Early-Life m6A RNA Demethylation by Fat Mass and Obesity-Associated Protein (FTO) Influences Resilience or Vulnerability to Heat Stress Later in Life

Visual Abstract


Introduction
The exposure to stressful experiences, including heat stress, during the critical sensory development period at early life can change neural architecture and modulate the stress response set point leading to either stress resilience or vulnerability later in life (Franklin et al., 2012;Braun et al., 2017;van Bodegom et al., 2017;Lux, 2018). Thermal control set point is regulated by thermosensitive neurons of the preoptic anterior hypothalamus (PO/AH; Boulant, 2006;Wang et al., 2019). Thermal input during the critical period of thermal-control establishment causes a plastic change in the ratio between thermosensitive neurons and innate PO/AH cells and can modulate temperature tolerance (Tzschentke and Basta, 2002). In chicks, the critical period of thermal-control establishment is between days 3 and 5 posthatch (Yahav and Hurwitz, 1996). Chicks exposed to moderate heat conditioning on day 3 posthatch displayed heat resilience during heat challenge on day 10, whereas chicks conditioned at high ambient temperature were vulnerable to heat stress (Kisliouk et al., 2014(Kisliouk et al., , 2017Cramer et al., 2015Cramer et al., , 2019. We have previously shown that epigenetic mechanisms, such as DNA methylation, histone modifications, and miRNAs, mediate the long-term effect of these stressful experiences in driving experience-dependent gene expression in the AH, underlying the heat stress memory formation (Yossifoff et al., 2008;Kisliouk and Meiri, 2009;Kisliouk et al., 2010Kisliouk et al., , 2011Kisliouk et al., , 2014Kisliouk et al., , 2017Cramer et al., 2015Cramer et al., , 2019. Recent evidence suggests additional, epitranscriptomic level of regulation, conferring further flexibility to fine-tune gene expression on top of the epigenetic one (Cao et al., 2016;Roignant and Soller, 2017;Widagdo and Anggono, 2018;Shi et al., 2019). Like the epigenetic code surrounding DNA, RNA modifications located in coding sequences of mRNAs can influence the fate of RNA and thereby serve as potential regulators of mRNA expression (Gilbert et al., 2016;Nainar et al., 2016). N 6 -methyladenosin (m 6 A) is the most abundant and the best-characterized internal mRNA modification playing an important regulatory role in various aspects of mRNA metabolism including the splicing, nuclear export, transcription, translation, and decay (Cao et al., 2016;Gilbert et al., 2016;Roignant and Soller, 2017;Widagdo and Anggono, 2018). Majority of m 6 A methylation on mRNA is installed by a methyltransferase complex including methyltransferase-like 3 and 14 (METTL3 and METTL14) and removed by the demethylases, fat mass and obesity-associated protein (FTO), and ALKB homolog 5 (ALKBH5; Roignant and Soller, 2017;Widagdo and Anggono, 2018;Du et al., 2019;Shi et al., 2019). Cellular functions of m 6 A modification are exerted by its direct recognition by m 6 A-specific binding proteins including YTH-domain containing proteins, heterogeneous nuclear ribonucleoprotein (HNRNP), and common RNA binding proteins like IGF2BP1-3 and FMR1 (Du et al., 2019;Shi et al., 2019). The m 6 A RNA modification is highly distributed in the brain (Meyer et al., 2012;Chang et al., 2017), and it was found to be essential regulatory element in various biological processes concerning neural development (Li et al., 2017;Yoon et al., 2017), synaptic plasticity (Chang et al., 2017;Merkurjev et al., 2018;Yu et al., 2018a), neurodegeneration Chen et al., 2019b), and axon regeneration . Imbalanced m 6 A RNA methylation was shown to affect diverse brain physiological functions, including learning, and memory (Widagdo et al., 2016;Walters et al., 2017;Shi et al., 2018;Zhang et al., 2018), addiction and reward (Hess et al., 2013;Ruud et al., 2019), and stress response (Engel et al., 2018). Here, we suggest a role for m 6 A RNA methylation in chick AH in long-term regulation of heat stress response.

Bird housing
Male Cobb chicks were obtained on the first day of life from Brown Hatcheries and raised in climate-controlled rooms at 30°C under continuous artificial illumination with ad libitum access to food and water. All experiments were performed according to the guidelines of the European Community Council and approved by the Volcani Institute Committee for Animal Use in Research.

Heat treatment and tissue collection
Heat conditioning was performed on day 3 posthatch. The chicks were arbitrarily divided into two groups and transferred into either 36°C or 40°C preheated rooms for 24 h, giving "mild-temperature-conditioned chicks" and "hightemperature-conditioned chicks," respectively. Body temperature was measured for each group of chicks using a digital thermometer (Extech Instruments) with 60.1°C accuracy that was inserted 1.5 cm into the cloaca. Chicks were killed by decapitation 6 and 24 h into the thermal treatment conditioning. Non-treated age-matched chicks served as controls. The brain areas matching the AH, the mesopallium intermediomediale (IMM), and the frontal area of the brain (FB; schematically represented in Fig.  1B) were dissected from the whole brain. In brief, the skull was cut along the l suture and the sagittal suture, the brain was removed from the skull and set on a small plastic cube with the lateral side upwards. First, the AH was dissected. The boundaries of the dissections were determined by the optic chiasma and the clear boundaries of the hypothalamus (A 8-10 L 0-1.4 on both hemispheres coordinates according to Kuenzel, 1988). After dissection, the tissues were immediately immersed in RNALater (Ambion).
Heat challenge was performed on day 10 posthatch. Both experimental chick groups and their naive counterparts were thermally challenged by exposure to 36°C for 6 h. The chicks were measured for body temperature and killed by decapitation at 0 and 6 h into the heat challenge. The brain subregions AH, IMM, and FB were dissected and immersed in RNALater.

