Regulation of BDNF chromatin status and promoter accessibility in a neural correlate of associative learning

Abstract

Brain-derived neurotrophic factor (BDNF) gene expression critically controls learning and its aberrant regulation is implicated in Alzheimer's disease and a host of neurodevelopmental disorders. The BDNF gene is target of known DNA regulatory mechanisms but details of its activity-dependent regulation are not fully characterized. We performed a comprehensive analysis of the epigenetic regulation of the turtle BDNF gene (tBDNF) during a neural correlate of associative learning using an in vitro model of eye blink classical conditioning. Shortly after conditioning onset, the results from ChIP-qPCR show conditioning-dependent increases in methyl-CpG-binding protein 2 (MeCP2) and repressor basic helix-loop-helix binding protein 2 (BHLHB2) binding to tBDNF promoter II that corresponds with transcriptional repression. In contrast, enhanced binding of ten-eleven translocation protein 1 (Tet1), extracellular signal-regulated kinase 1/2 (ERK1/2), and cAMP response element-binding protein (CREB) to promoter III corresponds with transcriptional activation. These actions are accompanied by rapid modifications in histone methylation and phosphorylation status of RNA polymerase II (RNAP II). Significantly, these remarkably coordinated changes in epigenetic factors for two alternatively regulated tBDNF promoters during conditioning are controlled by Tet1 and ERK1/2. Our findings indicate that Tet1 and ERK1/2 are critical partners that, through complementary functions, control learning-dependent tBDNF promoter accessibility required for rapid transcription and acquisition of classical conditioning.

Introduction

Reduced function of brain-derived neurotrophic factor (BDNF) mRNA and protein in brain is implicated in Alzheimer's disease and a host of neurodevelopmental and learning disorders,^1,2 while BDNF elevated expression is a marker for high cognitive function during aging.³ The BDNF gene is a target of several known DNA regulatory mechanisms, such as methylation/demethylation and chromatin remodeling, but the details of its activity-dependent regulation during synaptic plasticity and learning have yet to be fully characterized. There has been intense interest in active DNA methylation/demethylation and the function of the methyl-CpG-binding protein 2 (MeCP2) and ten-eleven translocation (Tet) proteins in gene expression. While MeCP2 binds to methylated CpG dinucleotides to control transcription, the Tet proteins (Tet1–3) convert 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), thought to be an intermediate product in an oxidative demethylation mechanism.^2,4-5 Mutations in the MeCP2 gene underlie the progressive neurodevelopmental disorder Rett syndrome characterized by mental retardation.² Tet1 activity functions in active DNA demethylation and gene regulation during learning and memory.^6,7 Recently, the signaling protein extracellular signal-regulated kinase 1/2 (ERK1/2) has been shown to bind specific DNA sequence motifs and is implicated in regulation of chromatin accessibility and transcription in embryonic stem cells.^8,9 ERK1/2 binds to promoters of developmental genes and supports an open chromatin configuration making them competent for approach by RNA polymerase II (RNAP II) and transcription. This function for ERK1/2 is different from its kinase activity and has not yet been described in mature brain tissue undergoing activity-dependent DNA modifications. It is conceivable that, through their complementary functions, Tet1 and ERK1/2 work together to promote a permissive chromatin state suitable for rapid gene induction during learning.

Expression of mature BDNF protein is a critical signaling element required for synaptic AMPA receptor (AMPAR) delivery and acquisition of learned conditioned responses (CRs) in a neural correlate of eye blink classical conditioning.^10-12 In this isolated preparation of the pons, the cranial nerves are electrically stimulated in place of delivering real stimuli such as a tone or air puff. We have previously characterized portions of the BDNF gene from the pond turtle (tBDNF) and identified several non-coding and one protein coding exon.^13,14 This in vitro model system is uniquely suited for studies of BDNF gene regulation during learning because there is specific up- and downregulation of mRNA transcripts during training, a significant strength of this model, and neuronal pathways underlying the behavior can be experimentally manipulated. Here, we performed a comprehensive analysis of the methylation status, associated histone modifications, and tBDNF promoter binding by regulatory proteins and RNAP II during induction of classical conditioning. The results show that tBDNF promoter occupancy by the transcriptional repressor basic helix-loop-helix binding protein 2 (BHLHB2) and activator cAMP response element-binding protein (CREB), as well as deposition of the histone marks H3K4me3 and H3K27me3 characteristic of active and inactive genes, respectively, are regulated by Tet1 and ERK1/2. These factors drive RNAP II to initiate or repress conditioning-dependent transcription of selective promoters. Our findings indicate that Tet1 and ERK1/2 are critical partners that through complementary functions control learning-dependent tBDNF promoter accessibility required for rapid transcription and acquisition of conditioning.

Results

Rapid changes in tBDNF DNA methylation during conditioning and effects of DNMT inhibitors

The epigenetic mechanisms that regulate gene expression are highly conserved.¹⁵ and have been studied, other than in mammals, in organisms as diverse as Drosophila, zebrafish, birds, and in our model system, the turtle. This is also true of the structure of the BDNF gene (see ref. 13). Each model offers its own advantages for molecular characterization of genetic regulation, which has been exploited to reveal underlying factors involved in normal human cellular function and disease. Detailed study of epigenetic changes in the tBDNF gene in response to associative learning is readily accessible using our in vitro model of eye blink classical conditioning. The experimental set-up for these experiments is shown in Figure 1A. Suction electrodes were used for stimulation and recording of the cranial nerves. A single shock delivered to the trigeminal nerve [the “air puff” unconditioned stimulus (US)] evokes a burst discharge in the abducens nerve (Fig. 1B, upper panel) that represents a neural correlate, or fictive, eye blink response. Pairing of a one second duration train stimulus to the auditory nerve [the “tone” conditioned stimulus (CS)] immediately prior to stimulation of the trigeminal nerve results in burst discharge in the abducens nerve in response to the auditory stimulus. This represents a learned blink CR in response to the “tone” (Fig. 1B, lower panel, arrow). As training proceeds using paired stimuli, acquisition of CRs is rapid and robust (Fig. 1C). Importantly, when the stimuli are unpaired, as during random pseudoconditioning control trials, learning fails to be acquired (Fig. 1C). Therefore, this in vitro model of eye blink conditioning has significant behavioral relevance for understanding epigenetic modifications that underlie learning.

