neurons. Casp9, Casp3, Casp7, and apoptosis-associated factor 1 (Apaf1) were significantly stimulated by hypoxia, but all of them were reduced with T3 treatment (Fig. 4C). With the exception of Casp6 mRNA, it was significantly elevated in T3-treated cells. Casp6 has been reported as a direct activator of Casp8. However, Casp8 mRNA level was barely detectable in cortical neurons after 7 hours of hypoxia either with or without T3 treatment. Although apoptosis was not trigged after 7 hours of hypoxia, there was a shift of increased proapoptotic caspase gene expression in hypoxic neurons without T3 presence. With continuous hypoxic exposure up to 9 hours, the Casp-3/7–positive cells were significantly increased by 56% in the absence of T3 compared with in the presence of T3 (Fig. 4D and 4E). T3 treatment resulted in a minimal increase of Casp-3/7 staining. Taken together, the mRNA profile of the genes in the mitochondria-mediated apoptosis pathway indicated that T3 treatment provides protection from apoptosis during hypoxic exposure.
T3 treatment promotes neuronal survival by enhancing 5-hmc and Tet
Hypoxia increases neuronal DNA methylation by increasing DNA 5-mc and Dnmt. The presence of 5-mc in CpG islands is a hallmark of DNA methylation, and conversion of 5-mc to 5-hmc reverses the DNA methylation. We analyzed global 5-hmc and 5-mc levels in primary neural cells. After 7 hours of hypoxic exposure, the DNA 5-mc level was increased 37% (P < 0.003) in the absence of T3, but only increased 13% (P < 0.064) in the presence of T3, compared with control cells grown in normoxia (Fig. 5A). Conversely, DNA 5-hmc was augmented 75% in the presence of T3 and 23% in the absence of T3, compared with controls (Fig. 5B). The 5-hmc level in neurons was also determined by immunofluorescence staining with anti–5-hmc antibody (Fig. 5C). Hypoxia resulted in both methylation and demethylation of gene promoters; however, T3 treatment induced greater demethylation, opposite to the hypoxia-induced changes.
T3 attenuates hypoxia-induced expression of factors that promote DNA methylation. The culture and treatment conditions for primary cortical neuron culture and hypoxia exposure are the same as described in Fig. 2. (A–C) After 7 h of exposure to hypoxia, (A and B) cells were harvested and analyzed for DNA 5-hmc and DNA 5-mc levels and (C) stained with 5-hmc antibody conjugated with Alexa Fluor 568 (orange-red). After merging with DAPI (blue), 5-hmc is shown as purple. Scale bar, 50 μM. *P < 0.05, hypoxia/with T3 (+T3) compared with hypoxia/without T3 (−T3) or control. (D and E) Tet and Dnmt gene expression was compared between hypoxia/−T3 and hypoxia/+T3. (F) The expression levels of Dnmt3a and Dnmt3b in neurons grown in normoxia were compared. *P < 0.05, hypoxia/+T3 vs hypoxia/−T3.
5-hmc conversion is catalyzed by ten-eleven translocation (TET) enzymes. After hypoxic exposure, Tet1 mRNA was increased 0.84-fold (P < 7.7 × 10−5), Tet2 0.37-fold (P < 1.1 × 10−4), and Tet3 0.67-fold (P < 1.6 × 10−5) in response to T3 treatment, compared with no T3 treatment (Fig. 5D). Tet is highly expressed and active during neural differentiation (34). Cultured primary neurons demonstrate neurite branching, elongating axons as is seen during neural differentiation. Hypoxic-induced reduction of Tet mRNA and protein levels leads to cessation of neural cell growth and fragmentation of the existing axons. Dnmts, responsible for DNA methylation, were downregulated in T3-treated cells. Dnmt1 mRNA was reduced 0.16-fold (P < 7.1 × 10−1), Dnmt3a 0.32-fold (P < 6.3 × 10−3), and Dnmt3b 0.23-fold (P < 1.1 × 10−3) in T3-treated cells compared with non–T3-treated cells (Fig. 5E). In cortical neurons, the mRNA levels of Dnmt3a were 7.1-fold (log2 scale) greater than those of Dnmt3b (Fig. 5F), suggesting that Dnmt3a has a dominant role in DNA de novo methylation. Even modest change of Dnmt3a may alter the DNA methylation status and subsequent gene expression patterns.
T3-mediated neural protection in hypoxia is associated with T3-induced gene expression
T3 induction of DNA 5-hmc and Tet mRNA may preferentially enhance T3-regulated genes. Hif2α (also known as Epas1) is a known thyroid hormone receptor (THR)–regulated gene, and the THR binding site has been located in the HIF2α promoter (35). T3 treatment of hypoxic neurons stimulated Hif2α 1.2-fold (P < 2.6 × 10−5), compared with hypoxia alone, and was increased 1.4-fold (P < 8.1 × 10−4) compared with control cells (Fig. 6A). The HIF2α protein was significantly increased after T3 treatment (Fig. 6B, lower panel) owing to both increased RNA level and hypoxia-induced degradation of prolyl hydroxylase, which stabilizes the HIF protein. Although the mRNA levels of both HIF1α and HIF1β were not significantly changed, HIF1α protein was increased due to hypoxic-induced protein stabilization (Fig. 6B, top panel). Some of the hypoxia-inducible genes are stimulated by both HIF2α and HIF1α, including vascular endothelial growth faction (Vegf), TGFβ (Tgfβ), enolase (Eno), Juns, Myc, neurophilin (Nrp), and integrin αvβ3 (Itgαv and Itgαβ3). Among them, Tgfβ1, Vegfα, Eno2, and Myc were significantly stimulated in T3-treated cells, compared with hypoxia without T3 treatment (Fig. 6C).