Total RNA isolation and real-time PCR
Total RNA was isolated with Total RNA Purification Plus kit (Norgen Biotek) according to the manufacturer's instructions. Isolated RNA (0.5 mg) was reverse transcribed to single-stranded cDNA by SuperScript II Reverse Transcriptase and oligo(dT) plus random primers (Invitrogen). Real-time PCR was performed with 10-ng cDNA in a StepOnePlus Real Time PCR System (Applied Biosystems) with PerfeCta SYBR Green FastMix, ROX (Quanta BioSciences). Dissociation curves were analyzed following each real-time PCR to confirm the presence of only one product and the absence of primer dimer formation. The threshold cycle number (Ct) for each tested gene (X) was used to quantify the relative abundance of that gene using the formula 2 (Ct gene X -Ct standard). Hydroxymethylbilane synthase (HMBS) was used as the standard for mRNA expression. The primers used for real-time PCR were as follows (59!39): HMBS, F-CGTTTGGAGGGTGGCTGTAG, R-TGTCAAGT ACAACTGGCCATCTTT; FTO, F-TAACATGCCTCTGCCA CTTG, R-GGCTGGAAGGTGACCTGATA; enhancer of zeste homolog 2 (EZH2), F-CACTGAACAGCAGCTTCCAGG, R-AAGAATGCAGGCTTTGCTCC; BDNF, F-GCTTGGCTTA CCCAGGTCTTC, R-TCAAAAGTGTCCGCCAGTG; heatshock protein70 (HSP70; HSPA2), F-TGGGTGTCTTCCAG CATGG, R-GATGAGGCGCTCTGTATCGG.

Quantification of global m 6 A RNA methylation
Total RNA was isolated with Total RNA Purification Plus kit (Norgen Biotek). The purified RNA (200 ng of each sample) was processed for detecting m 6 A RNA methylation status using the EpiQuik m 6 A RNA Methylation Quantification Colorimetric kit (EpiGentek) according to the manufacturer's instructions. The detected signal was quantified colorimetrically by reading the absorbance at 450 nm (Infinite M200 PRO microplate plate reader, Tecan). The absolute percentage of m 6 A in total RNA was Figure 1. Effect of heat stress on global m 6 A RNA methylation in the AH. A, Schematic diagram of the experimental setup. The chicks were heat-conditioned on day 3 at moderate (36°C) or extreme (40°C) ambient temperatures for 24 h. One week after (day 10), the chicks were challenged by exposure to 36°C for 6 h. Chicks were killed at 6 and 24 h into heat conditioning and at 0 and 6 h into heat challenge. Dissected brain subregions were analyzed. B, Illustrative diagram demonstrating the dissection of the brain area matching the AH, mesopallium IMM and FB. C, Absolute amount (%) of m 6 A in total RNA during heat conditioning on day 3. Total RNA was isolated from the AH of 3-d-old naive chicks at 6 h into the treatment and subjected to quantification of global m 6 A RNA methylation. Results are mean 6 SEM of 15-17 chicks in each group (F (2,44) = 3.39; p = 0.043; one-way ANOVA; Tukey's HSD test). D, E, FTO mRNA expression during heat conditioning on day 3. Total RNA was isolated from the AH (D) and FB and IMM (E) at indicated time points. FTO mRNA expression was evaluated by Syber green real-time PCR. HMBS was used as a standard gene. The relative mRNA expression in naive chicks was set to 1. Each value is the mean 6 SEM of 10-23 individual chicks (F (4,103) = 2.870; p = 0.026; one-way ANOVA; Dunnett's multiple comparisons test relative to naive). F, FTO mRNA expression in AH throughout the first week postconditioning. Total RNA was isolated from the AH 1, 2, 3, and 7 d following heat conditioning (recovery) and subjected to Syber Green real-time PCR. Each value is the mean 6 SEM of 10-30 individual chicks normalized to that of naive agematched chicks (recovery effect F (4,134) = 7.156, p , 0.0001; conditioning effect F (4,134) = 0.176, p = 0.676; interaction F (4,134) = 0.371, p = 0.829; two-way ANOVA; Dunnett's multiple comparisons test relative to 24 h). G, FTO mRNA expression in AH during heat challenge on day 10. Non-conditioned chicks (non-cond) and chicks that were exposed to moderate (36°C; 36°C-cond) or calculated using a standard curve prepared from the samples with known amount of m 6 A.
The m 6 A-RIP protocol was adopted from the method described by Dominissini et al. (2013) with several modifications. Purified RNA (20 mg) of two chicks within the same treatment was pooled and shared to ;100 nucleotide fragments with Magnesium RNA fragmentation Module (New England Biolabs) at 94°C for 5 min.
Fragmented RNA was incubated with 20 ml of Magna ChIP Protein A 1 G magnetic beads (Millipore) precoated with 2 mg of anti-m 6 A antibody (Abcam) or mouse IgG (Millipore; used as background RIP) in RIP buffer (150 mM NaCl, 10 mM TRIS-HCl, and 0.1% NP-40) supplemented with RNasin Plus (Promega) for 2 h at 4°C. The beads were washed three times with RIP buffer, and m 6 A RNA was eluted twice with RIP buffer including 20 mM of m 6 A 5'-monophosphate sodium salt (Cayman Chemical). The samples were cleaned up by ethanol precipitation, and 70 ng of each sample was reverse transcribed to singlestranded cDNA. The m 6 A enrichment was determined by real-time PCR using the primers mentioned in the upper paragraph.

FTO-antisense "knock-down"
An antisense oligodeoxyribonucleotide was designed to hybridize to the AUG translation initiation codon of the mRNA encoding FTO. A blast search revealed no significant homology to any sequence in GenBank other than the FTO sequence. The antisense and sense sequences were 59-CpCpTGCTCTCCTCTTCATGCpT-39 and 59-ApGpCATG AAGAGGAGAGCAGpG-39, respectively. The oligodeoxyribonucleotides were protected by double-phosphorothiolation at both the 39-end and 59-end (p represents protected nucleotides) and were purified by high-performance liquid chromatography (Sigma-Aldrich). Both oligonucleotides were dissolved in saline to a final concentration of 1 mg/ml. FTO-antisense or FTO-sense (2.5 mg) was injected into the third ventricle of 3-d-old chicks, as previously described (Kisliouk and Meiri, 2009). The effectiveness of FTO-antisense inhibition was evaluated by measuring of chick body temperature, FTO mRNA levels, and quantification of global m 6 A RNA methylation in the AH at 2, 6, and 24 h after injection.
