SEARCH

SEARCH BY CITATION

Keywords:

  • chick;
  • critical developmental period;
  • DNMT 3a;
  • sensory development;
  • thermotolerance

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Thermal control establishment develops during a critical period by alterations in cellular properties in the frontal hypothalamus. These alterations may be modulated by the epigenetic code that determines the repertoire of transcribed proteins. Here we demonstrate transient changes in the expression of brain-derived neurotrophic factor (Bdnf) during both thermal conditioning and re-exposure of conditioned chicks to heat stress, relative to their age-matched naive counterparts. These changes coincide with changes in CpG methylation pattern in the avian Bdnf promoter region. Reduction in methylation during heat conditioning was observed at a cAMP response element-binding (CREB) site which coincided with both elevation in phospho-CREB levels and its binding to the Bdnf promoter. At the same time, an increase in methylation was observed at two other CpG sites, accompanied by elevation of the DNA methyltransferase 3a (DNMT3a) expression. DNMT3a was also found to bind to the two elevated methyl CpG sites, but not to the CREB binding site. These data suggest that complex and dynamic changes in DNA methylation are involved in the regulation of Bdnf expression during thermotolerance acquisition.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Thermal sensation and temperature control develop, like other sensory mechanisms, during a critical period (Hensch, 2004; Knudsen, 2004; Boulant, 2006). Excessive thermal input on day 3 of life improves the acquisition of thermotolerance in chicks, resulting in a significant reduction in heat production during subsequent exposure to acute thermal challenge throughout their lives (Yahav & McMurtry, 2001; Labunskay & Meiri, 2006).

Neuroanatomically, body temperature is balanced by the preoptic anterior hypothalamus (PO/AH) and controlled by specific neurons (Griffin et al., 1996). Exposure to environmental stress during the critical periods of hypothalamic development causes a plastic change in the hypothalamic–neuronal networks and can modulate stress responses (Tzschentke & Basta, 2002). We recently described components of the biochemical pathway leading to thermal stress response set-point establishment (Labunskay & Meiri, 2006; Tirosh et al., 2007; Meiri, 2008). We further demonstrated that the pathway is activated by brain-derived neurotrophic factor (Bdnf), showing not only that its mRNA is induced during thermal control establishment, but also that antisense knock-down of Bdnfimpairs thermal control establishment (Katz & Meiri, 2006), indicating the importance of this factor in developmental plasticity.

Cellular properties in the PO/AH which determine the stress response set-point are realized by the subset of proteins that are translated. The repertoire of proteins that are expressed is modulated by an epigenetic code, based on both short- and long-term posttranslational modifications of histones and of DNA methylation in promoters (Kouzarides, 2007; Miranda & Jones, 2007).

Methylation of cytosine–phosphate–guanine (CpG) dinucleotides within gene promoters is thought to control the transcription mechanism (Bernstein et al., 2007; Berger, 2007; Miller & Sweatt, 2007). The promoter area contains CpG dinucleotides that can be dynamically methylated or unmethylated under different developmental situations (Dennis & Levitt, 2005). Attempts to resolve the epigenetic code were made by identifying the mammalian Bdnf promoter CpG code but the mammalian sequence consists of five exons that are separately regulated. As the avian Bdnf is made up of only one exon with a promoter that contains several stand-alone CpG dinucleotides (Timmusk et al., 1993) it is an optimal system in which to study the CpG code. Methylation of cytosine residues is catalyzed by a family of enzymes: DNA (cytosine-5) methyltransferases (DNMTs), among which are DNMT1 which plays a maintenance role and DNMT3a and -3b which are involved in de novo methylation (Goll & Bestor, 2005). Recent evidence has indicated that DNMTs are abundant in differentiated neurons and, in addition to the permanent CpG methylation pattern, there is a transient CpG methylation code (Miller & Sweatt, 2007). This transcription–regulation mechanism has been shown to play a role in neuronal plasticity and long-term memory (Levenson et al., 2006; Miller & Sweatt, 2007).

In the study presented here, the level of CpG methylation in the promoter of the hypothalamic Bdnf was evaluated during heat conditioning in the critical period of thermal control establishment and later in life, during heat challenge, and was correlated with DNMT expression and with induction of both the phosphorylation of the transcription factor cAMP response element-binding (CREB) and its binding to the Bdnf promoter.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Birds, diets and environment

Cobb male chicks were obtained on day of hatch from the Brown commercial hatchery (Hod-Hasharon, Israel) and raised in controlled-climate rooms at 30°C under continuous artificial illumination and ad libitum access to food and water. All experiments were carried out according to the guidelines of the European Community Council.

Heat exposure

Heat conditioning

On day 3 of life, the test chicks were transferred to 37.5°C for 24 h while control birds were left at 30°C. At 2, 6, 12 and 24 h, a group of chicks was weighed, a different group each time, and their colonic temperatures were measured (∼3.5 cm into the colon).

Thermal challenge

On day 10 of life, both heat-conditioned and naive chicks where heat-challenged by exposure to 37.5°C for 24 h. At 2, 6, 12 and 24 h into the heat exposure, groups of chicks were weighed, a different group at each time point, and their colonic temperatures measured.

Tissue collection, and mRNA, DNA and protein purification

Chicks were killed by cervical dislocation after 2, 6, 12 or 24 h of thermal conditioning. For RNA, DNA and protein purification, the anterior hypothalamus was dissected and immediately immersed in RNALater (Ambion, Austin, TX, USA). RNA and DNA were purified using Tri reagent (Molecular Research Center, Cincinnati, OH, USA). For immunostaining, whole brains were dissected and fixed in 4% paraformaldehyde in PBS for 24 h.