T3 enhances HIF2α mRNA and protein levels. The culture and treatment conditions for primary cortical neuron culture and hypoxia exposure are the same as described in Fig. 2. (A) Hif mRNA profile analyzed by RNA-seq. (B) Western blot analysis of HIF2α and HIF1α protein levels. (C) HIF-induced genes in hypoxia. *P < 0.05, hypoxia/+T3 vs hypoxia/−T3. −T3, without T3; +T3, with T3.
We subsequently examined the genes differentially regulated in hypoxic neurons treated with or without T3 compared with control neurons grown in normoxic conditions. T3 treatment modified the distribution of genes upregulated and downregulated in response to hypoxia (Fig. 7A and 7B). Thra mRNA was fivefold greater than Thrb in normoxia conditions. Hypoxic exposure reduced Thrb mRNA 50% logCPM (P < 2.1 × 10−4) either in the presence or absence of T3, and it stimulated Thra mRNA 18% logCPM (P < 1.3 × 10−2) and 15% logCPM (P < 2.3 × 10−2), compared with controls, respectively (Fig. 7C). We performed parallel analyses for selected genes that are important to this study by qPCR (36).
Identification of T3-regulated genes in hypoxia by RNA-seq. (A and B) Venn diagrams showing the number of genes that changed expression level in hypoxia exposure compared with control cells, using a threshold of log2FC >1.5, P < 0.05. (C) Differential responses of thyroid hormone receptor genes Thra and Thrb to hypoxia are shown as log2CPM. (D) T3 stimulation of its target genes was not affected by hypoxia. mRNA level was analyzed by RNA-seq. (E) Western blot detection of T3 targets KLF9 and hairless, which have the highest magnitude of induction by T3 after hypoxia. (F and G) Genes with the highest magnitude of (F) stimulation and (G) repression by T3 after hypoxia, which are identified using a stringent cutoff (i.e., log CPM ≥ 1, log FC ≥ 1.5, P < 0.05, FDR < 0.05). *P < 0.05. −T3, without T3; +T3, with T3.
To identify the genes most strongly regulated by T3 under hypoxic conditions, we applied stringent criteria in RNA-seq analysis with parameters of CPM ≥1, log FC ≥1.5, P value <0.05, and false discovery rate (FDR) <0.05. With these criteria, we identified 26 genes that were upregulated and 15 genes that were downregulated by T3, compared with expression without T3 treatment (Tables 1 and 2). Several known T3-regulated genes were highly stimulated by T3 in hypoxia, including hairless (Hr), KLF9 (Klf9), oxytocin receptor (Oxtr), and cytochrome P450 family 26 b1 (Cyp26b1) (Fig. 7D and 7E). These genes are known to be regulated by T3 in neural differentiation and brain development (37–43). Genes previously not know to be regulated by T3 were induced by the hypoxia/with T3 condition and barely detectable in controls and the hypoxia/without T3 condition, including Gm5643 coding for a predicted protein, ChrX and Baylor 18 (DxBay18), regulatory solute carrier 1a1 (Rsc1a1), and histone cluster 2, h3c2 (hist2h3c2) coding for a member of histone 3 family (Fig. 7F). The function of DxBay18 is not known. Based on the predicted structure, it should functionally associate with zinc finger proteins Zfp275 and Zfp92 (http://string-db.org). Rsc1a1 protein inhibits the both gene transcription and protein release of newly synthesized glucose-dependent Slc5A1 and Slc22A2 proteins from the trans-Golgi network (44–46).
Significantly Upregulated Genes for Hypoxia/+T3 vs Hypoxia/−T3 Conditions
| Gene Symbol | Gene Name | Log2FC +T3 vs −T3 | P Value | FDR |
|---|---|---|---|---|
| 5730405O15Rik | RIKEN cDNA 5730405O15 gene | 1.738252 | 4.67 × 10−3 | 1.11 × 10−2 |
| Aadat | Aminoadipate aminotransferase | 2.207737 | 1.07 × 10−5 | 1.82 × 10−24 |
| Atp8b4 | ATPase, class I, type 8B, member 4 | 2.710237 | 3.87 × 10−7 | 1.18 × 10−5 |
| B930025P03Rik | RIKEN cDNA B930025P03 gene | 1.515859 | 1.47 × 10−2 | 3.72 × 10−4 |
| Cpt2 | Carnitine palmitoyltransferase 2 | 1.622774 | 2.59 × 10−2 | 1.82 × 10−24 |
| Cyp26b1a | Cytochrome P450, family 26, subfamily b, polypeptide 1 | 2.018399 | 2.45 × 10−3 | 2.89 × 10−3 |