Both groups of injected chicks were kept at a constant temperature of 30°C until day 10, when they subjected to heat challenge by exposure to 36°C for 6 h. Their AH was dissected and immediately immersed in RNALater for RNA and protein isolation. The changes in chick body temperature were evaluated before and at the end of heat challenge.

Chromatin immunoprecipitation (ChIP) assays
Anterior hypothalamic tissues were crosslinked with 1% formaldehyde for 10 min and then sonicated in cell lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris; pH 8.1) supplemented with protease inhibitor cocktail from Cell Signaling Technology for nine rounds of 10 pulses each using a Vibracell Sonix (maximal power 750W; Sonics & Materials Inc.) at 30% maximal power to obtain fragments of 200-1000 bp. Sheared chromatin samples were diluted in ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl, and protease inhibitor cocktail) and incubated with anti-H3K27me2 antibody or mouse IgG as background IP (each 3 mg/sample; Millipore) overnight at 4°C. Immunoprecipitates were separated by Magna ChIP Protein A 1 G magnetic beads (20 ml/sample; Millipore) for 2 h at 4°C and reverse crosslinked in ChIP elution buffer [1% SDS, 100 mM NaHCO 3 , 0.2 M NaCl, and proteinase K (50 mg/sample)] for 2 h at 62°C. DNA was isolated from each immunoprecipitate with Simple ChIP DNA Purification Buffers and Spin Columns (Cell Signaling continued extreme (40°C; 40°C-cond) ambient temperature on day 3 posthatch were re-exposed to moderately high ambient temperature of 36 6 0.5°C a week later. Total RNA was isolated from the AH before, and at 6 h into the challenge. FTO mRNA expression was evaluated by Syber Green real-time PCR. Each value is the mean 6 SEM of 20-30 individual chicks normalized to the respective control at 0 h, which was set to 1. H, Absolute amount (%) of m 6 A in total RNA during heat challenge on day 10. Total RNA was isolated from the AH of 10-d-old naive (non-cond), 36°C-conditioned (36°C-cond) and 40°C-conditioned (40°C-cond) chicks, before, and at 6 h into the challenge. Each bar represents average m 6 A RNA methylation percentage of 10-23 individual chicks (conditioning effect F (2,96) = 2.689, p = 0.073; challenge effect F (1,96) = 2.869, p = 0.093; interaction F (2,96) = 2.366, p = 0.099; two-way ANOVA; Fisher's LSD multiple comparisons test); p indicates significant difference from naive; # indicates significant difference from 24-h conditioning along the recovery period; 1 indicates significant difference between 0 and 6 h into heat challenge. Technology) and subjected to real-time PCR using BDNF primers aligning at the following positions: À869 to À801 bp upstream of the coding region, F-TGGTTTTCAT GAGGAGCCCT and R-TTTCCCAGAGCCCCATATCA; and 191 to 1190 bp into the coding region (for sequences see previous section) and 11623 to 11698 bp [located at the 39-untranslated region (39-UTR)], F-GTCCCCTCCCCTT TCCTCTC and R-CAAGCTCCAGTTGTATGCTGAGTG. The data were normalized to an input control that consisted of PCRs from 1% crosslinked chromatin before immunoprecipitation.

Statistical analysis
Data were analyzed using IBM SPSS statistics (version 20, Statistical Package for the Social Sciences) and GraphPad Prism 6 software (GraphPad Software). No animals were excluded from the experiments or statistical analysis. Statistical power analysis was not applied for sample sizes determination. Nevertheless, the sample size was determined on the basis of previous studies (Kisliouk et al., 2014;Cramer et al., 2015). Sample number (N) includes the number of individual chicks in each treatment group. Data are presented as mean 6 SEM. The Shapiro-Wilk W test was used to examine the distribution. When the distribution was normal, i.e., Shapiro-Wilk W test gave a significance p . 0.05, parametric tests such as two-way ANOVA (for multiple effects and their interaction), Student's t test (for two separate comparisons), or one-way ANOVA (for multiple comparisons) were used. Tukey's HSD test, Dunnett's, or Fischer's LSD tests were run for multiple comparisons, according to the data. Differences were considered significant at p , 0.05. The statistic values are indicated in figure legends.