Figure 1.

In vitro model of eye blink classical conditioning. (A) Isolated preparation of the pons in which suction electrodes are used for paired stimulation of the auditory nerve CS (the “tone”) and the trigeminal nerve US (the “air puff”) while recording neural activity from the abducens nerve that generates the neural correlate of the blink response. (B) Representative physiological records show neural discharge in the abducens nerve characteristic of the unconditioned blink response (upper panel) and after acquisition of conditioning in which a burst is recorded during the auditory nerve CS that represents a conditioned response (CR, arrow; lower panel). (C) Acquisition curve of learned responses generated by paired conditioning stimulation. CRs are typically recorded by the second pairing session and asymptote with continued training. Unpaired pseudoconditioning stimuli generate no CRs.

The tBDNF gene structure from the pond turtle T. scripta elegans was analyzed previously^13-14 and is illustrated in Figure 2A. We have identified 4 exons, of which exons I-III are non-coding 5' flanking regions to exon IV, which is the BDNF protein coding region with a 3' UTR. We have previously shown that mRNA transcripts from exon I are not regulated during conditioning, while those from exon II are significantly downregulated and exon III are upregulated.¹³ To assess the impact of epigenetic regulation on tBDNF gene expression during classical conditioning, the pattern of CpG methylation for non-coding promoters and exons I-III was analyzed using bisulfite sequencing PCR (BSP) for naïve brainstem preparations and those that received one session (50 paired or unpaired stimuli lasting 25 min in duration) of pseudoconditioning control or conditioning training trials (Fig. 2B; 3 × 7 clones/group). Primers for BSP analysis are given in Table 1. The positions of the methylated CpG sites for these exons and their promoters are indicated in Figure 2A. Promoter and exon I undergo no significant conditioning-related changes in methylation status after conditioning or control training compared to naïve preparations (Fig. 2B) even though the promoter region has a CREB binding site. This was not surprising since expression of mRNA transcripts derived from exon I do not change with conditioning.¹³ On the other hand, tBDNF promoter and exon II undergo significantly increased methylation following conditioning, while promoter and exon III are demethylated at specific CpG dinucleotide sites (Fig. 2B). For promoter II, 4 sites are primarily methylated (7, 8, 10, 12) including the repressive BHLHB2 transcription factor binding site (P < 0.05) and a CCAAT-enhancer binding protein (C/EBP) binding site (P = 0.005) located just ahead of the transcription start site. Promoter III has several CpG sites that are demethylated most notably the CREB binding site (site 5, P = 0.02; Fig. 2B). These sites undergo surprisingly rapid methylation and demethylation during conditioning. Significantly increased methylation of CpGs occurs for promoter II only 5 min after the onset of conditioning and is further increased after 15 and 25 min (C1) of training (P = 0.01; Fig. 2C). This is observed for both the BHLHB2 and C/EBP binding sites (sites 8 and 10; Fig. 2C). Promoter III is rapidly demethylated 15 to 25 min after training (P = 0.01), as shown for the CREB binding site (site 5; Fig. 2C). Therefore, once pairing is initiated during training, there is remarkably rapid and coordinated regulation of methylation status of tBDNF gene promoters.

Figure 2.

Rapid conditioning-dependent methylation and demethylation of tBDNF promoter regions and differential effect of DNMT inhibitors. (A) tBDNF partial gene structure showing 3 5' non-coding exons (I-III) and the 3' protein coding sequence (IV) with its 3' UTR. Transcription start sites are indicated by the arrows. Promoter and exon regions that contain the CpG sites shown in B are indicated by the gray bars. Diagonal lines represent gaps in sequence data. (B) Bisulfite sequencing PCR (BSP) analysis of the methylation status of CpG sites in promoter and exonic regions I-III from naïve preparations (N) and those pseudoconditioned (Ps1) or conditioned for one training session (C1; 25 min in duration). 3 × 7 clones/group. Transcription start sites are indicated by the arrows. Promoter and exon I show relatively low levels of methylation that do not change with conditioning. Promoter II undergoes significant methylation of specific CpG sites after conditioning, particularly those localized to the BHLHB2 and C/EBP transcription factor binding sites (one-way ANOVA followed by Fisher's post-hoc test, P < 0.05 and P = 0.005, respectively). Promoter III undergoes significant demethylation notably at the CREB binding site (one-way ANOVA followed by Fisher's post-hoc test, P = 0.02). Values in this and all subsequent figures represent means ± SEM. (C) Timing of promoter II methylation and promoter III demethylation after the onset of conditioning training. Total methylation levels for promoter II (CpG sites 1–13) and III (sites 3–13) are shown. Levels of methylation for promoter II are significantly elevated after just 5 min of conditioning and are further increased after 15 and 25 min (F_(3,48) = 3.72, P = 0.01). Data for the BHLHB2 (CpG site 8; one-way ANOVA followed by Fisher's post-hoc test, P = 0.03) and C/EBP (10; P = 0.03 and P = 0.008 at C_15min and C1, respectively) sites are shown individually. Promoter III is demethylated after 15 min of conditioning, which is significantly different from naïve by one session of conditioning (C1 or 25 minutes; F_(3,40) = 3.96, P = 0.01). Data for the CREB binding site (5; one-way ANOVA followed by Fisher's post-hoc test, P = 0.007 and P = 0.002 at C_15min and C1, respectively) is also shown. (D) Treatment with DNMT inhibitors zebularine (100 µM) or RG108 (200 ng/µl) during conditioning suppressed the overall level of methylation of promoters and exons II and III. (E) Total methylation levels of promoter and exon II are moderately suppressed by zebularine and RG108 treatment during conditioning compared to normal conditioning (upper left). However, when specific CpG sites are examined (7, 8, 10, and 12), both drugs significantly inhibit methylation after one session of conditioning (F_(3,12) = 10.09, P = 0.001; upper right). For promoter and exon III, which are normally demethylated in conditioning, drug treatment results in reduced levels of methylation for naïve compared to normal naïve groups but did not affect the conditioned preparations already demethylated (F_(5,60) = 5.98, P = 0.0002).