| Cys1 | Cystin 1 | 1.570307 | 2.58 × 10−2 | 2.89 × 10−3 |
| DXBay18 | DNA segment, chromosome X, Baylor 18 | 40.55547 | 1.54 × 10−8 | 2.41 × 10−3 |
| Eif3j2 | Eukaryotic translation initiation factor 3, subunit J2 | 2.214333 | 3.51 × 10−6 | 3.66 × 10−3 |
| Gm1821 | Predicted gene 1821 | 40.62769 | 1.02 × 10−18 | 2.90 × 10−2 |
| H2-T22 | Histocompatibility 2, T region locus 22 | 1.719636 | 3.70 × 10−2 | 3.72 × 10−4 |
| Hist1h2ag | Histone cluster 1, H2ag | 1.593244 | 1.79 × 10−2 | 4.74 × 10−12 |
| Hist2h3c2 | Histone cluster 2, H3c2 | 39.8925 | 3.01 × 10−8 | 1.77 × 10−3 |
| Hra | Hairless | 3.028367 | 6.96 × 10−7 | 3.09 × 10−3 |
| RSC1a1 | Regulatory solute carrier protein, family 1, member 1 | 584.71290 | 2.46 × 10−9 | 8.09 × 10−9 |
| Klf9a | Krüppel-like factor 9 | 2.2111 | 5.94 × 10−3 | 5.52 × 10−3 |
| Lgals3 | Lectin, galactose binding, soluble 3 | 1.693164 | 2.45 × 10−2 | 1.11 × 10−2 |
| Lrch4 | Leucine-rich repeats and calponin homology (CH) domain containing 4 | 1.6734 | 3.28 × 10−3 | 9.54 × 10−3 |
| Mc1r | Melanocortin 1 receptor | 1.769616 | 4.26 × 10−2 | 0.000725 |
| Oxtr1a | Oxytocin receptor 1 | 2.097431 | 2.86 × 10−2 | 2.03 × 10−5 |
| Prss53 | Protease, serine 53 | 2.207737 | 2.13 × 10−3 | 1.79 × 10−11 |
| Rbm20 | RNA binding motif protein 20 | 1.529665 | 4.00 × 10−3 | 7.25 × 10−4 |
| Slc35g1 | Solute carrier family 35, member G1 | 1.622774 | 3.25 × 10−2 | 2.03 × 10−5 |
| Snurf | SNRPN upstream reading frame | 2.029851 | 3.53 × 10−2 | 4.74 × 10−12 |
| Tcf7l1 | Transcription factor 7 like 1 (T cell specific, HMG box) | 1.567689 | 3.65 × 10−2 | 2.92 × 10−12 |
| Trim12c | Tripartite motif–containing 12C | 2.430129 | 3.87 × 10−7 | 1.03 × 10−15 |
| Gene Symbol | Gene Name | Log2FC +T3 vs −T3 | P Value | FDR |
|---|---|---|---|---|
| 5730405O15Rik | RIKEN cDNA 5730405O15 gene | 1.738252 | 4.67 × 10−3 | 1.11 × 10−2 |
| Aadat | Aminoadipate aminotransferase | 2.207737 | 1.07 × 10−5 | 1.82 × 10−24 |
| Atp8b4 | ATPase, class I, type 8B, member 4 | 2.710237 | 3.87 × 10−7 | 1.18 × 10−5 |
| B930025P03Rik | RIKEN cDNA B930025P03 gene | 1.515859 | 1.47 × 10−2 | 3.72 × 10−4 |
| Cpt2 | Carnitine palmitoyltransferase 2 | 1.622774 | 2.59 × 10−2 | 1.82 × 10−24 |
| Cyp26b1a | Cytochrome P450, family 26, subfamily b, polypeptide 1 | 2.018399 | 2.45 × 10−3 | 2.89 × 10−3 |
| Cys1 | Cystin 1 | 1.570307 | 2.58 × 10−2 | 2.89 × 10−3 |
| DXBay18 | DNA segment, chromosome X, Baylor 18 | 40.55547 | 1.54 × 10−8 | 2.41 × 10−3 |
| Eif3j2 | Eukaryotic translation initiation factor 3, subunit J2 | 2.214333 | 3.51 × 10−6 | 3.66 × 10−3 |
| Gm1821 | Predicted gene 1821 | 40.62769 | 1.02 × 10−18 | 2.90 × 10−2 |
| H2-T22 | Histocompatibility 2, T region locus 22 | 1.719636 | 3.70 × 10−2 | 3.72 × 10−4 |
| Hist1h2ag | Histone cluster 1, H2ag | 1.593244 | 1.79 × 10−2 | 4.74 × 10−12 |
| Hist2h3c2 | Histone cluster 2, H3c2 | 39.8925 | 3.01 × 10−8 | 1.77 × 10−3 |
| Hra | Hairless | 3.028367 | 6.96 × 10−7 | 3.09 × 10−3 |
| RSC1a1 | Regulatory solute carrier protein, family 1, member 1 | 584.71290 | 2.46 × 10−9 | 8.09 × 10−9 |
| Klf9a | Krüppel-like factor 9 | 2.2111 | 5.94 × 10−3 | 5.52 × 10−3 |
| Lgals3 | Lectin, galactose binding, soluble 3 | 1.693164 | 2.45 × 10−2 | 1.11 × 10−2 |
| Lrch4 | Leucine-rich repeats and calponin homology (CH) domain containing 4 | 1.6734 | 3.28 × 10−3 | 9.54 × 10−3 |
| Mc1r | Melanocortin 1 receptor | 1.769616 | 4.26 × 10−2 | 0.000725 |
| Oxtr1a | Oxytocin receptor 1 | 2.097431 | 2.86 × 10−2 | 2.03 × 10−5 |
| Prss53 | Protease, serine 53 | 2.207737 | 2.13 × 10−3 | 1.79 × 10−11 |
| Rbm20 | RNA binding motif protein 20 | 1.529665 | 4.00 × 10−3 | 7.25 × 10−4 |
| Slc35g1 | Solute carrier family 35, member G1 | 1.622774 | 3.25 × 10−2 | 2.03 × 10−5 |
| Snurf | SNRPN upstream reading frame | 2.029851 | 3.53 × 10−2 | 4.74 × 10−12 |
| Tcf7l1 | Transcription factor 7 like 1 (T cell specific, HMG box) | 1.567689 | 3.65 × 10−2 | 2.92 × 10−12 |
| Trim12c | Tripartite motif–containing 12C | 2.430129 | 3.87 × 10−7 | 1.03 × 10−15 |
Cutoff stringency: log FC ≥ 1.5, CPM ≥ 1, P < 0.05, FDR < 0.05.