Western blot analysis

Equal amounts of protein from PO/AH homogenate at each time point were prepared in sodium dodecyl sulfate (SDS) sample buffer, separated by SDS–polyacrylamide gel electrophoresis and subjected to Western blot analysis. After electrophoresis and electroblotting, the blots were blocked with 3% bovine serum albumin for 1 h at room temperature. The blots were then cut horizontally into three parts: the upper part, with peptides in the range 60–200 kDa, was reacted either with polyclonal antiserum to DNMT3a (ab16704; Abcam, Cambridge, UK) or with a polyclonal DNMT3b antibody (sc-52922; Santa Cruz Biotechnology, Santa Cruz, CA, USA); the middle part, with peptides in the range 30–60 kDa, was reacted either with a polyclonal antibody to β-actin (Cell Signaling Technology, Beverly, MA, USA), with a polyclonal antibody to CREB (ab5803; Abcam) or with a polyclonal antibody to phospho-CREB (Ser133) (Cell Signaling Technology); and the bottom part, peptide range 10–30 kDa, was reacted with a polyclonal antibody to Bdnf (1 : 150; Santa Cruz Biotechnology). Incubations were performed overnight at 4°C. After three 5-min washes, the blots were incubated for 1 h at room temperature with horseradish peroxidase-linked antirabbit IgG (Amersham Biosciences, Little Chalfont, UK). The blots were then exposed to enhanced chemiluminescence substrate (Pierce, Rockford, IL, USA). A chemiluminecent signal was detected using Image Master VDS-CL (Amersham Pharmacia Biotech). The densitometry analyses of protein expression were performed using imagej 1.3 image analysis software.

Reverse transcription and real-time polymerase chain reaction (PCR)

Hypothalamic RNA was reverse-transcribed to single-stranded cDNA using reverse primers and Moloney murine leukemia virus RT (Invitrogen, Carlsbad, CA, USA). Real-time PCR was performed on a model 7000 sequence analysis system (Applied Biosystems, Foster City, CA, USA). Quantification was established using the SYBR green method as previously described. Briefly, PCRs were performed in a total volume of 18 μL, consisting of (in μL) cDNA,1; iQ Absolute Blue SYBR green ROX Mix (ABgene, Epsom, UK), 9; primers (10 nm each), 1; and H2O, 7. Expression was determined by comparing the Bdnf levels with those of 18S. The primers were designed such that there would be no primer dimerization, and the amplification curves of the genes would be parallel. The following primers were used for real-time PCR: Bdnf, L-GCTTGGCTTACCCAGGTCTTC, R-TTCAAAAGTGTCCGCCAGTG; and 18S, L-CGGGTTGGTTTTGGTCTGAT, R-ATGGTTCCTTTGGTCGCTCC.

Immunofluorecent staining

Brains were removed and fixed in 4% paraformaldehyde in 0.1 m phosphate-buffered saline (PBS; pH 7.3), dehydrated in a series of graded alcohol solutions, cleared in chloroform and embedded in paraffin. Sections (5 μm) were cut with a microtome, dewaxed in xylene and rehydrated using decreasing ethanol concentrations. Antigen retrieval was performed by treating sections for 5–10 min in boiling 10 mm sodium citrate buffer (pH 6.0). Tissue sections were then washed in PBS and incubated for 1 h with normal donkey serum (10%), which served as a blocking agent for nonspecific binding. The Bdnf antibody (Santa Cruz Biotechnology Inc.), diluted 1 : 100 in blocking solution, was added for 18 h. After three washes in PBS, the slides were incubated for 2 h with Cy3-conjugated donkey antirabbit IgG (H + L; Jackson ImmunoResearch Laboratories, West Grove, PA, USA), diluted 1 : 400, counterstained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma Aldrich) and then photographed under a fluorescence microscope at 100× magnification.

DNA methylation assay

Purified DNA was processed for bisulfite modification (CpGenome DNA modification kit; Chemicon,). Semiquantitative PCR was used to determine the DNA methylation status of the Bdnf promoter. Methylation-specific PCR primers were as shown in Table 1.

Table 1.   Methylation-specific PCR primers
CpG positionL/RSequence
  1. L, left; R, right; M + U, methylated and unmethylated.

M1 (−186 bp)
 MethylatedLGGAAATTTAAATCGAAGAAATA
 UnmethylatedLGGAAATTTAAATTGAAGAAATA
 M + URCAAAATTATCAAAATTCACCAAAA
M2 (−251 bp)
 MethylatedLGTAAGATATTGGTATATACGAATTG
 UnmethylatedLGTAAGATATTGGTATATATGAATTG
 M + URACAAAATTATCAAAATTCACCA
M3 (−370 bp)
 MethylatedLGTTGAAACGTTGTGTTGTTAAATAG
 UnmethylatedLGTTGAAATGTTGTGTTGTTAAATAG
 M + URAAATCAAATACTACATAAACTCCTT
M4 (−474 bp)
 MethylatedLGATTAAATTATTTTTGGCGTAGAGAG
 UnmethylatedLGATTAAATTATTTTTGGTGTAGAGAG
 M + URCAAAAACAAAACTACTATTTAACAAC
M9 (−1082 bp)
 MethylatedLGTGTTGGTAGGAATGACGTTTTG
 UnmethylatedLGTGTTGGTAGGAATGATGTTTTG
 M + URAACACCAACTAACAACATCAATAAA
M12 (−1363 bp)
 MethylatedLGTGAATAATGTCGTTGTTTTTTAG
 UnmethylatedLGTGAATAATGTTGTTGTTTTTTAG
 M + URCCCCCAAATCCTACTATAAC

Samples were normalized to an unmethylated region from the reading frame of the gene using the following primers: Reading frame (+402): L, TGGGGAATTGAGTGTTTGTG; R, TACCCCTACAACCTTCCTTT.