Table 1.

Primers used for BSP analysis

Target gene	Orientation	Sequence (5′→3′)
BDNF Promoter I	For	TAGTTAATTAATTTATTGGAAGAGGAGATA
	Rev	ACCAAAAATATAAAAAATTCAACCTC
	For	TAAGGAAGAATGTTTTTTGATTAGTTGA
	Rev	CCACTCTCCTTTTAACACAACCTTA
	For	ATATTTTGTAGGAATTAGTATGTAATAAGT
	Rev	AAAATTTTTAAAAACCTAAAAATTTAC
BDNF Promoter II	For	TTTGATATGTTGTTTAATAATAGTATTTAA
	Rev	CTATACCCACATAAAAAAAATAAAC
	For	TTTATTTATAGATGTTGGGGGAAAG
	Rev	AATACCCCTATCCAAAATACTAATC
	For	TTTTTGGAGTATTTTGTGTAGTTTTG
	Rev	CTAAAAATTCAACTTCAACCAACTC
	For	GAGTTGGTTGAAGTTGAATTTTTAG
	Rev	TCACAACTAACACCACAAACAATAAC
BDNF Promoter III	For	TTAAAGTAAAATTTTGTAGTGTAGG
	Rev	ATTAAAAATAATACATCTTTTATTAACAAA
	For	TTTATTAGAGGTAGTTAGGGTATTATATGA
	Rev	AAAACATAACAACAAACAAAAATCCA

We next examined the effect of DNMT inhibitors on the pattern of tBDNF methylation during conditioning. Both zebularine (100 µM) and RG108 (200 ng/µl) treatment results in a dramatic overall reduction in CpG methylation of promoters II and III and their exons (Fig. 2D). The total methylation status of promoter II after one session of conditioning in zebularine or RG108, which is significantly elevated during normal conditioning, is reduced to levels just above those observed for naïve preparations (Fig. 2E). Greater scrutiny of specific CpG sites normally methylated (sites 7, 8, 10, 12) shows a significant reduction after drug treatment compared to normal conditioning (P = 0.001; Fig. 2E). For promoter III, which is highly methylated in the naïve state, DNMT inhibitors reduced the methylation status of treated naïve preparations but essentially had no effect on conditioned preparations as these normally undergo demethylation (P = 0.0002; Fig. 2E). These findings show that DNMT inhibitors differentially interfere with the normal pattern of tBDNF methylation and demethylation of promoters II and III when applied during conditioning.

Conditioning-dependent tBDNF promoter occupancy by regulatory proteins and upstream signaling mechanisms

The pattern of tBDNF DNA methylation observed here suggests that methylation of tBDNF promoter II results in the downregulation of mRNA transcripts while demethylation of promoter III upregulates transcript expression. A critical mechanistic issue is how tBDNF II and III methylation status contributes to access and binding of DNA regulatory proteins to promoter regions. To investigate this, we used ChIP assays and quantitative real-time PCR (qPCR) to analyze naïve and conditioned brainstem samples obtained from the pons. Primers used for promoter II flanked the BHLHB2 and C/EBP binding sites (−158 to −55 nt) while primers for promoter III flanked the CREB binding site (−66 to +22 nt; Table 2).¹⁴ First we examined the activity of the DNA regulatory proteins MeCP2, which binds to methylated CpGs to exert transcriptional control, and Tet1, which is involved in an oxidative demethylation mechanism that may facilitate gene activation. Tet protein orthologs are found throughout metazoan genomes that have retained cytosine methylation. Analysis using ChIP-qPCR showed significantly increased binding of MeCP2 to promoter II after only 15 min of conditioning (P < 0.0001) and corresponding release of binding from promoter III during the same time period (P = 0.02, C vs. N; Fig. 3). In contrast, Tet1 released promoter II during conditioning (P = 0.001) and bound strongly to promoter III (P < 0.001; Fig. 3). These observations correlate well with the rapid conditioning-induced methylation of tBDNF promoter II that would be expected to recruit MeCP2 binding and the active demethylation of tBDNF III by Tet1. Both tBDNF II and III promoters contain functionally significant transcription factor binding sites, namely BHLHB2 and C/EBP for promoter II and CREB for promoter III (Fig. 2B).¹⁴ BHLHB2 has previously been shown to act as a repressor for the mouse Bdnf promoter 4,¹⁶ while CREB is a well known activator of Bdnf. As expected, ChIP-qPCR showed a significant increase in BHLHB2 binding to promoter II (P < 0.0001) and CREB to promoter III after conditioning for 15 min (P < 0.0001; Fig. 3). Finally, recent data has shown that ERK2 binds to specific sequence motifs within human and mouse embryonic stem cell gene promoters.^8,9 We examined this possibility for the tBDNF gene since ERK1/2 has a critical signaling role during conditioning.^17,18 While there was no detectable binding of ERK1/2 to the tBDNF II promoter (data not shown; P = 0.82) binding was increased substantially for tBDNF III after conditioning (P = 0.005; Fig. 3). These data profiling conditioning-dependent tBDNF promoter occupancy using ChIP-qPCR were confirmed by gel shift assays for ERK1/2, CREB, and BHLHB2, but not for Tet1, which may not bind directly to DNA in vivo, or MeCP2, which binds methylated sites directly (data not shown).^5,19

Table 2.

Primers used for tBDNF ChIP assays and qPCR

Target gene	Orientation	Sequence (5′→3′)	Antibody
BDNF Promoter II	For	AGCAAAGGGTCCAGTCAAC	DEC	MeCP2, Tet1, H3K4me3, H3K27me3,  RNAPIISer2, RNAPIISer5
	Rev	CGCTTTAGACAGTTCCTTTCATTT
BDNF Promoter III	For	GTGTCTATTAGAGGCAGCCA	CREB, ERK
	Rev	CGCTCTGAATGGGAGTGT
BDNF Exon II	For	GCGGTGTAGGCTGGAATAG	RNAPIISer2
	Rev	CAGTGGCTAGATCTTGGAGATAAA
BDNF Exon III	For	AAATCAGAGCCAATTTGGGAGT	RNAPIISer2
	Rev	GGCTACTTGTATAAGTCAATATCAACAG

Figure 3.