Abbreviations: +T3, with T3; −T3, without T3.
Known T3 targets.
Significantly Upregulated Genes for Hypoxia/+T3 vs Hypoxia/−T3 Conditions
| Gene Symbol | Gene Name | Log2FC +T3 vs −T3 | P Value | FDR |
|---|---|---|---|---|
| 5730405O15Rik | RIKEN cDNA 5730405O15 gene | 1.738252 | 4.67 × 10−3 | 1.11 × 10−2 |
| Aadat | Aminoadipate aminotransferase | 2.207737 | 1.07 × 10−5 | 1.82 × 10−24 |
| Atp8b4 | ATPase, class I, type 8B, member 4 | 2.710237 | 3.87 × 10−7 | 1.18 × 10−5 |
| B930025P03Rik | RIKEN cDNA B930025P03 gene | 1.515859 | 1.47 × 10−2 | 3.72 × 10−4 |
| Cpt2 | Carnitine palmitoyltransferase 2 | 1.622774 | 2.59 × 10−2 | 1.82 × 10−24 |
| Cyp26b1a | Cytochrome P450, family 26, subfamily b, polypeptide 1 | 2.018399 | 2.45 × 10−3 | 2.89 × 10−3 |
| Cys1 | Cystin 1 | 1.570307 | 2.58 × 10−2 | 2.89 × 10−3 |
| DXBay18 | DNA segment, chromosome X, Baylor 18 | 40.55547 | 1.54 × 10−8 | 2.41 × 10−3 |
| Eif3j2 | Eukaryotic translation initiation factor 3, subunit J2 | 2.214333 | 3.51 × 10−6 | 3.66 × 10−3 |
| Gm1821 | Predicted gene 1821 | 40.62769 | 1.02 × 10−18 | 2.90 × 10−2 |
| H2-T22 | Histocompatibility 2, T region locus 22 | 1.719636 | 3.70 × 10−2 | 3.72 × 10−4 |
| Hist1h2ag | Histone cluster 1, H2ag | 1.593244 | 1.79 × 10−2 | 4.74 × 10−12 |
| Hist2h3c2 | Histone cluster 2, H3c2 | 39.8925 | 3.01 × 10−8 | 1.77 × 10−3 |
| Hra | Hairless | 3.028367 | 6.96 × 10−7 | 3.09 × 10−3 |
| RSC1a1 | Regulatory solute carrier protein, family 1, member 1 | 584.71290 | 2.46 × 10−9 | 8.09 × 10−9 |
| Klf9a | Krüppel-like factor 9 | 2.2111 | 5.94 × 10−3 | 5.52 × 10−3 |
| Lgals3 | Lectin, galactose binding, soluble 3 | 1.693164 | 2.45 × 10−2 | 1.11 × 10−2 |
| Lrch4 | Leucine-rich repeats and calponin homology (CH) domain containing 4 | 1.6734 | 3.28 × 10−3 | 9.54 × 10−3 |
| Mc1r | Melanocortin 1 receptor | 1.769616 | 4.26 × 10−2 | 0.000725 |
| Oxtr1a | Oxytocin receptor 1 | 2.097431 | 2.86 × 10−2 | 2.03 × 10−5 |
| Prss53 | Protease, serine 53 | 2.207737 | 2.13 × 10−3 | 1.79 × 10−11 |
| Rbm20 | RNA binding motif protein 20 | 1.529665 | 4.00 × 10−3 | 7.25 × 10−4 |
| Slc35g1 | Solute carrier family 35, member G1 | 1.622774 | 3.25 × 10−2 | 2.03 × 10−5 |
| Snurf | SNRPN upstream reading frame | 2.029851 | 3.53 × 10−2 | 4.74 × 10−12 |
| Tcf7l1 | Transcription factor 7 like 1 (T cell specific, HMG box) | 1.567689 | 3.65 × 10−2 | 2.92 × 10−12 |
| Trim12c | Tripartite motif–containing 12C | 2.430129 | 3.87 × 10−7 | 1.03 × 10−15 |
| Gene Symbol | Gene Name | Log2FC +T3 vs −T3 | P Value | FDR |
|---|---|---|---|---|
| 5730405O15Rik | RIKEN cDNA 5730405O15 gene | 1.738252 | 4.67 × 10−3 | 1.11 × 10−2 |
| Aadat | Aminoadipate aminotransferase | 2.207737 | 1.07 × 10−5 | 1.82 × 10−24 |
| Atp8b4 | ATPase, class I, type 8B, member 4 | 2.710237 | 3.87 × 10−7 | 1.18 × 10−5 |
| B930025P03Rik | RIKEN cDNA B930025P03 gene | 1.515859 | 1.47 × 10−2 | 3.72 × 10−4 |
| Cpt2 | Carnitine palmitoyltransferase 2 | 1.622774 | 2.59 × 10−2 | 1.82 × 10−24 |
| Cyp26b1a | Cytochrome P450, family 26, subfamily b, polypeptide 1 | 2.018399 | 2.45 × 10−3 | 2.89 × 10−3 |
| Cys1 | Cystin 1 | 1.570307 | 2.58 × 10−2 | 2.89 × 10−3 |
| DXBay18 | DNA segment, chromosome X, Baylor 18 | 40.55547 | 1.54 × 10−8 | 2.41 × 10−3 |
| Eif3j2 | Eukaryotic translation initiation factor 3, subunit J2 | 2.214333 | 3.51 × 10−6 | 3.66 × 10−3 |
| Gm1821 | Predicted gene 1821 | 40.62769 | 1.02 × 10−18 | 2.90 × 10−2 |
| H2-T22 | Histocompatibility 2, T region locus 22 | 1.719636 | 3.70 × 10−2 | 3.72 × 10−4 |