Electromobility shift assay

Protein extracts were prepared from eight hypothalamic tissue samples from 3-day-old chicks, homogenized in ice-cold buffer containing NaCl, 0.1 m; Tris–HCl (pH 7.4), 20 mm; ethylenediaminetetraacetic acid (EDTA), 0.2 mm; glycerol, 20% (v/v); DTT, 0.5 mm; leupeptin, 15 g/mL; and phenylmethylsulphonyl fluoride, 1 mm. After homogenization, samples were centrifuged for 30 min (4°C, 12,000 g) and supernatants were collected, frozen in liquid nitrogen and stored at 70°C. The digoxigenin (DIG) gel shift kit for 3′-end labeling of oligonucleotides (Roche Applied Science, Indianapolis, IN, USA) was used for protein–DNA binding assays. Oligomers (M1, 5′-TATGATTTATCTGGAAATCTAAATCGAAGAAAC-3′; M3, 5′- CTTTTAATGCTGAAACGCTGTGTTGCTAAATA-3′; M9, 5′- TTTGGAGTGCTGGTAGGAATGACGTTTTGCTAAT-3′) were labeled and used in gel-shift reactions according to the manufacturer’s instructions (Roche). Briefly, 6 μg of cellular protein were incubated with DIG-labeled oligomers, electrophoretically separated by 8% native polyacrylamide gel in 0.5× Tris–borate–EDTA buffer and transferred onto a nylon membrane. Chemiluminescence of DIG-labeled DNA–protein complexes on the nylon membranes was achieved using DIG Wash and Block Buffer Set (Roche) and detected using Image Master VDS-CL (Amersham Pharmacia Biotech).

For the supershift reactions, 4 μL DNMT3a antiserum (ab16704; Abcam) or 2 μg DNMT3b antibody (sc-52922; Santa Cruz Biotechnology) were incubated with 6 μg of cellular protein for 60 min on ice prior to the gel shift reaction.

Chromatin immunoprecipitation (ChIP) assays

ChIP assays were performed using EZ ChIP Chromatin Immunoprecipitation Kit (Upstate Cell Signaling Solutions, Temecula, CA, USA) according to the manufacturer’s instructions, with several modifications. Briefly, anterior hypothalamic tissues were crosslinked with 1% formaldehyde for 10 min followed by addition of 700 μL per sample of SDS lysis buffer (SDS, 1%; EDTA, 10 mm; and Tris, pH 8.1, 50 mm) and sonication for seven rounds of 10 pulses each with a Vibracell Sonix (maximal power 750 watts; Sonics & Materials Inc, Newtown, CT, USA) at 30% maximal power to obtain 200- to 1000-bp fragments. Afterwards, 100 μL of sheared chromatin sample was used for immunoprecipitation with antibodies directed against p-CREB (4 μL/sample each; Cell Signaling Technology, Beverly, MA, USA). For mock immunoprecipitation (background), no antibody was used. DNA was isolated from p-CREB immunoprecipitates and subjected to real-time PCR using Bdnf primers aligning at the following positions: −793 up to −869 bp upstream of the coding region, F-TGGTTTTCATGAGGAGCCCT; R-TTTCCCAGAGCCCCATATCA; and +1623 to +1698 bp (located at the 3′-untranslated region), F-GTCCCCTCCCCTTTCCTCTC; R-CAAGCTCCAGTTGTATGCTGAGTG. The data were normalized to an input control which consisted of PCRs from 1% cross-linked chromatin before immunoprecipitation.

Statistical analysis

anova was used to examine the differences between the groups in all the biochemical studies. Results are presented as mean ± SEM.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Bdnf expression was induced during both heat conditioning and heat challenge of previously conditioned chicks

Plastic events in the hypothalamus during the critical period of temperature control establishment are responsible for long-term changes in the thermal control set-point. That Bdnf is crucial for synaptic plasticity and maintenance of long-term memory (Kang & Schuman, 1995; Patterson et al., 1996; Barco et al., 2005), that it plays an important role in critical periods of sensory development (Huang et al., 1999; Katz & Meiri, 2006), and that we have previously demonstrated its critical role in thermal control establishment (Katz & Meiri, 2006), led us to investigate changes in the expression of this protein during the time window known to be most effective for heat conditioning in chicks, i.e. on the third day of life. As revealed by Western blot analysis, Bdnf expression was induced during heat conditioning (heat exposure of the chicks to 37.5°C): the amplification began 2 h into the treatment and peaked 6–12 h into the heat treatment, at a level that was ∼60% higher than in naïve chicks. At 24 h, the expression level had decreased back to that in naïve chicks (Fig. 1A and B; Bdnf expression relative to that in naïve chicks was 1.58 ± 0.19 after 6 h, F1,27 = 6.33, < 0.02; 1.63 ± 0.23 after 12 h, F1,26 = 5.65, < 0.03).

image

Figure 1.  Induction of Bdnf protein expression in chick PO/AH during heat conditioning on day 3. (A) Representative Western blot analysis of heat treatment. Each time point represents Bdnf protein expression from a single chick. (B) Densitometry of Western blot analysis evaluated by imagej 1.30 image-analysis software. Each time point represents the level of Bdnf protein relative to naïve age-matched chicks (to minimize loading or exposure differences, Bdnf was compared with β-actin in each gel lane; n = 15 at each time point; bars are + SEM). *P < 0.05 between conditioned and naïve chicks. (C) Representative immunofluorescent staining of the PO/AH in sagittal sections reacted with antibody to Bdnf. Immunohistochemistry was performed with polyclonal antibody to Bdnf (Santa Cruz Biotechnology Inc.) and with Cy3-conjugated donkey antirabbit IgG. Enlargements (100×) of the PO/AHs from 6-h-conditioned and naïve chicks are presented in the left panels. The same fields counterstained with DAPI, which stains nuclei, are presented in the right panel. Coordinates of the section according to Kuenzel & Masson (1988), lateral distance from midline, L = 0.5 mm.