Results
Heat stress reduces global m 6 A RNA methylation in the AH To test whether acute heat stress influences m 6 A RNA methylation, 3-d-old chicks were conditioned at either 36°C or 40°C (Fig. 1A). These temperatures were chosen because they have been shown to induce heat resilience or vulnerability, respectively, later in life (Cramer et al., 2015). As demonstrated in Figure 1C, global m 6 A RNA methylation was reduced in the AH of 3-d-old chicks in a temperature-dependent manner, with significant decrease (;25%; p = 0.043) at 6 h into the high-temperature (40°C) conditioning. The decrease in m 6 A RNA levels during heat conditioning was inversely correlated with FTO mRNA expression in the AH: after 6 h of heat conditioning at 40°C, the expression level of FTO mRNA was 1.5-fold higher than that in naive chicks (p = 0.034), whereas a 1.3-fold increasing trend in FTO mRNA expression was measured at the same time in 36°C-conditioned chicks (Fig. 1D). At the end of conditioning (24 h), FTO expression reached to its maximum level (;1.5-fold trend) and ;1.7 increase (p = 0.016) in 36°C-and 40°C-conditioned chicks, respectively, compared with naive chicks (Fig. 1D). To assess regional specificity concerning m 6 A RNA methylation in response to heat stress, FTO mRNA levels were also evaluated in two additional brain areas, FB and IMM (Fig. 1B), at 6 h into the 40°C conditioning. As a result of heat exposure, in contrast to the expression in the AH, there was no change in FTO mRNA expression in either of these two areas (Fig. 1E). FTO mRNA expression in the AH of both conditioned groups gradually declined after the end of the heat conditioning (recovery effect p , 0.0001; two-way ANOVA; Fig. 1F). During recovery on the first day postconditioning (day 4 posthatch), FTO mRNA levels in 40 and 36°C-conditioned chicks tended to be 1.4-and 1.2-fold higher than those in naive age-matched chicks (Fig. 1F). However, from the second to seventh day postconditioning (days 5-10 posthatch), FTO mRNA expression significantly decreased in 40°Cconditioned chicks (p = 0.028, p = 0.039, and p = 0.0002 on recovery days 2, 3, and 7, respectively). FTO mRNA levels in 36°C-conditioned during recovery period presented similar dynamics (p = 0.029 on recovery day 7). It should be noted that no interaction was found between the effect of heat conditioning and heat recovery measured from the second to seventh day postconditioning. Moreover, heat challenge on day 10 did not significantly affect FTO mRNA levels in both conditioned chick groups as well as in 10-d-old naive chicks (Fig. 1G). The general effects of heat conditioning and heat challenge on global m 6 A RNA levels in the AH evaluated by two-way ANOVA had a trend toward significance. However, by the LSD post hoc test, one week after heat conditioning, global m 6 A RNA levels in the AH of both groups of conditioned chick were lower, by approximately 30% and 40% for 36°C -and 40°C-conditioned chicks, respectively, than those of naive age-matched counterparts (p = 0.005 and p = 0.0006 for 36°C-and 40°C-conditioned chicks, respectively; 0 h; Fig. 1H). Moreover, m 6 A RNA methylation in the AH was not further altered in both conditioned chick groups at 6 h into heat challenge (Fig. 1H). Interestingly, exposure to heat stress of 10-d-old naive chicks, similarly to the heat conditioning on day 3 posthatch, resulted in reduction (;35%; p = 0.01) of global m 6 A RNA methylation, (Fig. 1H). The m 6 A RNA levels of 10-d-old non-conditioned chicks after 6 h of heat challenge were similar to those of conditioned age-matched chicks (Fig. 1H). These data indicate that acute heat stress, independent of age, leads to decrease in global m 6 A RNA methylation in the AH.
M 6 A RNA methylation profile of EZH2 and BDNF in the AH of harsh-temperature-conditioned chicks is opposite to that of mild-heat-conditioned ones Based on the fact that alterations in global m 6 A RNA methylation reflect the changes in m 6 A level along total RNA, including all the RNA in a cell, we aimed to identify m 6 A methylation along specific transcripts in correlation with heat stress. As potential candidates were chosen EZH2, BDNF, and HSP70, because their role in both thermal control establishment and heat stress response was previously demonstrated (Kisliouk and Meiri, 2009;Kisliouk et al., 2017). We analyzed m 6 A levels along EZH2, BDNF, and HSP70 transcripts at their coding region by m 6 A-RIP followed by PCR. A strong interaction between heat conditioning and challenge was determined by two-way ANOVA (p = 0.026 for EZH2, p = 0.004 for BDNF; Fig. 2A,B). A week after heat conditioning (0 h into heat challenge), m 6 A levels along both EZH2 and BDNF transcripts in the AH were 40% and 35% lower [p = 0.06 for EZH2 ( Fig. 2A) and p = 0.04 for BDNF (Fig. 2B)], respectively, in 40°C-conditioned chicks than those in their 36°C-conditioned counterparts. However, when heat challenged, the m 6 A levels in the AH were opposite between harsh-and mild-heat-conditioned chicks. While a 6-h heat challenge of the 40°C-conditioned chicks resulted in 2-and 1.8-fold increase in m 6 A levels at EZH2 and BDNF transcripts, respectively (p = 0.037 for EZH2 and p = 0.005 for BDNF; Fischer's LSD), a 6-h heat challenge of the 36°C-conditioned chicks led to decrease in methylation of both transcripts (1.7-and 1.6-fold decrease of EZH2 and BDNF, respectively; p = 0.076 for EZH2 and p = 0.049 for BDNF; Fischer's LSD; Fig. 2A,B). No changes in m 6 A RNA methylation were detected in the AH of the same samples at HSP70 during heat challenge (Fig. 2C). It should be noted that exposure to heat of 10d-old non-conditioned chicks did not affect m 6 A levels at EZH2, BDNF, and HSP70 transcripts in the AH ( Fig. 2A-C). Since the amount of mRNA in the cell might depend on the level of m 6 A RNA methylation, we also analyzed EZH2, BDNF, and HSP70 mRNA expression in the AH during heat challenge on day 10 posthatch. An interaction between heat conditioning and challenge on EZH2, BDNF, and HSP70 mRNA expression was evaluated by two-way ANOVA (p = 0.056 for EZH2, p = 0.084 for BDNF, and p = 0.032 for HSP70; Fig. 2D-F). As depicted in Figure  2D,E, EZH2 and BDNF mRNA baseline expression (0 h into heat challenge) in the AH was not affected by either harsh or moderate heat conditioning and was not different from that of non-conditioned age-matched chicks. However, the levels of both transcripts at 6 h into the heat re-exposure were reduced by 35 and 30%, respectively, comparing with those before the heat challenge [p = 0.0002 for EZH2 mRNA (Fig. 2D) and p = 0.023 for BDNF mRNA, Fischer's LSD (Fig. 2E)]. Heat challenge (6 h) of 36°C-conditioned chicks as well as non-conditioned counterparts had no significant effect on the mRNA expression of both EZH2 and BDNF in the AH (Fig. 2D,E). As expected (Kisliouk et al., 2017), the HSP70 mRNA expression during heat challenge was the highest in chicks conditioned under harsh ambient temperature: at 6 h into the challenge, HSP70 mRNA level in 40°C-conditioned chicks was .1.3 times higher than that in non-conditioned chicks (p = 0.002) and almost two times higher than that in the 36°C-conditioned counterparts (p , 0.0001; Fischer's LSD; Fig. 2F). These data imply on a role for m 6 A RNA signature on transcript-specific regulation of gene expression, specifically, an increase in m 6 A RNA levels negatively correlated with the mRNA expression. To confirm the AH regional specificity of m 6 A RNA methylation profile, m 6 A levels at EZH2, BDNF, and HSP70 transcripts were also evaluated in FB and IMM during heat challenge of 10-d-old 40°C-conditioned chicks and their non-conditioned counterparts. There were no changes detected in m 6 A levels at EZH2, BDNF, and HSP70 transcripts in either of these two areas before (0 h) and at 6 h into heat challenge (Fig. 2G-L).