| Hist1h2ag | Histone cluster 1, H2ag | 1.593244 | 1.79 × 10−2 | 4.74 × 10−12 |
| Hist2h3c2 | Histone cluster 2, H3c2 | 39.8925 | 3.01 × 10−8 | 1.77 × 10−3 |
| Hra | Hairless | 3.028367 | 6.96 × 10−7 | 3.09 × 10−3 |
| RSC1a1 | Regulatory solute carrier protein, family 1, member 1 | 584.71290 | 2.46 × 10−9 | 8.09 × 10−9 |
| Klf9a | Krüppel-like factor 9 | 2.2111 | 5.94 × 10−3 | 5.52 × 10−3 |
| Lgals3 | Lectin, galactose binding, soluble 3 | 1.693164 | 2.45 × 10−2 | 1.11 × 10−2 |
| Lrch4 | Leucine-rich repeats and calponin homology (CH) domain containing 4 | 1.6734 | 3.28 × 10−3 | 9.54 × 10−3 |
| Mc1r | Melanocortin 1 receptor | 1.769616 | 4.26 × 10−2 | 0.000725 |
| Oxtr1a | Oxytocin receptor 1 | 2.097431 | 2.86 × 10−2 | 2.03 × 10−5 |
| Prss53 | Protease, serine 53 | 2.207737 | 2.13 × 10−3 | 1.79 × 10−11 |
| Rbm20 | RNA binding motif protein 20 | 1.529665 | 4.00 × 10−3 | 7.25 × 10−4 |
| Slc35g1 | Solute carrier family 35, member G1 | 1.622774 | 3.25 × 10−2 | 2.03 × 10−5 |
| Snurf | SNRPN upstream reading frame | 2.029851 | 3.53 × 10−2 | 4.74 × 10−12 |
| Tcf7l1 | Transcription factor 7 like 1 (T cell specific, HMG box) | 1.567689 | 3.65 × 10−2 | 2.92 × 10−12 |
| Trim12c | Tripartite motif–containing 12C | 2.430129 | 3.87 × 10−7 | 1.03 × 10−15 |
Cutoff stringency: log FC ≥ 1.5, CPM ≥ 1, P < 0.05, FDR < 0.05.
Abbreviations: +T3, with T3; −T3, without T3.
Known T3 targets.
Significantly Downregulated Genes for Hypoxia/+T3 vs Hypoxia/−T3 Conditions
| Gene Symbol | Gene Name | LogFC (+T3 vs –T3) | P Value | FDR |
|---|---|---|---|---|
| Amd1 | S-adenosylmethionine decarboxylase 1 | −8.78436 | 6.28 × 10−8 | 4.06 × 10−5 |
| Ass1 | Argininosuccinate synthetase 1 | −41.5177 | 3.26 × 10−2 | 9.15 × 10−4 |
| Cdca4 | Cell division cycle–associated 4 | −8.7824 | 4.34 × 10−3 | 3.21 × 10−2 |
| Dcun1d1 | DCN1, defective in cullin neddylation 1, domain containing 1 (Saccharomyces cerevisiae) | −2.1846 | 2.35 × 10−3 | 5.58 × 10−3 |
| Foxr2 | Forkhead box R2 | −1.79226 | 3.20 × 10−3 | 5.27 × 10−3 |
| Gm3500 | Predicted gene 3500 | −40.4921 | 2.59 × 10−2 | 5.65 × 10−4 |
| Gm6644 | Predicted gene 6644 | −1.82046 | 4.51 × 10−4 | 2.17 × 10−4 |
| Hist1h2ap | Histone cluster 1, H2ap | −3.52697 | 1.48 × 10−6 | 6.70 × 10−15 |
| Hist2h2aa2 | Histone cluster 2, H2aa2 | −40.4921 | 2.46 × 10−9 | 5.65 × 10−3 |
| Il4i1 | IL-4–induced 1 | −40.0771 | 7.81 × 10−20 | 6.59 × 10−5 |
| Lgals4 | Lectin, galactose binding, soluble 4 | −1.76954 | 4.46 × 10−4 | 8.94 × 10−5 |
| Prr16 | Proline-rich 16 | −1.96219 | 3.31 × 10−5 | 1.55 × 10−4 |
| Taf4b | TAF4B RNA polymerase II, TATA box–binding protein (TBP)-associated factor | −2.83666 | 2.20 × 10−6 | 3.33 × 10−6 |
| Tmem132c | Transmembrane protein 132C | −1.69915 | 1.93 × 10−2 | 3.01 × 10−4 |
| Ttr | Transthyretin | −1.74646 | 1.01 × 10−2 | 6.12 × 10−5 |
| Gene Symbol | Gene Name | LogFC (+T3 vs –T3) | P Value | FDR |
|---|---|---|---|---|
| Amd1 | S-adenosylmethionine decarboxylase 1 | −8.78436 | 6.28 × 10−8 | 4.06 × 10−5 |
| Ass1 | Argininosuccinate synthetase 1 | −41.5177 | 3.26 × 10−2 | 9.15 × 10−4 |
| Cdca4 | Cell division cycle–associated 4 | −8.7824 | 4.34 × 10−3 | 3.21 × 10−2 |
| Dcun1d1 | DCN1, defective in cullin neddylation 1, domain containing 1 (Saccharomyces cerevisiae) | −2.1846 | 2.35 × 10−3 | 5.58 × 10−3 |
| Foxr2 | Forkhead box R2 | −1.79226 | 3.20 × 10−3 | 5.27 × 10−3 |
| Gm3500 | Predicted gene 3500 | −40.4921 | 2.59 × 10−2 | 5.65 × 10−4 |
| Gm6644 | Predicted gene 6644 | −1.82046 | 4.51 × 10−4 | 2.17 × 10−4 |
| Hist1h2ap | Histone cluster 1, H2ap | −3.52697 | 1.48 × 10−6 | 6.70 × 10−15 |