Conditioning-dependent binding of DNA regulatory proteins to tBDNF promoters II and III and inhibition by PKA and PDK1 blockers. Data are from ChIP-qPCR analysis of tissue samples from naïve (N), pseudoconditioned (Ps), or conditioned (C) preparations for 15 min, or those conditioned for 15 min in the presence of the PKA inhibitor Rp-cAMPs (Rp; 50 µM) or the PDK1 blocker BX-912 (BX; 0.3 µM; n = 5/group). Significant P values between N and C_15min are given in the text. Individual comparisons for promoters II and III between the C_15min and Rp treated groups were, promoter II: MeCP2, P < 0.001; Tet1, P < 0.01; BHLHB2, P = 0.001; and promoter III: MeCP2, P = 0.005; Tet1, P = 0.001; CREB, P < 0.0001; ERK1/2, P = 0.001. Comparisons between C_15min and BX treated groups were, promoter II: MeCP2, P < 0.001; Tet1, P < 0.0001; BHLHB2, P = 0.002; and promoter III: MeCP2, P = 0.07; Tet1, P = 0.001; CREB, P < 0.0001; ERK1/2, P < 0.001. *, Significant differences from naïve in normal saline; +, significant differences from normal conditioning. Representative agarose gels are also shown of the resulting PCR products from the different experimental conditions.

Little is known about how DNA modifications controlling gene expression are regulated by signal transduction mechanisms. Previously, we showed that phosphorylation of PKA initiates classical conditioning within 15 min of training onset and induces expression of BDNF protein that is required for learning.¹⁸ More recently, phosphoinositide-dependent kinase-1 (PDK1) was identified as an early coincidence detector that is activated within 5 min by paired stimuli and primes PKA for downstream function.²⁰ Here, we examined whether PKA and PDK1 signaling are essential for the observed changes in regulatory protein binding of the tBDNF gene during conditioning. Preparations were conditioned in the presence of the competitive PKA inhibitor Rp-cAMPs (50 µM) or the PDK1 antagonist BX-912 (0.3 µM) and analyzed by ChIP-qPCR. The results show that treatment with both of these compounds completely inhibits the conditioning-induced pattern of binding that would normally occur during conditioning for all of the proteins examined and returns these levels to naïve values (Fig. 3). Hence, activation of PKA and PDK1 during conditioning is required to initiate DNA regulatory protein binding to tBDNF promoters during conditioning.

Inhibition of Tet1 and ERK1/2 alters promoter occupancy and suppresses transcription factor binding

Proteins that bind DNA recruit other partners to form activator or repressor complexes. Here, using co-immunoprecipitation (co-IP), we investigated whether Tet1 forms complexes with other tBDNF binding proteins. As shown in Figure 4A, MeCP2 interacts with Tet1 weakly in naïve preparations but this association is significantly increased after conditioning (P = 0.01, C groups vs. N). In contrast, CREB immunoprecipitates strongly with Tet1 in naïve preparations but declines sharply to near undetectable levels with conditioning (P = 0.0002, C groups vs. N; Fig. 4B). Finally, an ERK1/2-Tet1 interaction is significantly increased with conditioning for 15 min (P = 0.004, C_15min vs. N; Fig. 4C) but declines with longer training periods to naïve levels. An ERK1/2-CREB interaction was also examined but showed minimal immunoprecipitates suggesting little to no association of these proteins. Tet1, therefore, shows conditioning-dependent partnering with MeCP2, CREB, and ERK1/2.

Figure 4.

Tet1 partners with MeCP2 and ERK1/2 but dissociates from CREB during conditioning. (A–C) Co-immunoprecipitation of the protein-protein interactions between (A) MeCP2, (B) CREB, and (C) ERK1/2 with Tet1 protein in naïve and conditioned preparations that were trained for 15 or 25 (C1) min. P values are given in the text (n = 5/group). Input lanes are whole cell lysates from naïve samples and control IgG lanes are shown. Loading controls are also shown, except for MeCP2 that is detected at the same molecular weight as IgG heavy chain.

tBDNF promoter III is strongly demethylated during conditioning that corresponds with the binding of Tet1, ERK1/2 and the activator CREB, and increased mRNA expression. In contrast, tBDNF promoter II is methylated in conditioning and bound by MeCP2 and the repressor BHLHB2, and is associated with downregulated transcripts. We further examined whether Tet1 and ERK1/2 have a critical role in regulating tBDNF gene expression during conditioning using inactivation procedures. First, an anti-Tet1 siRNA was employed (see Methods for details). Western blots for Tet1 show a single ~255 kDa molecular weight band in naïve turtle brainstem tissue (Fig. 5) close to the predicted molecular weight for Tet1 in the pond turtle Chrysemys picta bellii of 244 kDa. The predicted molecular weights of Tet2 and Tet3 in Chrysemys are much lower, at 215 kDa and 211 kDa, respectively, making them easily identifiable in Westerns, similar to the situation in human and mouse. The siRNA was designed to target a conserved sequence in turtle that corresponds to human exon 4, which is exclusive to the Tet1 isoform. Bath application of anti-Tet1 siRNA results in an average knockdown of Tet1 protein to 54% after 24 hr of incubation, as shown by Western blot (150 nM; Fig. 5). Treatment of preparations with Tet1 siRNA does not result in reduction in protein expression for Tet2, Tet3, CREB, or ERK1/2 (Fig. 5), thereby verifying its specificity. ChIP-qPCR assays of tBDNF promoter III in these conditions show not only a decrease in Tet1 binding, as expected, but also a significant reduction in CREB and ERK1/2 binding, compared to normal conditioning (F_(1,28) = 23.03, P < 0.0001), while MeCP2 increased to naïve values (P = 0.03; Fig. 5). In contrast, examination of promoter II by ChIP shows that the repressor BHLHB2 is significantly reduced compared to normal conditioning after anti-Tet1 siRNA treatment (P = 0.0002; Fig. 5). To investigate whether ERK1/2 inactivation also alters tBDNF promoter occupancy during conditioning, the highly selective MEK1/2 inhibitor PD0325901 (PD; 1 µM) was tested. Bath application of PD during 15 min of conditioning resulted in reduced binding of CREB, Tet1, and ERK1/2 (F_(1,28) = 23.22, P < 0.0001) and increased MeCP2 binding (P < 0.0001) to promoter III compared to normal conditioning (Fig. 5). PD also significantly reduces BHLHB2 binding to promoter II (P = 0.02; Fig. 5). These findings indicate that inhibition of either Tet1 or ERK1/2 severely alters the efficiency of regulatory protein binding to tBDNF during conditioning. Moreover, the treatment interferes with binding of the transcriptional repressor BHLHB2 to tBDNF promoter II and the transcriptional activator CREB to promoter III, which ultimately generates BDNF protein required for conditioning.