FTO-antisense knock-down alters EZH2 and BDNF mRNA levels in the AH
To explore whether an increase in the m 6 A RNA methylation down-regulates EZH2 and BDNF mRNA levels, and furthermore, affects heat stress response, FTOantisense DNA was intracranially injected into the third ventricle of 3-d-old chicks (Fig. 3A). The time course of inhibition was determined by measuring the FTO mRNA levels in the AH between 2 and 24 h after the injection and comparing them with those in chicks injected with FTO-sense-specific sequence (Fig. 3B). As depicted in Figure 3B, a 35% inhibition of FTO mRNA (p = 0.016) was observed at 6 h after injection, at 24 h after the antisense treatment, FTO mRNA levels were similar to those in sense-treated chicks. Since inhibition of FTO mRNA expression is expected to lead to increase in m 6 A modification, we also evaluated global m 6 A RNA methylation level in AH at 6 and 24 h following FTOantisense injection. Global m 6 A RNA methylation was augmented by 40% at 24 h after the FTO-antisense injection (p = 0.023; Fig. 3C). At the next step, expression levels of EZH2, BDNF, and HSP70 were quantified. FTO-antisense knock-down resulted in reduction of both EZH2 and BDNF by 25 and 15%, respectively, at 24 h after the treatment (p = 0.005 for EZH2 and p = 0.053 for BDNF), but one week after mRNA, levels of both transcripts corresponded to those in sensetreated chicks (Fig. 3D,E). It should be noted that no changes were detected in HSP70 mRNA levels following FTO-antisense injection (Fig. 3F).

Effect of FTO-antisense knock-down on histone 3 lysine 27 (H3K27) methylation in the AH
To further evaluate the downstream effects of the FTOantisense inhibition mediated by EZH2 function, EZH2 protein expression was analyzed in the AH of FTO-antisense-injected chicks and compared with its expression level in FTO-sense-injected ones. The effect of intracranial injection of the FTO-antisense on EZH2 protein levels was similar to that on EZH2 mRNA expression, with ;25% inhibition of EZH2 observed at 24 h after FTO-antisense treatment (p = 0.059), but no changes in EZH2 levels one week after (Fig. 4A). Since, EZH2 is an H3K27-specific methyltransferase, (Laugesen et al., 2019), we analyzed global methylation of H3K27 after FTO-antisense injection. FTO-antisense knock-down resulted in long-term decrease in H3K27 dimethylation (H3K27me2; Fig. 4B), but did not affect H3K27 trimethyl levels (H3K27me3; Fig.  4C). A significant decrease in H3K27me2 was detected 24 h after FTO-antisense injection (;50%; p = 0.037), and one week after the treatment, H3K27me2 levels in antisense-injected chicks remained lower by ;30% than in sense-treated counterparts (p = 0.089; Fig. 4B). Given that H3K27me2 is involved in transcriptional regulation of BDNF expression (Kisliouk and Meiri, 2009), a ChIP assay was performed in FTO-knock-down chicks 24 h after the antisense treatment. H3K27me2 level at the 59-proximal regions, À869 to À801 bp upstream and 191 to 1190 bp into the BDNF coding region, was ; 50% lower in FTOantisense-injected chicks comparing with sense-injected Figure 2. Long-term effect of heat stress on m 6 A methylation of EZH2, BDNF, and HSP70 transcripts in the AH. A-C, m 6 A enrichment along the EZH2 (A), BDNF (B), and HSP70 (C) transcripts in the AH during heat challenge on day 10. Total RNA was isolated from the AH of 10-d-old chicks before (0 h) and 6 h into heat challenge and then subjected to m 6 A-RIP with anti-m 6 A antibody followed by real-time PCR. The levels of EZH2, BDNF, and HSP70 mRNA in the RIP samples were normalized against the HMBS ones (EZH2 m 6 A-IP, BDNF m 6 A-IP, and HSP70 m 6 A-IP, specifically). Each value is the mean 6 SEM of 9-13 pools of the two chicks normalized to that of non-conditioned chicks (non-cond) at 0 h, which was set to 1. For EZH2 m 6 A-IP conditioning effect F (2,54) = 0.111, p = 0.89; challenge effect F (1,54) = 0.045, p = 0.83; interaction F (2,54) = 3.885, p = 0.026; two-way ANOVA. For BDNF m 6 A-IP conditioning effect F (2,54) = 0.297, p = 0.74; challenge effect F (1,54) = 0.102, p = 0.75; interaction F (2,54) = 6.196, p = 0.004; two-way ANOVA. Fisher's LSD multiple comparisons tests: # significant difference between 36°C-and 40°C-conditioned chicks (36°C-cond and 40°C -cond, respectively); p significant difference relative to respective control (0 h). D-F, mRNA levels of EZH2 (D), BDNF (E), and HSP70 (F) in the AH during heat challenge on day 10. Total RNA from the AH of the same samples described in A-C was subjected to real-time PCR with EZH2 (D), BDNF (E), and HSP70 (F) primers, respectively. HMBS mRNA expression was used as the standard. counterparts (p = 0.019 for upstream part and p = 0.040 for coding region; Fig. 4D). No significant decrease in H3K27me2 levels was measured at the BDNF 39-UTR (11623 to 11698 bp downstream of the translation start site), using as a control (Fig. 4D). These data can imply a long-term effect of FTO-antisense inhibition resulted in reduction of EZH2 level and decrease in H3K27 dimethylation.