| Hist2h2aa2 | Histone cluster 2, H2aa2 | −40.4921 | 2.46 × 10−9 | 5.65 × 10−3 |
| Il4i1 | IL-4–induced 1 | −40.0771 | 7.81 × 10−20 | 6.59 × 10−5 |
| Lgals4 | Lectin, galactose binding, soluble 4 | −1.76954 | 4.46 × 10−4 | 8.94 × 10−5 |
| Prr16 | Proline-rich 16 | −1.96219 | 3.31 × 10−5 | 1.55 × 10−4 |
| Taf4b | TAF4B RNA polymerase II, TATA box–binding protein (TBP)-associated factor | −2.83666 | 2.20 × 10−6 | 3.33 × 10−6 |
| Tmem132c | Transmembrane protein 132C | −1.69915 | 1.93 × 10−2 | 3.01 × 10−4 |
| Ttr | Transthyretin | −1.74646 | 1.01 × 10−2 | 6.12 × 10−5 |
Cutoff stringency: log FC > 1.5, CPM > 1, P < 0.05, FDR < 0.05.
Abbreviations: +T3, with T3; −T3, without T3.
Significantly Downregulated Genes for Hypoxia/+T3 vs Hypoxia/−T3 Conditions
| Gene Symbol | Gene Name | LogFC (+T3 vs –T3) | P Value | FDR |
|---|---|---|---|---|
| Amd1 | S-adenosylmethionine decarboxylase 1 | −8.78436 | 6.28 × 10−8 | 4.06 × 10−5 |
| Ass1 | Argininosuccinate synthetase 1 | −41.5177 | 3.26 × 10−2 | 9.15 × 10−4 |
| Cdca4 | Cell division cycle–associated 4 | −8.7824 | 4.34 × 10−3 | 3.21 × 10−2 |
| Dcun1d1 | DCN1, defective in cullin neddylation 1, domain containing 1 (Saccharomyces cerevisiae) | −2.1846 | 2.35 × 10−3 | 5.58 × 10−3 |
| Foxr2 | Forkhead box R2 | −1.79226 | 3.20 × 10−3 | 5.27 × 10−3 |
| Gm3500 | Predicted gene 3500 | −40.4921 | 2.59 × 10−2 | 5.65 × 10−4 |
| Gm6644 | Predicted gene 6644 | −1.82046 | 4.51 × 10−4 | 2.17 × 10−4 |
| Hist1h2ap | Histone cluster 1, H2ap | −3.52697 | 1.48 × 10−6 | 6.70 × 10−15 |
| Hist2h2aa2 | Histone cluster 2, H2aa2 | −40.4921 | 2.46 × 10−9 | 5.65 × 10−3 |
| Il4i1 | IL-4–induced 1 | −40.0771 | 7.81 × 10−20 | 6.59 × 10−5 |
| Lgals4 | Lectin, galactose binding, soluble 4 | −1.76954 | 4.46 × 10−4 | 8.94 × 10−5 |
| Prr16 | Proline-rich 16 | −1.96219 | 3.31 × 10−5 | 1.55 × 10−4 |
| Taf4b | TAF4B RNA polymerase II, TATA box–binding protein (TBP)-associated factor | −2.83666 | 2.20 × 10−6 | 3.33 × 10−6 |
| Tmem132c | Transmembrane protein 132C | −1.69915 | 1.93 × 10−2 | 3.01 × 10−4 |
| Ttr | Transthyretin | −1.74646 | 1.01 × 10−2 | 6.12 × 10−5 |
| Gene Symbol | Gene Name | LogFC (+T3 vs –T3) | P Value | FDR |
|---|---|---|---|---|
| Amd1 | S-adenosylmethionine decarboxylase 1 | −8.78436 | 6.28 × 10−8 | 4.06 × 10−5 |
| Ass1 | Argininosuccinate synthetase 1 | −41.5177 | 3.26 × 10−2 | 9.15 × 10−4 |
| Cdca4 | Cell division cycle–associated 4 | −8.7824 | 4.34 × 10−3 | 3.21 × 10−2 |
| Dcun1d1 | DCN1, defective in cullin neddylation 1, domain containing 1 (Saccharomyces cerevisiae) | −2.1846 | 2.35 × 10−3 | 5.58 × 10−3 |
| Foxr2 | Forkhead box R2 | −1.79226 | 3.20 × 10−3 | 5.27 × 10−3 |
| Gm3500 | Predicted gene 3500 | −40.4921 | 2.59 × 10−2 | 5.65 × 10−4 |
| Gm6644 | Predicted gene 6644 | −1.82046 | 4.51 × 10−4 | 2.17 × 10−4 |
| Hist1h2ap | Histone cluster 1, H2ap | −3.52697 | 1.48 × 10−6 | 6.70 × 10−15 |
| Hist2h2aa2 | Histone cluster 2, H2aa2 | −40.4921 | 2.46 × 10−9 | 5.65 × 10−3 |
| Il4i1 | IL-4–induced 1 | −40.0771 | 7.81 × 10−20 | 6.59 × 10−5 |
| Lgals4 | Lectin, galactose binding, soluble 4 | −1.76954 | 4.46 × 10−4 | 8.94 × 10−5 |
| Prr16 | Proline-rich 16 | −1.96219 | 3.31 × 10−5 | 1.55 × 10−4 |
| Taf4b | TAF4B RNA polymerase II, TATA box–binding protein (TBP)-associated factor | −2.83666 | 2.20 × 10−6 | 3.33 × 10−6 |
| Tmem132c | Transmembrane protein 132C | −1.69915 | 1.93 × 10−2 | 3.01 × 10−4 |
| Ttr | Transthyretin | −1.74646 | 1.01 × 10−2 | 6.12 × 10−5 |
Cutoff stringency: log FC > 1.5, CPM > 1, P < 0.05, FDR < 0.05.