Download figure to PowerPoint

To visualize the localization of Bdnf protein during heat exposure, sagittal sections prepared from naïve and 6-h heat-treated chicks were immunofluorescently stained. The cellular distributions on the two representative slides were similar, as demonstrated by DAPI staining of the nuclei (Fig. 1C). Bdnf signal was visualized in more cells in the PO/AH area after 6 h of heat treatment than in naïve untreated chicks (Fig. 1C). It should be noted that this method is not quantitative.

To correlate the timing of Bdnf expression with the critical period of thermal control establishment, its expression was also evaluated in the PO/AH of two groups of 10-day-old chicks, one group that had been heat-exposed on day 3 (37.5°C for 24 h; the heat re-exposed group) and a group of naïve 10-day-old untreated chicks. Using real-time PCR, we found that Bdnf mRNA expression was not altered by heat treatment in chicks that had passed the critical age for thermal control establishment (Fig. 2A). However, Bdnf mRNA expression was significantly induced during heat challenge in chicks that had previously undergone heat exposure (Fig. 2A). A significant increment relative to the respective controls was observed 6 h into the challenge, reaching a maximum level 12 h into the treatment (2.51 ± 0.52 after 6 h, F1,15 = 6.99, < 0.05; 4.27 ± 1.04 after 12 h, F1,13 = 11.21, < 0.01). At 24 h, the level of expression had declined to that observed at 6 h (3.04 ± 0.66, F1,15 = 8.06, < 0.02; Fig. 2A). The induction of Bdnf expression in the challenged (re-exposed) group relative to that in their naive age-matched counterparts was significant throughout the experiment (< 0.05).

image

Figure 2. Bdnf mRNA expression pattern during heat exposure of naive chicks and heat re-exposure (challenge) of previously conditioned chicks on day 10 of age. (A) Real-time PCR quantification of Bdnf mRNA expression in the PO/AH of two groups of 10-day-old chicks: a group that was conditioned on day 3 by exposure to 37.5°C for 24 h (heat re-exposed), and a group of naïve untreated chicks. Bdnf expression level was compared to that of 18S using the SYBR green method. Each time point is an average + SEM of six naive chicks or nine re-exposed chicks. (B) Densitometry of Western blot analysis of Bdnf expression in the PO/AH of the two groups of 10-day-old chicks: naive and heat re-exposed chicks, evaluated by imagej 1.30 image analysis software. Each time point represents the levels of Bdnf protein relative to naïve age-matched chicks (to minimize loading or exposure differences, Bdnf was compared with β-actin in each gel lane; n = 6 naive and 8–9 re-exposed chicks at each time point; bars are + SEM). In the figure, aP < 0.05 within the heat re-exposed group and *P < 0.05 between naive and heat re-exposed chicks.

Download figure to PowerPoint

As induction of a particular gene does not always reflect a change in the expression of the protein it encodes, the expression of Bdnf protein was evaluated during heat challenge of previously heat-conditioned chicks and of age-matched naive untreated chicks by Western blot analysis. Bdnf protein level was consistent with its mRNA expression: it was not altered in naive chicks during heat exposure on day 10 (Fig. 2B). However, its expression was induced upon heat challenge in previously heat-conditioned chicks. The Bdnf protein levels in the PO/AH from 10-day-old heat-challenged, previously conditioned, chicks was compared to that from age-matched naive heat-exposed chicks: as a result of the re-exposure, Bdnf levels were induced from 2 to 12 h into the heat challenge, with peak induction at 12 h of 2.23 ± 0.38 (F1,11 = 48.48 and F1,14 = 9.54, < 0.01 after 6 and 12 h respectively, and F1,11 = 5.72, < 0.05 after 2 h (Fig. 2B). Similar to the expression of the Bdnf mRNA, there was a decline in Bdnf protein expression in the heat-conditioned chicks after 24 h of heat re-exposure.

DNA methylation was dynamically altered at the Bdnf promoter during heat exposure

As Bdnf was induced during heat exposure of 3-day-old chicks, we investigated the regulation mechanism which might underlie this induction. The reading frame of the chick Bdnf gene is delineated on one exon. Upstream of the Bdnf coding region’s initiation site there are several CpG dinucleotides which can be methylated and therefore are potentially involved in the regulation of Bdnf expression (Fig. 3A). Methylation of DNA in the PO/AH was measured at six CpG positions detected in silico during heat exposure of 3-day-old chicks (Fig. 3B–G). A complete sequence of the Bdnf promoter region was performed in order to control for a complete bisulfite conversion processes. Analyzing the DNA sequence in which these methylation sites reside we found that M1, located 186 bp upstream of the coding region (−186), is a potential CEB/P site, which is known to be involved in energy metabolism; M2 (−251) resides within an AP1 site which is the binding site of Jun; M3 (−370) is not located within a known transcription-regulation site; M4 (−474) is located in an ENKCRE and c-Myc site; M9 (−1082) is a CREB site, and M12 (−1363) is a c-Myb site (Fig. 3A).

image

Figure 3.  Alteration of CpG methylation upstream of the Bdnf coding sequence during heat conditioning. Chicks were thermally conditioned on day 3 of life. At 2, 6, 12 or 24 h into conditioning, their PO/AH was dissected, the DNA extracted and modified using a CpGenome DNA modification kit, and the amount of methylated CpG was determined using PCR with specific primers. Methylation was determined by comparing the PCR product from the CpG site with that from a site in the coding region. (A) A schematic chart depicting the CpG locations upstream of the Bdnf coding region. The analyzed methylated sites are marked in black and their exact location is delineated. (B–G) Methylation levels at positions M1, 2, 3, 4, 9 and 12, respectively. Each time point represents the methylation levels in 8–11 chicks and is + SEM. (B, D and F) *P < 0.05 between conditioned and naïve chicks.