continued Results are mean 6 SEM of 20-25 chicks in each group. The PCR relative value of non-conditioned chicks at 0 h was set to1. For EZH2 mRNA conditioning effect F (2,129) = 2.97, p = 0.055; challenge effect F (1,129) = 10.47, p = 0.001; interaction F (2,129) = 2.98, p = 0.054; two-way ANOVA. For BDNF mRNA conditioning effect F (2,128) = 2.49, p = 0.086; challenge effect F (1,128) = 0.861, p = 0.35; interaction F (2,128) = 2.515, p = 0.085; two-way ANOVA. For HSP70 mRNA conditioning effect F (2,129) = 14.06, p , 0.0001; challenge effect F (1,129) = 158, p , 0.001; interaction F (2,129) = 3.54, p = 0.032; Fisher's LSD multiple comparisons tests: p significant difference relative to respective control (0 h); # significant difference in HSP70 mRNA level between 36°C-and 40°C-cond at 6 h into the challenge. G-I, m 6 A enrichment along the EZH2 (G), BDNF (H), and HSP70 (I) transcripts in the FB during heat challenge on day 10. J-L, m 6 A enrichment along the EZH2 (J), BDNF (K), and HSP70 (L) transcripts in the IMM during heat challenge on day 10. Total RNA isolated from the FB (G-I) and IMM (J-L) was subjected to m 6 A-RIP with anti-m6A antibody followed by real-time PCR as described in A-C. Each value is the mean 6 SEM of six pools of the two chicks normalized to that of non-conditioned chicks (non-cond) at 0 h, which was set to 1. Figure 3. FTO-antisense knock-down affects EZH2 and BDNF mRNA expression in the AH. A, Schematic representation of the treatment protocol. FTO-antisense was injected into the third ventricle on day 3 posthatch. Chicks injected with FTO-sense were used as controls. One week after (day 10), the chicks heat challenged by exposure to 36°C for 6 h. Chicks were killed 2 h, 6 h, 24 h, and 7 d following FTO-sense/antisense injection, and their AH was dissected for following analyses. B, Pharmacokinetic evaluation of FTO-antisense knock-down in the AH by measuring FTO mRNA expression. Total RNA was isolated from the AH and subjected to real-time PCR with FTO-specific primers. HMBS was used as a standard gene. The relative PCR values of antisense-treated chicks were normalized to respective sense ones, set as 1. Each bar represents mean 6 SEM (n = 8 at 2 h; n = 20 at 6 h; and n = 22 at 24 h); p indicates significant difference between sense and antisense counterparts 6 h following antisense treatment (t (41) = 2.50, p = 0.016; Student's t test). C, Quantification of global m 6 A RNA methylation in the AH following FTO-antisense treatment. Results are absolute amount (%) of m 6 A in total RNA at each indicated time. Each bar represents average m 6 A RNA methylation percentage (n = 10 at 6 h and n = 21 at 24 h); p indicates significant difference between sense and antisense counterparts 24 h following antisense treatment (t (40) = 2.37, p = 0.023 Student's t test). D-F, Effect of FTO-antisense knock-down on EZH2 (D), BDNF (E), and HSP70 (F) mRNA expression. Total RNA was isolated from the AH and subjected to real-time PCR with EZH2-, BDNF-, and HSP70-specific primers. HMBS was used as a standard gene. The relative PCR values of antisense injected chicks were normalized to those of respective sense, set as 1. Each bar represents mean 6 SEM (n = 22 at 6 and 24 h; and n = 33 at 7 d); p p = 0.053 (t (42) = 1.99, Student's t test) and p p = 0.005 (t (42) = 2.95, Student's t test) compared with respective control.

FTO-antisense knock-down affects the thermal control establishment and leads to heat vulnerability later in life
After evaluating the biochemical effect of the FTO-antisense inhibition on EZH2 and BDNF expression, we examined the phenotypic effect of FTO-antisense injection by measuring the injected chicks' body temperature. Intracranial injection of FTO-antisense resulted in elevation of the chicks' basal body temperatures. Six hours after the treatment, the body temperature of FTO-antisense-injected chicks was 0.3°C higher than that in FTOsense-treated counterparts (41.0 6 0.05°C in antisensetreated chicks vs 40.7 6 0.07°C in sense-treated chicks; p = 0.001), and at 24 h, it remained at the same level (41.0 6 0.07°C in antisense-treated chicks vs 40.7 6 0.05°C in sense-treated chicks; p = 0.003; Fig. 5A). Although no significant differences were observed in the body temperatures of the examined chick groups one week following FTO-antisense injection (40.9 6 0.05°C in antisensetreated chicks vs 41.0 6 0.04°C in sense-treated chicks), at 6 h of heat challenge on day 10 posthatch, the body temperature of FTO-antisense-injected chicks was 0.5°C higher than that in FTO-sense-treated counterparts (42.8 6 0.11°C in antisense-treated chicks vs 42.3 6 0.11°C in sense-treated chicks; p = 0.004; Fig. 5B). Hence, FTO-antisense-injected chicks demonstrated vulnerable response to heat stress. To further prove the role of FTOantisense knock-down in heat vulnerability, the mRNA level of the heat stress marker, HSP70 (Kisliouk et al., 2017), was measured in the AH of FTO-antisense-treated chicks during the heat challenge. The highest level of the HSP70 mRNA was observed in FTO-antisense-injected chicks at 6 h into the challenge, at which time, it was almost 1.4 times higher than that in FTO-sense-treated Figure 4. FTO-antisense knock-down affects methylation of the EZH2 substrate, histone H3 at lysine 27 (H3K27), in the AH. A, Time course of EZH2 protein inhibition following FTO-antisense injection by Western blot analysis. Total protein was isolated from the AH samples at 24 h and 7 d following FTO-antisense or sense injection and subjected to immunoblotting with anti-EZH2 antibody. Upper panels, Representative blots. Lower panels, Densitometric quantification of EZH2 levels relative to ACTB expression. The ratio between EZH2 and ACTB levels in continued sense-injected chicks at each indicated time was set to 1. Each bar represents the mean 6 SEM of 9-11 individual chicks; p p = 0.059 compared with respective sense (t (18) = 2.01, Student's t test). B, C, Evaluation of global levels of dimethyl and trimethyl histone H3 lysine27 (H3K27me2 and H3K27me3, respectively) after FTO-antisense injection. The protein samples described in A were immunoblotted with anti-H3K27me2 (B) and anti-H3K27me3 (C) antibodies. Upper panels, Representative blots. Lower panels, Densitometric quantification of H3K27me2 (B) and H3K27me3 (C) levels relative to ACTB expression in the same samples. Each bar represents the mean 6 SEM of 9-10 individual chicks; p p = 0.089 (t (16) = 1.81, Student's t test) and pp p = 0.037 (t (17) = 2.27, Student's t test) compared with respective sense. D, Effect of FTO-antisense inhibition on dimethylation of H3K27 (H3K27me2) along BDNF coding region. AH samples were collected 24 h after FTO-antisense or sense treatment, immunoprecipitated with anti-H3K27me2 antibody (H3K27me2-IP), and subjected to real-time PCR with BDNF-specific primers aligning at position À869 to -801 bp upstream of the coding region, 191 to 1190 bp into the coding region and 11623 to 11698 bp downstream of the translation start site. Immunoprecipitation with normal mouse IgG was used as a background (IgG-IP). Results are mean 6 SEM of 9-10 chicks in each group; pp p = 0.019 (t (18) = 2.65, Student's t test) for the area upstream of the BDNF coding region and p = 0.040 for the coding region (t (18) = 2.27, Student's t test) compared with sense. counterparts (p = 0.023; Fig. 5C). Thus, FTO-antisense knock-down on day 3 posthatch revealed vulnerability to heat challenge on day 10 posthatch.