Abbreviations: +T3, with T3; −T3, without T3.
Genes highly downregulated by T3 in hypoxic conditions were identified, including adenosylmethionine decarboxylase 1 (Adm1), a key enzyme for polyamine biosynthesis, argininosuccinate synthetase 1 (Ass1), histone cluster 2, H3c1 (Hist2haa2), IL-4–induced 1 (iL4i1), cell division cycle–associated 4 (Cdc4), and Gm 3500 (Fig. 7G). These genes may play important roles in T3-mediated neural cell protection after hypoxia.
The T3-mediated Klf9 gene is a critical factor for neural cell survival in hypoxia
KLF9 has been shown to interact with multiple transcription factors and regulates >1000 genes during brain development (18, 19, 38). KLF9 promotes a number of processes, including T3-mediated neurite outgrowth and Purkinje cell survival, maturation, and myelination (47, 48). We found that T3 treatment induced several KLF9-mediated genes in hypoxia, including adenomatous polyposis coli 2 (Apc2; a negative regulator for β-catenin), nuclear receptor corepressor 2 (Ncor2)/SMRT, TH receptor interactor 4 (Trip4), and transfer RNA nucleotidyltransferase 1 (Trnt1) associated with mitochondria OXPHOS (18) (Fig. 8A). Several known KLF9-repressed genes were differentially affected by hypoxia and the hypoxia/with T3 condition. Klf11 and Klf13, important for cell growth and erythropoiesis, were further downregulated by T3. Glucocorticoid receptor (Nr3c1) and G protein–coupled receptor 161 (Gpr161) regulates neural tube development and were also downregulated by T3 (Fig. 8B).
KLF9 knockdown diminishes T3-mediated neural protective effects. (A and B) KLF9 regulates a range of genes in primary cortical neurons in response to hypoxia, including those that are (A) stimulated and (B) repressed. *P < 0.05, hypoxia/+T3 vs hypoxia/−T3. (C and D) Primary rat NSCs were transfected with siRNA Kilf9 or siRNA control (see “Materials and Methods” for transfection reagents). Two days after transfection, the kif9 RNA knockdown was confirmed by (C) fluorescence imaging with anti-KLF9 antibody and (D) qPCR. *P < 0.05, siklf9 knockdown compared with control and siRNA negative control (siNC). (E) Cells were then differentiated for 7 d and subsequently exposed to hypoxia for 7 h. After exposure, cells were fixed and imaged with anti-Map2 antibody conjugated with Alexa Fluor 488. (F) Quantification of cell survival rate. For details, see “Materials and Methods.” *P < 0.05, statistical significance within the comparison groups, as indicated by each line, and all conditions compared to 21% oxygen control (top line). −T3, without T3; +T3, with T3.
To investigate the role of KLF9 in T3-mediated neuronal protection, we performed gene siRNA Klf9 knockdown in primary rat NSCs. We used rat NSCs in siRNA knockdown experiments, rather than mouse primary cortical neurons, because unlike primary neurons, they can proliferate after siRNA knockdown. After KLF9 silencing, cells were grown under proliferation conditions for 2 days. Klf9 knockdown was confirmed by immunofluorescence staining of KLF9 protein and qPCR analysis of Kif9 mRNA (Fig. 8C and 8D). Cells were then differentiated for 7 days. KLF9 knockdown cells had greater reduction in axon branching, compared with siRNA control cells in control and hypoxic neurons (Fig. 8E). This is in agreement with previous studies indicating that KLF9 controls axon growth (47, 49). In siKFL9 knockdown cells, axon dissolving and apoptosis was observed after 7 hours of hypoxic exposure, with or without T3 treatment (Fig. 8E, lower penal). The quantification data (Fig. 8F) indicated that, after KLF9 knockdown, only 17% of cells survived hypoxic exposure, either with or without T3 treatment, which was a significant reduction compared with the siControl cells (48%). These data demonstrated that KLF9 knockdown completely abolished the protective effects of T3 on primary neurons after hypoxic exposure. These data also support that KLF9 confers a T3-mediated protective effects in hypoxic neurons, likely due to both direct regulation of gene expression, as well as effects on downstream pathways.