Figure 5.

Inhibition of Tet1 or ERK1/2 suppresses regulatory DNA protein binding to tBDNF promoters as determined by ChIP assays. Preparations were treated with an siRNA directed against turtle Tet1 (24 hr incubation, 150 nM) or the MEK1/2 inhibitor PD0325901 (1 hr, 1 µM) during 15 min of conditioning. The Western blots show a substantial reduction in Tet1 protein expression by application of the siRNA to naïve preparations to a mean of 54% (n = 4). The marker for Tet1 indicates 250 kDa. Westerns for Tet2, Tet3, CREB, and ERK1/2 show no reduction in protein expression after siRNA treatment verifying the specificity of the Tet1 siRNA. Both the Tet1 siRNA and PD resulted in significant alterations in tBDNF binding by all of the proteins examined compared to normal conditioning. Data from the conditioned group shown in Figure 3 are repeated here for direct comparison. P and F values are given in the text (n = 5/group). Individual comparisons for promoters II and III between the C_15min and Tet1 siRNA treated groups were, promoter II: BHLHB2, P = 0.0002; and promoter III: MeCP2, P = 0.03; Tet1, P = 0.001; CREB, P < 0.001; ERK1/2, P = 0.003. Comparisons between C_15min and PD treated groups were, promoter II: BHLHB2, P = 0.02; and promoter III: MeCP2, P < 0.0001; Tet1, P < 0.05; CREB, P < 0.0001; ERK1/2, P < 0.001. *, Significant differences from naïve in normal saline; +, significant differences from normal conditioning.

Regulation of histone modifications and RNAP II promoter binding by Tet1 and ERK1/2

Distinct histone modifications are associated with various gene transcriptional states. Active promoters are marked by trimethylation of histone H3 lysine 4 (H3K4me3), whereas repressed promoters are associated with trimethylation of histone H3 lysine 27 (H3K27me3).²¹ We examined the rapid modifications of these marks in tBDNF promoter regions during the initiation of conditioning in order to better understand the regulation of transcription factor binding observed here. Antibodies that recognize the H3K4me3 and H3K27me3 histone modifications show rapid and extremely robust increases after conditioning for 15 min but not after 5 min (P < 0.0001, 15 min vs. 5 min for both marks) that are maintained during the training procedure for a total of 80 min (2 pairing sessions or C2; Fig. 6A). Therefore, these specific histone modifications are exquisitely responsive to the application of the conditioning stimuli. We then used ChIP-qPCR assays to detect changes in these marks with the same primers used to examine tBDNF promoter II and III occupancy, which are in the regions flanking the BHLHB2 and CREB transcription factor binding sites. The results show that, in the naïve state, promoters II and III are positive for both the active H3K4me3 and repressive H3K27me3 histone marks. After 15 min of conditioning, H3K4me3 in promoter III increases dramatically (P < 0.0001), while H3K27me3 declines (P < 0.01; Fig. 6B) relative to naïve. These histone changes are consistent with promoting access for CREB binding followed by transcriptional activation during conditioning. Interestingly, for promoter II, which undergoes conditioning-dependent transcriptional repression, both histone marks decreased after 15 min of conditioning (P < 0.0001 for both marks vs. N; Fig. 6B). A possible explanation for this is discussed below. Since Tet1 and ERK1/2 were found to regulate binding of the transcription factors BHLHB2 and CREB, we tested whether they might also alter the induction of histone modifications during conditioning. Indeed, application of either Tet1 siRNA or the MEK1/2 blocker PD during 15 min of conditioning significantly reduced levels of H3K4me3 compared to normal conditioning (F_(2,12) = 19.45, P < 0.001, comparison of C₁₅, Tet1 siRNA, PD groups) and elevated H3K27me3 (F_(2,12) = 4.70, P = 0.03) in promoter III. These findings on histone status for promoter III correspond with the inhibition of CREB binding observed following these treatments (Fig. 5). For promoter II, these agents had no effect on conditioning-related changes to H3K4me3 (F_(2,12) = 0.69, P = 0.52, comparison of C₁₅, Tet1 siRNA, PD) and partially reversed those for H3K27me3 (F_(2,12) = 8.21, P = 0.006), corresponding with a reduction in BHLHB2 binding.

Figure 6.