Discussion
Fine-tuning of the thermal-response set point during the critical postnatal sensory-developmental period can determine future reactions to heat stress, inducing either resilience or vulnerability. Epigenetic modifications are involved in early life environmental programming, establishing a specific chromatin state to specify gene expression patterns associated with cellular memory (Franklin et al., 2012;Braun et al., 2017;Lux, 2018). We have previously shown that specific epigenetic marking of CRH intron (Cramer et al., 2019) and HSP70 promoter (Kisliouk et al., 2017) in response to harsh or mild heat stress at the critical development period underlay their differential expression during heat challenge later in life. Here, we demonstrate a role for epitranscriptomic regulation affecting the thermal control establishment. Chemical modifications on RNA have been shown to influence mRNA metabolism and thus fine-tune gene expression on top of the epigenetic code (Meyer et al., 2012;Gilbert et al., 2016;Nainar et al., 2016;Shi et al., 2019). Given that m 6 A is the most abundant epitranscriptomic mark existing across different brain regions (Chang et al., 2017), and it was identified in different aspects of the brain functions, among them learning and memory (Widagdo et al., 2016;Walters et al., 2017;Li et al., 2018;Shi et al., 2018) and acute stress response (Engel et al., 2018), it was of interest to explore the role of m 6 A RNA methylation in thermal control establishment and long-term heat stress response.
Here, we demonstrate that acute heat stress diminished global m 6 A RNA levels in the AH. Moreover, heat exposure at the critical period of thermal-control establishment has a long-term effect on RNA methylation: one week after the treatment, m 6 A global levels in both harsh-and mild-temperature-conditioned chicks were significantly lower than those in non-conditioned counterparts. Although two other studies have shown an increase in m 6 A RNA methylation in sheep liver (Shi et al., 2018) and abdominal fat and liver of piglets (Heng et al., 2019) under exposure to high environmental temperature, the differential effect of heat stress on m 6 A modification may be reasoned by different experimental models (animals, organs, and tissues) as well as heat stress stringency. Regulation of m 6 A RNA levels in the brain has been shown to be highly specific in a contextand experience-dependent manner (Widagdo et al.,Figure 5. Effect of FTO-antisense knock-down on chick thermal response during heat challenge. A, Alterations in body temperature of 3-d-old chicks after FTO-antisense injection. Chick body temperature was measured 6 and 24 h after FTO-antisense or sense treatment. The graph represents the mean 6 SEM of 26 and 38 chicks at 6 and 24 h, respectively; p indicates significant difference in the body temperature between sense and antisense at the same time point (t (50) = À3.43, p = 0.001 Student's t test at 6 h and t (73) = À3.05, p = 0.003 Student's t test at 24 h). B, Changes in body temperature of 10-d-old chicks previously injected with FTO-antisense during heat challenge. Following FTO-antisense or sense injection on day 3, the chicks kept at 30°C until day 10. Body temperature was measured before and 6 h into heat challenge. Results are mean 6 SEM of 32-34 chicks at each time point; p indicates significant difference in the body temperature between treatments at the same time point (t (63) = 3.03, p = 0.004; Student's t test). C, HSP70 mRNA expression in the AH during heat challenge of FTO-antisensetreated chicks. The chicks were treated as described in B. Total continued RNA was isolated before and 6 h into the challenge and subjected to real-time PCR with HSP70-specific primers. HMBS was used as a standard gene. The relative PCR values of antisense-treated chicks were normalized to respective sense ones, set as 1. Each bar represents mean 6 SEM of 31-32 chicks at each time point; p indicates significant differences in HSP70 mRNA expression between treatments at the same time point (t (60) = 2.34, p = 0.023 Student's t test). 2016; Chang et al., 2017;Walters et al., 2017;Widagdo and Anggono, 2018). Time course of RNA methylation in response to acute restraint stress in mouse brain revealed a decrease in m 6 A levels in the medial prefrontal cortex simultaneously with their increase in the amygdala (Engel et al., 2018). In addition, global RNA methylation was transiently decreased in whole blood of mice after acute stress (Engel et al., 2018). Apparently, stress-regulation of m 6 A modification is time and region dependent.