Download figure to PowerPoint

Following heat exposure of 3-day-old chicks, there was a significant increase in the level of methylation in heat-exposed chicks compared to naïve chicks at positions M1 and M3 (Fig. 3B and D). There was a significant decrease in methylation in heat-exposed chicks compared to naïve chicks at position M9 (Fig. 3F). Positions M2, M4 and M12 did not show any significant changes in methylation during heat exposure (Fig. 3C, E and G). Methylation induction at the first CpG location upstream of the ATG initiation site (M1) began after 2 h of heat exposure (2.07 ± 0.37-fold increase compared to the level in naïve age-matched chicks; F1,17 = 6.13, < 0.02; Fig. 3B) and remained high throughout the heat treatment (2.38 ± 0.35 after 6 h, F1,17 = 11.24, < 0.01; 2.61 ± 0.41 after 12 h, F1,17 = 11.97, < 0.01; and 2.52 ± 0.41 after 24 h of heat treatment, F1,16 = 11.56, < 0.01; Fig. 3B). The methylation level at the third CpG location (M3) was only induced 12 h after the beginning of heat treatment, at which time it was 1.65 ± 0.15-fold that in naïve chicks (F1,18 = 9.83, < 0.01; Fig. 3D). The level of methylation at M3 was further elevated after 24 h (2.29 ± 0.23-fold higher than in naïve chicks, F1,18 = 28.32, < 0.01; Fig. 3D). In contrast, on M9 (−1082 bp) there was a significant decrease in the level of methylation, which started 6 h into the heat exposure and lasted for an additional 6 h (the level of methylation relative to that in naive chicks was 0.52 ± 0.05 after 6 h and 0.62 ± 0.17 after 12 h; F1,12 = 12.79, < 0.01 at 6 h, F1,15 = 9.01, < 0.01 at 12 h; Fig. 3F). However, after 24 h the expression level had returned to that in naïve chicks (Fig. 3F). At M2, M4 and M12 there was no change in methylation level during heat treatment (Fig. 3C, E and G).

Long-term effect of heat conditioning on Bdnf promoter methylation pattern

Heat conditioning has a long-term phenotypic effect on heat tolerance (Yahav & McMurtry, 2001; Labunskay & Meiri, 2006), which can be best evaluated by heat challenge later in life. In correlation with this phenotypic change, the expression pattern of Bdnf differed between heat-conditioned and naive age-matched chicks during heat challenge on day 10 (Fig. 2). We therefore evaluated the methylation pattern in the Bdnf promoter in two groups of 10-day-old chicks, a group that had been heat-conditioned on day 3 by exposure to 37.5°C for 24 h (the heat re-exposed group) and a group of naïve untreated chicks. As there was a slight but nonsignificant change in the baseline of Bdnf methylation between previously heat-conditioned and naïve chicks on day 10, the methylation level of the naive 10-day-old chicks was arbitrarily set to 1.

The DNA methylation pattern clearly differed in previously heat-conditioned vs. naïve 10-day-old chicks (Fig. 4). While CpG methylation was dynamically induced at all CpG positions on the Bdnf promoter during heat exposure of naive 10-day-old chicks, there were only minor changes in the Bdnf promoter during heat re-exposure of previously heat-conditioned chicks. These changes included a significant reduction in the amount of methylation at position M9 (Fig. 4E) and a single increment at position M4 (Fig. 4D). The difference in CpG methylation between conditioned and naïve heat-exposed chicks during heat re-exposure was significant in five out of the six positions checked (at M1 after 12 and 24 h, F1,11 = 10.73, < 0.01 and F1,18 = 6.72, < 0.02, respectively; at M2 after 2 h, F1,13 = 7.50, < 0.02, after 6 h, F1,13 = 6.58, < 0.02, after 12 h, F1,11 = 11.09, < 0.01 and after 24 h, F1,13 = 5.22, < 0.04; at M3 after 6 h, F1,8 = 5.77, < 0.04 and after 12 and 24 h, F1,11 = 9.00 and F1,9 = 15.21, respectively, < 0.01; at M9, after 2 h, F1,21 = 6.37, < 0.02, after 6 h, F1,18 = 5.76, < 0.03, after 12 h F1,15 = 9.22, < 0.01 and after 24 h, F1,17 = 15.23, < 0.01, and at M12 after 12 and 24 h, F1,12 = 4.94 and F1,15 = 4.89, respectively, < 0.05). There was no difference in CpG methylation patterns between previously heat-conditioned and naive chicks at position M4.

image

Figure 4.  Comparison between CpG methylation patterns upstream of the Bdnf coding sequence of previously conditioned chicks and aged-matched naive chicks during heat exposure on day 10 of life. Methylation in the Bdnf promoter was evaluated in two groups of 10-day-old chicks, a group that had been conditioned on day 3 by exposure to 37.5°C for 24 h (heat re-exposed) and a group of naïve untreated chicks, by exposing them to 37°C for 24 h. At 2, 6, 12 or 24 h into heat exposure their PO/AH was dissected, the DNA extracted and modified using a CpGenome DNA modification kit, and the amount of methylated CpG was determined using PCR with specific primers. (A–F) The amounts of methylation at M1, 2, 3, 4, 9 and 12, respectively. Each time point represents the methylation in 8–11 chicks and is + SEM.

Download figure to PowerPoint

In the first position upstream of the ATG initiation site (M1) there was a significant elevation in the methylation of the CpG dinucleotide throughout heat exposure of naïve 10-day-old chicks (Fig. 4A). The induction started 2 h into the heat exposure, at which time the methylation level was two-fold that in naive chicks; at 6 h, it had risen to 2.5-fold; the induction peaked at 12 h, at ∼3.5-fold the value in naive chicks and after 24 h the induction was 2.7 times that in naïve chicks (< 0.01 at all time points for heat-exposed vs. naive chicks, F1,16 = 6.76, F1,16 = 11.22, F1,13 = 13.95 and F1,16 = 12.19 at 2, 6, 12 and 24 h, respectively). The methylation of CpG at M1 in previously conditioned chicks was not altered during heat challenge (Fig. 4A).