Histone modifications and RNAP II tBDNF binding during conditioning are regulated by Tet1 and ERK1/2. (A) Western blots showing dramatically increased protein levels for both H3K4me3 (P < 0.0001) and H3K27me3 (P < 0.0001; n = 5/group) after 15 min of conditioning but not after 5 min. These levels are maintained for at least 80 min after 2 pairing sessions of conditioning (C2). (B) ChIP analysis reveals significantly elevated H3K4me3 (P < 0.0001) and reduced H3K27me3 (P < 0.01; n = 5/group) tBDNF promoter III levels after 15 min of conditioning. The H3K4me3 and H3K27me3 marks for promoter II were both significantly reduced by conditioning (P < 0.0001). Treatment of preparations with Tet1 siRNA or the MEK1/2 inhibitor PD during conditioning interfered with these histone modifications such that they were either suppressed or reversed (P values in the text). ChIP assays of phosphorylated forms of RNAP II tBDNF binding showed markedly elevated levels for promoter III after 15 min of conditioning (Ser5, Ser2 promoter, Ser2 exon, all P < 0.0001; Ser5 and Ser2 promoter, n = 5/group; Ser2 exon, n = 3/group) compared to naïve. Tet1 siRNA or PD treatment inhibited promoter III RNAP II binding to naïve values (Ser5, F_(2,12) = 138.52, P < 0.0001; Ser2 promoter, F_(2,12) = 16.76, P = 0.0003; Ser2 exon, F_(2,6) = 52.30, P = 0.0002; comparison of C_15min, Tet1 siRNA, PD). RNAP II binding to promoter II after conditioning showed surprisingly high values of the Ser5 phosphorylated form compared to naïve. However, Ser2 at both promoter and exonic sites was significantly lower than naïve (Ser2 promoter and exon, P < 0.001; Ser5 and Ser2 promoter, n = 5/group; Ser2 exon, n = 3/group). Tet1 and PD treatment also interfered with the normal activity of RNAP II during conditioning (Ser5, F_(2,12) = 20.42, P = 0.0001; Ser2 promoter, F_(2,12) = 4.55, P < 0.05; Ser2 exon, F_(2,6) = 1.13, P = 0.38; comparison of C_15min, Tet1 siRNA, PD). *, Significant differences from naïve in normal saline; +, significant differences from normal conditioning.

Permissive chromatin marks, such as H3K4me3, are thought to promote the accessibility of RNAP II at promoters to activate transcription. RNAP II phosphorylated at Ser5 (RNAPIISer5) is involved in initiation of transcription, while phosphorylation at Ser2 (RNAPIISer2) allows mRNA elongation, and both are associated with activated genes.²¹ Using ChIP, we next examined the association of these forms of RNAP II to tBDNF promoters after conditioning and treatment with Tet1 and ERK1/2 inhibitors. After 15 min of conditioning, promoter III showed greatly increased levels of both RNAPIISer5 and RNAPIISer2 (Ser5, P < 0.0001; Ser2 promoter, P < 0.0001, C₁₅ vs. N; Fig. 6B) indicating active transcription. To confirm this interpretation, ChIP assays were also performed for RNAPIISer2 on exonic regions of III and verified that mRNA elongation beyond the promoter and transcription start site was likely (Ser2 exon, P < 0.0001, C₁₅ vs. N; Fig. 6B). Significantly, treatment with either the Tet1 siRNA or PD dramatically inhibited RNAPIISer5 and RNAPIISer2 binding, as shown by ChIP (Ser5, P < 0.0001; Ser2 promoter, P = 0.0003; Ser2 exon, P = 0.0002, comparison of C₁₅, Tet1 siRNA, PD; Fig. 6B). Therefore, within 15 min of conditioning onset, favorable histone modifications allowed access of RNAP II for active transcription of tBDNF III and these processes were disrupted by interference with Tet1 or ERK1/2 activity. Interestingly, promoter II, which is normally repressed in conditioning, showed relatively high levels of RNAPIISer5 from the ChIP assays in both naïve and conditioned preparations. However, unlike tBDNF III, levels for RNAPIISer2 during conditioning were significantly lower compared to naïve for both the promoter and exonic regions (P < 0.001 both promoter and exon, C₁₅ vs. N; Fig. 6B). These results suggest that RNAP II is stalled at promoter II during conditioning. Conditioning during Tet1 siRNA and PD treatment also showed they interfered with the normal activity of RNAP II at tBDNF II (Ser5, P = 0.0001; Ser2 promoter, P < 0.05; Ser2 exon, non-significant, comparison of C₁₅, Tet1 siRNA, PD; Fig. 6B). Together, these results indicate that both Tet1 and ERK1/2 regulate the rapid histone modifications induced by conditioning that interface with the activity of RNAP II, resulting in stalled or active transcriptional activity at tBDNF promoters.

Epigenetic modifications control tBDNF mRNA expression and acquisition of classical conditioning

After only 15 min of conditioning, tBDNF2a-d promoter II mRNA transcripts are significantly downregulated (F_(7,16) = 26.49, P < 0.0001, C₁₅ vs. N) while tBDNF3b promoter III transcripts are dramatically upregulated, nearly threefold, compared to naïve preparations using semi-quantitative analysis (P < 0.0001; Fig. 7A). These findings are similar to those previously reported after one complete training session (25 min) of conditioning.¹³ To evaluate whether inhibition of Tet1 and ERK1/2 affects tBDNF transcript expression, mRNA was analyzed from preparations conditioned for 15 min in the presence of Tet1 siRNA or PD. The results show that both compounds clearly inhibit transcriptional expression of promoter III transcripts while they weakly enhance downregulation of promoter II transcripts. Notably, transcription of tBDNF3b is dramatically reduced in the treatment groups back to naïve levels compared to normal conditioning (F_(2,6) = 65.23, P < 0.0001, comparison C₁₅, Tet1 siRNA, PD; Fig. 7A).

Figure 7.

Disruption of tBDNF mRNA expression by Tet1 siRNA and an inhibitor of ERK1/2 blocks acquisition of conditioned responding. (A) During normal conditioning mRNA transcripts from promoter II are downregulated (F_(7,16) = 26.49, P < 0.0001; n = 3/group) while promoter III transcript 3b is significantly upregulated by threefold (P < 0.0001) compared to naïve. Tet1 siRNA treatment further lowered expression levels of promoter II transcripts (2c, P = 0.004; 2d, P = 0.001) compared to normal conditioning, while both Tet1 siRNA and PD blocked expression of promoter III transcript 3b (F_(2,6) = 65.23, P < 0.0001). *, Significant differences from naïve in normal saline; +, significant differences from normal conditioning. (B) Treatment of preparations with Tet1 siRNA (n = 5/group) or the MEK inhibitor PD (n = 3/group) completely blocks the acquisition of CRs during training for 2 pairing sessions. However, application of a negative control siRNA (NC siRNA; n = 4/group) resulted in robust conditioning that typically occurs in session 2. Electrophysiological records show extracellular recordings from the abducens nerve during application of the CS to the auditory nerve followed by the US to the trigeminal nerve which results in expression of a burst discharge characteristic of the unconditioned “blink” response. A CR, a burst discharge in response to the CS, is indicated by the arrow for a preparation treated with control siRNA.