Discussion
We have demonstrated that T3 treatment prolongs neuronal survival of cultured primary cortical neurons during hypoxic exposure. We showed that T3 treatment promoted demethylation of DNA, with increased 5-hmc levels and Tet mRNA, as well as reciprocally downregulated mRNA expression of Dnmt3a and Dnmt3b. These data suggest that T3 treatment attenuates hypoxia-induced DNA de novo methylation, which ultimately modified the pattern of gene expression, as shown in the RNA-seq analysis. T3 enhanced Hif2α mRNA and protein, resulting in enhanced expression of several transcription factors.
We determined the effects of T3 treatment on apoptosis at the mRNA level and on mRNA expression of proapoptotic and antiapoptotic genes, as well as caspase genes. Most antiapoptotic genes were stimulated by T3, compared with control cells without T3 treatment. T3 effects on the proapoptotic genes were mixed. Proapoptotic genes Bcl2l11 is a hypoxia-inducible gene, which was significant upregulated by T3. It has been reported that Bcl2l11 is involved in FOXO3a-iinduced reactive oxygen species (50). It is known that T3 treatment increase basal respiration and reactive oxygen species. Nevertheless, apoptosis was not observed after 7 hours of hypoxic exposure with or without T3 treatment. Although hypoxia induced caspases, the level of caspase expression was not sufficient to trigger the apoptosis program after hypoxic exposure for 7 hours. After 10 hours of hypoxic exposure, mitochondria-mediated apoptosis was observed (Fig. 2).
Brain injury is associated with dysfunction of BBB, resulting in reduced transport of TH into the brain (51–54). We previously reported downregulation of the Mct8 gene in the brain in response to injury (13). Triac and DIPA do not require MCT8 for transport through the BBB or into the neuron and have the potential to enhance hormone delivery to the brain. We studied the protective role of DITPA and Triac after hypoxia. Treatment with Triac and DITPA was inferior to T3 in neural protection of hypoxia-induced neural injury in cultured cortical neuron cells. Triac has higher selectivity for THRB, approximately threefold higher than that for THRA (22). However, in hypoxic cortical neurons, the THRA mRNA level is approximately fivefold (in log2 scal2) higher than that in THRB mRNA. Furthermore, hypoxia significantly reduced THRB mRNA. This may explain why Triac treatment did not provide sufficient protection to neurons. DITPA is a T3 analog that binds THRA and THRB with almost identical affinity (55) and can also act via a nongenomic mechanism (56). In the brain of MCT8−/− mice, only treatment with high-dose DITPA normalized the T3 target gene Hr mRNA level and other T3 target gene expression (57). Both drugs have been used in pilot studies involving patients with Mct8 gene mutations, with improved pituitary feedback, reduction in serum T3 concentration, and improved metabolic manifestations, but with limited benefit in reversing brain abnormalities (23, 57).
KLF9 is an important transcription factor expressed during central nervous system development. Chromatin immunoprecipitation sequencing has identified thousands of genes that bind KLF9 in the region 1 kb from the transcription start site (18). KLF9 can function as a corepressor and coactivator. KLF9 recruits mSin3A for repression and p300/CBP and p300/CBP-associated factor (PCAF) for activation of gene transcription (58). Mice lacking KLF9 showed impaired learning and motor skills (59). In our studies, knockdown of KLF9 in rat NSCs resulted in poor differentiation and inhibition of neurite branching. Our observations are similar to those previously reported that KLF9 is required for T3-stimulated neurite outgrowth (42). The T3-mediated beneficial effects on neuron survival after injury were eliminated after knockdown of Klf9. Our data suggest that KLF9 is essential for a T3-mediated protective role in neurons. This idea is worth pursuing further in KLF9-mediated pathways.
Rodent models of traumatic brain injury have consistently shown a protective benefit of TH when administered up to 1 hour after the brain injury (13, 15, 16). The mechanism that produces this benefit, however, has not been shown. Although the in vitro hypoxia model does not reproduce all elements of traumatic brain injury, it models neuronal damage, and the protective effect of TH was robust. T3 treatment protects hypoxic neuronal injury by reducing DNA methylation, stimulating Klf9 expression and stimulating antiapoptotic pathways.
Acknowledgments
Financial Support: This work was supported by US Department of Veterans Affairs Merit Review Grant BX001966 (to G.A.B.).
Additional Information
Current Affiliation: J. Li’s current affiliation is Department of Endocrinology, Union Hospital, Fujian Medical University, Fuzhou, Fujian, China.
Disclosure Summary: The authors have nothing to disclose.
Data Availability:
All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.
Abbreviations:
- 5-hmc
5-hydroxymethylcytosine
- 5-mc
5-methylcytosine
- BBB
blood–brain barrier
- Casp
caspase
- DAPI
4′,6-diamidino-2-phenylindole
- DITPA
diiodothyropropionic acid
- Dnmt
DNA methyltransferase
- FDR
false discovery rate
- GAPDH
glyceraldehyde 3-phosphate dehydrogenase
- KLF9
Krüppel-like factor 9
- log2CPM
log2 counts per minute
- log2FC
log2 fold change
- MCT8
monocarboxylate transporter 8
- NSC
neural stem cell
- qPCR
quantitative PCR
- RNA-seq
RNA sequencing
- siRNA
small interfering RNA
- TBI
traumatic brain injury
- TET
ten-eleven translocation
- TH
thyroid hormone
- THR
thyroid hormone receptor
- Triac
3,5,3′-triiodothyroacetic acid
References and Notes
Author notes
J.L. and K.A. contributed equally to this work.