At M2, the methylation induction in naïve 10-day-old chicks started 2 h into the heat exposure, at which time the methylation level was 1.3-fold that in naïve untreated chicks; at 6 h it was 1.6-fold; the level peaked at 12 h and was 2.5 times higher than in naive chicks, and after 24 h it was 1.5 times higher (F1,16 = 4.64, < 0.05 at 6 h and F1,13 = 17.29, < 0.01 after 12 h; Fig. 4B).

Whereas in naïve 10-day-old chicks there was a clear induction in methylation at M3 which stayed elevated throughout the heat exposure, methylation level in the age-matched heat-challenged group did not change (Fig. 4C; the amount of methylation in naïve 10-day-old heat-exposed chicks vs. naive age-matched chicks was 2.43 ± 0.45, F1,10 = 8.76, < 0.01 after 6 h, 2.31 ± 0.36, F1,10 = 10.61, < 0.01 after 12 h, and 2.72 ± 0.33, F1,10 = 20.72, < 0.01 after 24 h; the amount of methylation in heat re-exposed chicks remained the same throughout the heat challenge).

As mentioned above, the only position at which both heat-exposed naïve 10-day-old chicks and challenged, previously exposed, chicks’ methylation levels changed with time was M4. Both methylation levels were elevated by 40% compared to their respective controls (Fig. 4D).

At M9, the amount of methylation in previously heat-conditioned chicks was significantly reduced throughout heat re-exposure on day 10 (Fig. 4E). In these chicks, the amount of methylation was reduced by 40%, after 2, 6, 12 and 24 h (F1,24 = 4.37, F1,21 = 5.83, F1,18 = 5.31 and F1,20 = 7.54 after 2, 6, 12 and 24 h respectively; < 0.05 in all groups). In addition, the methylation in this challenged group was significantly lower than that in its counterpart aged-matched naïve 10-day-old group.

At M12, whereas in heat exposure of naive 10-day-old chicks there was a clear induction in methylation which stayed elevated throughout the heat exposure, methylation level in the heat-challenged group (chicks that were heat-exposed on day 3) did not change (Fig. 4F). The amount of methylation in naïve heat-exposed chicks compared to age-matched nonexposed counterparts was 1.42 ± 0.24, 2.16 ± 0.52, 2.54 ± 0.64 and 2.18 ± 0.47 after 2, 6, 12 and 24 h, respectively (F1,14 = 7.17, < 0.05 after 12 h and F1,16 = 6.03, < 0.03 after 24 h).

CREB bound to the Bdnf promoter at the M9 site

Given that transcription activation usually coincides with reduction of CpG methylation (Berger, 2007; Miller & Sweatt, 2007) and that CpG methylation decreased at M9 of the Bdnf promoter during heat exposure, in the next step we analyzed the binding ability of phosphorylated CREB (p-CREB) to the Bdnf promoter, around position M9 which is a potential CREB binding site (−869 bp), by ChIP assay. Indeed, binding levels of p-CREB to the Bdnf promoter increased significantly during heat exposure, while the binding to a part of the Bdnf sequence which is probably unrelated to transcription regulation was unaffected (i.e. 3′-untranslated region; +1623 bp; Fig. 5A). p-CREB levels after 12 h of incubation increased by 2.5-fold at the promoter region (F1,16 = 4.61, < 0.05; Fig. 5A).

image

Figure 5.  p-CREB and DNMT expression and binding capacity to the Bdnf promoter region during heat conditioning. (A) To assess the p-CREB binding to the Bdnf promoter, ChIP assays were performed. Collected PO/AH samples were immunoprecipitated with antibodies against p-CREB and subjected to real-time PCR with Bdnf-specific primers aligning at positions −869 to −801 bp upstream of the coding region (M9 at the promoter region), and +1623 to +1698bp (3′-untranslated region). (B) Representative gels of CREB and p-CREB, and densitometry of the Western blot. The densitometric analysis was performed by imagej 1.30 image analysis software. Each time point represents the levels of CREB or p-CREB relative to that in naïve age-matched chicks (to minimize loading or exposure differences, CREB levels were compared with β-actin in each gel lane; n = 6–8 at each time point and values are + SEM). (C) A representative gel of DNMT3a compared with β-actin, and densitometry of DNMT3a Western blot analysis. (D) A representative gel of DNMT3b compared with β-actin, and densitometry of DNMT3b Western blot analysis. The densitometric analysis was performed by imagej 1.30 image analysis software. Each time point represents the levels of DNMT3a or -b protein relative to that in naïve age-matched chicks (to minimize loading or exposure differences, DNMT levels were compared with β-actin in each gel lane; n = 12–14 at each time point and values are + SEM.) (E) Gel-mobility shift assay to evaluate DNMT binding capacity to the Bdnf promoter region. Samples were prepared from hypothalamic tissues of 3-day-old chicks. After incubation of lysates (6 μg) with DNMT3a antibody (2 μL of antisera), DNMT3b antibody (2 μg) or no antibody, on ice for 60 min, DIG-labeled oligonucleotides corresponding to CpG sites M1, M3 and M9 were added. The mixtures were separated on 8% native polyacrylamide gels and blotted onto a nylon membrane. Visualization was performed using anti-DIG antibody followed by exposure to X-ray film (Ab, antibody; 3a, DNMT3a; 3b, DNMT3b). (A, B and C) *< 0.05 between conditioned and naïve chicks.

Download figure to PowerPoint

CREB was phosphorylated during heat exposure of 3-day-old chicks.

As there was an elevation in p-CREB binding to position M9 at the Bdnf promoter region during heat conditioning, it was tempting to assume that the involvement of CREB would be accompanied by its own regulation. Therefore, we next checked whether CREB was induced and/or activated during heat exposure. Western blot analysis revealed that, although the total level of the CREB protein did not change, there was a significant increase in phosphorylation of CREB which began 2 h into heat exposure and peaked 12 h into conditioning (the levels of p-CREB relative to that in naïve chicks were: 1.51 ± 0.19, F1,13 = 7.43, < 0.02 after 2 h; 1.59 ± 0.24, F1,14 = 5.73, < 0.03 after 6 h; 1.99 ± 0.34, F1,13 = 9.69, < 0.01 after 12 h; and 1.37 ± 0.16, F1,12 = 6.53, < 0.03 after 24 h; Fig. 5B).