Whereas global RNA methylation describes an average of m 6 A modifications along the transcriptome, and therefore gives little information about the regulation of the repertoire of proteins that are relevant to the physiological state of the cells, we measured m 6 A levels at specific transcripts, EZH2, BDNF, and HSP70, the essential role of which was previously explored in heat stress response (Yossifoff et al., 2008;Kisliouk and Meiri, 2009;Kisliouk et al., 2011Kisliouk et al., , 2017. Here, we showed that exposure to harsh heat stress during the critical period of thermal control establishment (day 3 posthatch) resulted in significant decrease in methylation of the EZH2 and BDNF transcripts in the AH one week after the treatment. Methylation level of the same transcripts in mild-heat-conditioned chicks was unaffected. However, heat challenge on day 10 posthatch resulted in reduction of m 6 A levels at both transcripts in mild-heat-conditioned chicks and its elevation in high-temperature-conditioned ones. It should be noted that these alterations in m 6 A levels at the EZH2 and BDNF transcripts were AH specific and were not detected in other brain regions, such as FB or IMM, indicating the distinct role of the AH in heat stress processing and reprogramming the response to further challenges. In addition, changes in m 6 A modification may provide a long-term effect on the regulation of both EZH2 and BDNF expression under different stress conditions, for example, harsh heat stress versus mild heat stress. Moreover, an increase of this modification at the EZH2 and BDNF transcripts was accompanied by decrease in their mRNA levels, arguing for elevated m 6 A levels correlating with mRNA decay (Widagdo et al., 2016;Engel et al., 2018). While a several in vitro studies previously demonstrated that HSP70 expression is modulated by m 6 A level (Zhou et al., 2015;Yu et al., 2018b), no changes were found in m 6 A levels at the HSP70 transcript in response to aforementioned heat treatments. Perhaps, m 6 A is not involved in long-term regulation of the HSP70 RNA expression in chick hypothalamus in response to heat stress. In contrast, epigenetic marking at the HSP70 promoter in the AH, as we have previously shown, differentiate between its heat resilience and vulnerability (Kisliouk et al., 2017). Furthermore, differential m 6 A regulation of EZH2 and BDNF, on the one hand, and HSP70, on the other hand, point to a targetspecific role of the m 6 A in the environmental programming of gene expression.
To more specifically explore the role of m 6 A RNA methylation in heat stress regulation, FTO-antisense DNA was intracranially injected into the third ventricle during the critical period for the establishment of thermal control. FTO-antisense knock-down in the AH resulted in transient decrease in both EZH2 and BDNF mRNA expression, further supporting the observation that an increase in m 6 A levels can lead to mRNA degradation. Our findings are consistent with recent studies demonstrating reduction of BDNF expression as well as several key components of BDNF signaling pathway in the hippocampus of Ftoknock-out mice, which was also attended by impaired adult neurogenesis and learning and memory (Li et al., 2017;Spychala and Rüther, 2019). On the other hand, Mettl3 knock-down, resulted in a decrease in m 6 A levels, has been shown to reduce both Ezh2 protein expression and consequent H3K27me3 levels and altered the proliferation and cell cycle progression of adult neural stem cells (Chen et al., 2019a). Probably, m 6 A levels below or above a certain threshold impair the regulation of mRNA expression and/or degradation.
Since the functional consequence of m 6 A methylation of the EZH2 transcript can be evaluated by H3K27 methylation, we analyzed the levels of H3K27me2 and H3K27me3 in the Ant Hyp following FTO-antisense knock-down. Interestingly, transient FTO-antisense inhibition had longterm inhibitory effect on H3K27me2 but did not affect H3K27me3 levels. Long-term decline of H3K27me2, comparing with EZH2 transient inhibition by FTO-antisense knock-down, can be explained by the presence of other regulatory mechanisms balancing between methylation and demethylation processes (Lee et al., 2007). Moreover, increase in m 6 A RNA methylation, following FTO-antisense inhibition, may also influence the expression level of histone demethylases. These results also support our previous findings, demonstrating a significant increase in H3K27me2 simultaneously with EZH2 expression but no changes in H3K27me3, in chick hypothalamus during heat conditioning (Kisliouk and Meiri, 2009). Furthermore, EZH2 transient inhibition by Mir-138 in chick hypothalamus caused more profound inhibitory effect on H3K27me2 than on H3K23me3 (Kisliouk et al., 2011). Apparently, H3K27me2 is highly susceptible histone modification involved in transcriptional regulation of stress-activated genes in hypothalamic neuronal circuits. Indeed, a decrease in H3K27me2 level at the BDNF gene coding area following FTO-antisense injection suggests a role for H3K27me2 mark in this context. Moreover, our previous study has emphasized a specific role of the H3K27me2 modification at the BDNF promoter and coding region in the establishment of thermal control set point (Kisliouk and Meiri, 2009).
Phenotypically, FTO-antisense knock-down on day 3 posthatch led to heat stress vulnerability later in life, manifested by higher increase in the body temperature and HSP70 mRNA levels at heat challenge on day 10 posthatch. Similarly, Fto-knock-out in mice was resulted in a hyperactivation of the hypothalamic-pituitary-adrenal axis (Spychala and Rüther, 2019). Vulnerability to heat stress in FTO-antisense knock-down chicks, on the one hand, and increase in fear memory (Widagdo et al., 2016;Walters et al., 2017;Engel et al., 2018) and anxiety-like behavior (Spychala and Rüther, 2019) in Ftoknock-down mice, on the other hands, highlight the role of m 6 A RNA methylation in different memory-related processes.
To summarize, FTO-antisense knock-down in chick hypothalamus partially "mimics" the effects of heat challenge of 40°C-conditioned chicks, exhibited in elevation of global m 6 A RNA methylation, reduction of EZH2 and BDNF mRNA levels, and decrease in global H3K27 dimethylation as well as dimethyl H3K27 level along the BDNF coding region, and, finally, lead to vulnerable response to heat stress. Here, we present a dual-level regulation of BDNF expression in response to heat stress, including m 6 A marks on BDNF transcript and H3K27me2 modifications on BDNF gene. We suppose that cross talk between epigenetic and epitranscriptomic regulation can balance the response of stress-related neuronal networks to the future challenges.