From these data, showing that inhibition of Tet1 and ERK1/2 alters transcription factor binding, histone modifications, RNAP II, and transcription of tBDNF II and III, it would be predicted that both Tet1 siRNA and PD should also interfere with acquisition of conditioning. This was observed to be the case. When Tet1 siRNA was applied to the bath prior to and during training, there was no acquisition of CRs (mean 0% CRs), even though the unconditioned reflex response was normal (Fig. 7B). The same results were obtained for preparations conditioned during PD treatment (mean 0% CRs). For comparison, treatment of preparations with a negative control siRNA resulted in robust conditioned responding (mean 95% ± 4 CRs). We additionally tested the effects of zebularine on conditioning and this treatment also resulted in 0% CRs (data not shown). These data confirm that expression of the appropriate pattern of tBDNF mRNA is essential for classical conditioning and indicate that Tet1 and ERK1/2 exert critical regulatory control over gene expression required for this form of learning.

Discussion

We have performed a comprehensive analysis of key epigenetic factors controlling tBDNF gene expression during a neural correlate of eye blink classical conditioning. The primary finding of this study is that there is a significant role of Tet1 and ERK1/2 in the rapid regulation of tBDNF during conditioning. We have shown that interfering with the actions of Tet1 and ERK1/2 altered conditioning-dependent histone modifications, transcription factor and RNAP II DNA binding, expression of tBDNF transcripts encoded from promoters II and III, and inhibited acquisition of conditioning. These data suggest that Tet1 and ERK1/2 are critical partners that, through complementary functions, control chromatin status, promoter accessibility for regulatory proteins and RNAP II, and, ultimately, rapid induction of gene expression required for learning.

Methylation and demethylation of tBDNF promoter CpG sites occurs surprisingly rapidly after the onset of the training procedure within regions of transcription factor binding sites. Although there are no comparable studies of eye blink classical conditioning, DNA demethylation was observed to occur in the rat Bdnf gene 2 hours after contextual fear training, which was associated with mRNA expression²² in as little as 30 min after fear training.²³ These data demonstrate that learning-dependent changes in DNA methylation may also occur rapidly in mammals. ChIP-qPCR assays show conditioning-dependent increases in binding of MeCP2 and transcriptional repressor BHLHB2 to tBDNF promoter II, which corresponds with methylation, and binding of Tet1, ERK1/2, and transcriptional activator CREB to promoter III, which corresponds with demethylation after only 15 min of conditioning. These processes are illustrated in the model in Figure 8. In the naïve state, promoter II is bound strongly by Tet1 to maintain it in a hypomethylated state for basal expression of tBDNF II mRNA transcripts.⁷ Shortly after conditioning onset, promoter II is rapidly methylated (within 5 min), released by Tet1, and bound by MeCP2 and BHLHB2 to suppress tBDNF II transcripts. The transcription factor BHLHB2 has been shown to be an activity-dependent repressor of Bdnf promoter 4 in mouse that recognizes DNA E-box sequences.¹⁶ Alternately, methylated sites in promoter III are bound by MeCP2 in naïve preparations. After conditioning, MeCP2 is released and Tet1 binds tightly resulting in active demethylation of promoter III allowing binding by ERK1/2 and CREB to initiate transcription of tBDNF III transcripts. MeCP2 is known to bind methylated CpGs and act as a transcriptional repressor.² On the other hand, Tet1 functions in an active oxidative demethylation process^4,5 that has recently been shown to have a critical role in gene expression during learning and memory.^6,7 However, the function of Tet1 in learning-dependent gene expression has recently come into question, as Tet3, but not Tet1, has been shown to be important in fear extinction learning.²⁴ Our co-IPs further indicate conditioning-dependent interactions among regulatory proteins, given the caveat that they evaluate the global status of these protein-protein interactions across the genome. Specifically, MeCP2 strongly associates with Tet1 during conditioning suggesting the possibility that MeCP2 recruits Tet1 to CpG sites to be demethylated. This interaction has been described previously but was not shown to be activity-dependent.¹⁹ It is interesting that the MeCP2-Tet1 interaction is sustained during conditioning but that they have opposite tBDNF binding profiles. We speculate that these 2 proteins rotate around one another to rapidly interact with DNA in an activity-dependent manner as is illustrated in Figure 8. Tet1 and CREB also show a significant association in the naïve state that is broken during conditioning allowing binding to separate DNA sites (Fig. 8). Finally, we observed a strong transient interaction between Tet1 and ERK1/2 suggesting that Tet1 may recruit ERK1/2 to active promoters. It is currently unclear which proteins recruit specific partners and the timing of these actions during conditioning, but this issue can be resolved by further analysis using ChIP-reChIP.

Figure 8.

Model of tBDNF transcriptional regulation during classical conditioning. The naïve state is shown to the left and 15 min after conditioning onset (C_15min) is shown to the right of the arrows. The DNA segments illustrated represent the regions covered by primer pairs for ChIP assays flanking transcription factor binding sites (promoter II, −158 to −55 nt; promoter III, −66 to +22 nt). Promoter II is hypomethylated in the naïve state by Tet1 and bound by RNAPIISer5 that initiates active transcription. After conditioning, Tet1 is released and tBDNF II undergoes transcriptional repression by increased methylation and binding by MeCP2 and the transcription factor BHLHB2. While RNAP II remains phosphorylated at Ser5 and bound to tBDNF, the levels of RNAPIISer2 decline and transcription is suppressed. Promoter III is methylated and bound by MeCP2 in naïve preparations. It is transcriptionally activated during conditioning following demethylation by Tet1 and release of MeCP2, which allows access of ERK1/2 and transcription factor CREB. There is also dramatically increased binding of RNAPIISer5 and Ser2 required for successful initiation and elongation for mRNA transcription. Since a Tet1-MeCP2 interaction is sustained during conditioning as shown by co-IP, we speculate that these proteins rotate around one another to rapidly interact with DNA as illustrated.