DNMT3a expression was induced during heat exposure of 3-days-old chicks and it may bind to specific sites at the Bdnf promoter region

DNMTs are a family of enzymes that catalyze the methylation of cytosine residues (Goll & Bestor, 2005). Given that we observed induction of methylation at two CpG sites during heat exposure of 3-day-old chicks at the Bdnf promoter (M1 and M3), we sought to corroborate these results by determining which DNMT is involved in this process. Therefore, we evaluated the protein expression of both DNMT3a and -3b in the PO/AH of 3-day-old chicks during heat exposure. Western blot analysis of DNMT3a expression indicated an induction in its expression, but there was no change in the expression level of DNMT3b (Fig. 5C and D, respectively). The expression of DNMT3a reached a peak induction of 30% at 6 and 12 h after heat-exposure (F1,25 = 4.85, < 0.04 and F1,24 = 5.35, < 0.03 respectively).

As CpG methylation was induced during heat exposure of 3-day-old chicks at two positions, M1 and M3, and reduced at position M9, we wanted to demonstrate that, indeed, DNMT3a but not -3b can bind to the two inducible sites and not to the site of reduced CpG methylation. Electrophoresis mobility shift assay was used to detect the binding abilities of DNMT3a and -3b to the DNA in the Bdnf promoter at the relevant CpG sites. PO/AH extracts were reacted with antibodies to DNMT3a or -3b, and then with DIG-labeled oligonucleotides, each containing one CpG position (corresponding to M1, M3 or M9); they were then separated on a polyacrylamide gel and visualized with anti-DIG antibody. As can be seen in Fig. 5E, there was a clear supershifting of the DIG-labeled oligonucleotide at positions M1 and M3 when reacted with DNMT3a antibody, whereas there was no binding of DNMT3a to the sequence in position M9. DNMT3b did not bind to any of these positions (Fig. 5E).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The mechanisms by which sensory information affects neural circuits during critical developmental periods are largely unknown. Like other sensory mechanisms, the thermal response set-point is fine-tuned during a critical developmental period by alterations in cellular properties in the frontal hypothalamus (Griffin et al., 1996; Boulant, 2006). These alterations may be regulated by the epigenetic code that determines the repertoire of transcribed proteins. Here, we correlated transient changes in the CpG code in the avian Bdnf promoter region with changes in Bdnf protein expression during heat conditioning and during heat challenge of previously conditioned chicks. Furthermore, we found a correlation between a reduction in CpG methylation at position M9, located −1082 bp downstream of the coding region, within a CREB binding site, CREB activation by phosphorylation and an induction of the p-CREB binding to the Bdnf promoter. In addition we demonstrated that the protein levels of the CpG methylase DNMT3a, but not -3b, were induced during heat conditioning and that DNMT3a binds specifically to the CpG sites on the Bdnf promoter that are methylated during heat conditioning but not to the CREB binding site.

Although all species, ranging from Drosophila to humans, exhibit critical periods for sensory development, very little is known about the epigenetic mechanism that determines the network wiring for this process (Hensch, 2004). The domestic chick offers an optimal model system for studying postnatal sensory development during critical periods as these birds are able to learn immediately after hatching and, more importantly, their behavioral repertoire after hatching can be easily measured (Matsushima et al., 2003).

Previous studies have demonstrated that thermal conditioning induces long-term changes in chicks’ ability to withstand heat stress later in life (Yahav & McMurtry, 2001; Labunskay & Meiri, 2006). We recently characterized the signal transduction pathways underlying this hypothalamic–neuronal network reorganization, showing that it is mediated by R-Ras3 and 14-3-3ε, which in turn activate transcription via Jun. The process is regulated at the translational level by induction of eukaryotic initiation factor 2B expression. It was further determined that Bdnf mRNA expression, but not that of nerve growth factor or neurotrophin 3, is induced during thermal conditioning (Katz & Meiri, 2006; Labunskay & Meiri, 2006; Tirosh et al., 2007; Meiri, 2008). Bdnf is regarded as a key player in many learning systems, including passive-avoidance memory in chicks (Johnston et al., 1999; Johnston & Rose, 2001), rat performance in the Morris water maze learning task (Linnarsson et al., 1997; Croll et al., 1999; Alonso et al., 2002) and during long-term potentiation which serves as a model for memory storage (Kang & Schuman, 1995). Here Bdnf was shown to be not only an effector during conditioning but also a mediator whose expression is correlated with long-term memory storage. We showed that Bdnf is differentially expressed during heat challenge in chicks that were previously conditioned during the critical period relative to age-matched naive chicks. While age limits the induction of Bdnf expression during heat exposure in the critical period (there was no change its expression in 10-day-old chicks that had passed the critical age for thermal control response), in chicks that were previously conditioned, Bdnf was prone to reconditioning by heat challenge.

The repertoire of proteins that are expressed in the cell and hence determine its physiological role is delineated by epigenetic regulation mechanisms which consist of two components, histone modification and DNA methylation (Miller & Sweatt, 2007; Miranda & Jones, 2007; Miller et al., 2008). The N-terminal tail of histones (15–30 amino acids) is subject to posttranslational modifications that modulate chromatin structure and regulate the feasibility of a gene’s transcription (Berger, 2007; Kouzarides, 2007). The second regulatory stage enables transcription by altering the methylation states of CpG dinucleotides within gene promoters (Berger, 2007). The CpG methylation template was initially thought to be restricted to the developmental stage in dividing cells, determining the cell’s phenotypic state (Tajima & Suetake, 1998; Illingworth et al., 2008). However, recent evidence indicates that both DNA methyltransferase activity and CpG methylation occur in mature nondividing neurons (Miller & Sweatt, 2007). CpG methylation has been shown to play a pivotal role in the ability of transcription factors to bind to DNA at specific sites on promoters (Weaver et al., 2007).

Investigation of the CpG code in the promoter regions of several genes, including Bdnf, in several behavioral tasks during both postnatal and adult training has demonstrated dynamic alteration in Bdnf methylation levels during learning tasks (Dennis & Levitt, 2005; Levenson et al., 2006). These studies point to the need for a suitable model to decipher the CpG code. Similar to the advantage of using the chick model for behavioral studies, there is a clear advantage to using this model in studying the epigenetic CpG methylation code in the Bdnf promoter region. In contrast to the complexity of the mammalian sequence, which spans five exons and five introns, each containing several CpG dinucleotides which can serve as regulation points, the chick sequence contains only one exon and all CpG promoter regulation sites are sequentially ordered. Furthermore, each CpG site is located separately and therefore its methylation can be evaluated individually. In this study, we concentrated on six CpG dinucleotide sites (M1–M4, M9 and M12 counting upstream from the Bdnf coding region). We showed that, during heat-conditioning on day 3, methylation levels are induced at positions M1 and M3 and reduced at M9. As the working hypothesis is that changes in CpG methylation affect the binding of transcription factors to the promoter region of the gene, the sequences in the altered methylation sites were analyzed. Indeed, the sequences in which these sites reside were found to be relevant to neuronal plasticity in general and to heat responses in particular. M1 is located within a CEB/P binding site. CEB/P is known to be involved in energy metabolism (Faisst & Meyer, 1992), and to be induced during long-term memory formation (Alberini et al., 1994). M2 resides within an AP1 site, the binding site of Jun, which we have shown to be involved in thermal control establishment (Labunskay & Meiri, 2006). M4 and M12 are located within c-Myb sites, which have been shown to play a synergistic role with the aforementioned CEB/P in activating choline acetyltransferase (Robert et al., 2002) . The most intriguing site is M9, which resides within a CREB site, as CREB is known to activate Bdnf transcription in different conditions. CREB has been correlated with neuronal plasticity and long-term memory (Yin et al., 1994; Barco et al., 2005; Carlezon et al., 2005). Furthermore, the CREB–Bdnf cascade has been implicated in modulating mood (Nair & Vaidya, 2006), and CREB activation has been correlated with Bdnf expression in postnatal development following maternal separation (Lippmann et al., 2007). Our findings show that M9 is demethylated, with the lowest methylation state being registered at 6 and 12 h into heat conditioning (Fig. 3). This time window coincides with an increase in the phosphorylation levels of CREB in the PO/AH and its binding to the Bdnf promoter during heat conditioning (Fig. 5), which corroborate the role of CREB in Bdnf activation. Furthermore, it also matches the induction of Bdnf mRNA (Katz & Meiri, 2006) and protein (Fig. 1) during conditioning, all pointing to the possible mechanism of reduced DNA methylation which enables p-CREB binding to the Bdnf promoter and hence activation of Bdnf transcription.

During heat challenge of previously heat-exposed chicks the methylation of this position was reduced below the naïve level (Fig. 4), again within a time frame similar to that of Bdnf induction (Fig. 2). In naïve 10-day-old heat-exposed chicks we observed an elevation in the methylation levels at all CpG sites in correlation with the fact that the levels of Bdnf in those chicks does not change during heat exposure. As DNA methylation is associated with transcription repression (Miranda & Jones, 2007) it is tempting to speculate that this increase in Bdnf-promoter methylation plays a role in the repression of BDNF induction seen in heat re-exposure of chicks that have been previously conditioned. It should be noted that there are reports of cases in which elevation in CpG methylation results in induction of gene expression (Dennis & Levitt, 2005; Brinkman et al., 2007).

DNMTs are a family of enzymes that catalyze the methylation of cytosine residues (Goll & Bestor, 2005), and their involvement in the observed alterations in DNA methylation pattern in the Bdnf promoter region was therefore explored. We found that DNMT3a is induced 6 and 12 h into conditioning, a time frame which coincides with the elevation in methylation at positions M1 and M3. Furthermore, the DNA at these positions was prone to binding DNMT3a but not DNMT3b, further corroborating the potential role of the former in the CpG methylation process in postdifferentiated neurons. Interestingly, M9 located in the CREB binding site, which was demethylated and therefore not expected to bind the methyl transferase, is not prone to binding either DNMT. This finding provides further support that lack of methylation enables the binding of CREB and activation of Bdnf transcription. It is therefore tempting to conclude that DNMT3a is involved in CpG methylation in the Bdnf promoter during heat conditioning. These results are in agreement with previous findings that indicate the importance of DNMT3a in behavioral plasticity (Levenson et al., 2006).

These data suggest that complex and dynamic changes in DNA methylation are involved in the regulation of thermotolerance acquisition and support the role of DNA methylation in the regulation of gene expression during neuronal plasticity.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We are grateful to the Volcani Institute chicken farm staff for their dedicated work. Supported by the Israeli Science Foundation. Contribution no. 513/07 from the ARO, the Volcani Center, Bet Dagan 50250, Israel.

Abbreviations
Bdnf

brain-derived neurotrophic factor

ChIP

chromatin immunoprecipitation

CpG

cytosine–phosphate–guanine

CREB

cyclic AMP response element binding protein

DAPI

4′,6-diamidino-2-phenylindole dihydrochloride

DIG

digoxigenin

DNMT

DNA (cytosine-5) methyltransferase

EDTA

ethylenediaminetetraacetic acid

PBS

phosphate-buffered saline

PCR

polymerase chain reaction

p-CREB

phosphorylated CREB

PO/AH

preoptic anterior hypothalamus

SDS

sodium dodecyl sulfate

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References