Recently, we found that coincidence detection of paired stimuli during the earliest stages of classical conditioning is mediated by PDK1 and PKA.²⁰ To demonstrate that this process is mechanistically linked to tBDNF gene expression, we performed ChIP assays in the presence of their inhibitors. The data show suppression of Tet1, ERK1/2, and CREB binding to promoter III, as well as BHLHB2 binding to promoter II, during conditioning in BX or Rp-cAMPs. Exactly how conditioning-dependent activation of PDK1 and PKA stimulates rapid changes in tBDNF gene expression requires further study but likely involves the calcium-calmodulin-dependent protein kinases²⁵ known to be activated in this form of conditioning.¹⁸

Conditioning-dependent histone modifications accompany the changes observed in tBDNF promoter binding to control access of regulatory proteins. Active gene promoters show distinctive chromatin signatures, such as the H3K4me3 mark, while repressed promoters are associated with H3K27me3.²¹ We observed a rapid and robust conditioning-induced global increase in both histone marks by Western blots after 15 min of conditioning but not after 5 min. Significantly, ChIP assays show tBDNF promoter III undergoes greatly increased H3K4me3 binding and a corresponding decrease in H3K27me3 during conditioning. These changes correlate well with enhanced access of CREB and activation of tBDNF III transcripts. This interpretation is corroborated by ChIP analysis of RNAP II showing dramatically increased binding of RNAPIISer5, the form that initiates transcription, at promoter III as well as RNAPIISer2 at both promoter and exonic sites, verifying successful mRNA elongation (Fig. 8). Nearly the opposite is true for promoter II which is repressed in conditioning. While ChIP assays reveal high levels of H3K4me3 and H3K27me3 in the naïve state, both marks are significantly reduced during conditioning. At first glance, it is perplexing why both, particularly the repressive H3K27me3 mark, would decline. However, tBDNF II must be accessible to the transcriptional repressor BHLHB2. A recent study of nucleosome-interacting proteins and their regulation by chromatin modifications provided evidence that BHLHB2 was very strongly excluded when the H3K4me3 or H3K27me3 marks were present.²⁶ Therefore, access of BHLHB2 to tBDNF II may require a significant reduction in both marks. Correspondingly, while RNAPIISer5 shows strong binding to promoter II after conditioning, indicating that RNAP II is poised to initiate transcription, RNAPIISer2 is dramatically reduced at both promoter and exonic sites suggesting it is stalled and unable to engage in active mRNA elongation. Together, these findings provide evidence that rapid changes in methylation status accompanied by histone modifications and transcription factor binding control tBDNF transcript expression required for classical conditioning. The functional significance of exon-specific regulation of tBDNF transcripts during learning is unknown. In the case of conditioning in the turtle, tBDNF exon II transcripts are downregulated, while those from III are significantly upregulated and exon I transcripts are not altered during conditioning.¹³ This feature of BDNF gene expression has now been repeatedly described across many different behavioral paradigms and it has been speculated to be related to brain region, type of neuron, or specific environmental stimuli.

From our ChIP assays, we additionally demonstrated the presence of both H3K4me3 and H3K27me3 histone marks in high levels at tBDNF promoters II and III in the naïve state. These findings suggest that tBDNF may contain bivalent domains. Bivalent promoter domains contain chromatin signatures characteristic of both gene activation and repression and poise genes for rapid transcriptional regulation.²¹ They have been largely recognized in developmental genes in embryonic stem cells. However, the BDNF gene in mature brain may be uniquely suited to contain bivalent domains in order to be poised for rapid transcriptional activation by environmental stimuli, such as is necessary for learning. Such domains may also be present for Bdnf promoters in rat.²⁷ While ChIP-reChIP assays may provide compelling data, the presence of bivalent domains cannot be unequivocally demonstrated unless homogeneous cell populations or single-cell approaches are used to establish that both marks occur simultaneously.²¹

Significantly, inhibition of Tet1 by siRNA or ERK1/2 by the MEK antagonist PD during conditioning revealed that both are critical for controlling tBDNF gene expression. Application of either inhibitor alone interfered with histone modifications, binding of DNA regulatory proteins and RNAP II, transcription, and blocked acquisition of conditioning. Surprisingly and most compellingly, inhibition of either Tet1 or ERK1/2 had similar effects on their targets suggesting that the two work together as regulators of gene expression. Evidence in favor of this idea comes from observations that not only do Tet1 and ERK1/2 co-IP in early conditioning, they occupy the tBDNF III CREB binding site during conditioning, although promoter co-occupancy has not been firmly demonstrated here. The signaling protein ERK1/2 binds specific DNA sequence motifs and has recently been implicated in regulation of chromatin accessibility in embryonic stem cells.^8,9 Promoter binding by ERK2 is associated with high levels of histone modifications related to active genes, such as H3K4me3⁸ and in vitro evidence indicates that it phosphorylates RNAP II at Ser5,⁹ thus placing developmental genes in a poised state to be readily activated during differentiation. We have established here that ERK1/2 binds tBDNF promoter III in mature brain tissue and has a significant regulatory role in shaping chromatin features and promoting transcription by RNAP II. Our results for tBDNF are consistent with findings from these developmental studies. ERK1/2 binding is detected at activated tBDNF promoter III but not repressed promoter II, and RNAPIISer5 binding is severely suppressed by inhibition of ERK1/2 by PD during conditioning. In fact, Western blots showed RNAPIISer5 was markedly reduced during conditioning in PD (data not shown), suggesting that RNAP II at Ser5 may be a substrate for ERK1/2 kinase activity. For in vitro classical conditioning, we hypothesize that activated tBDNF promoter III is demethylated by Tet1 which promotes access of ERK1/2 and CREB to initiate transcription for rapid gene induction during conditioning. Tet1 may additionally facilitate H3K4me3 histone deposition by recruiting SET proteins²¹ while ERK1/2 phosphorylates RNAP II at Ser5⁹ to prime the transcriptional machinery.

Materials and Methods

Animals and conditioning procedures

Freshwater pond turtles, Trachemys scripta elegans, of either sex were purchased from commercial suppliers and anesthetized by hypothermia until torpid and decapitated. All experiments involving the use of animals were performed in accordance with the guidelines of the National Institutes of Health and were approved by the Institutional Animal Care and Use Committee. Brainstems were transected at the levels of the trochlear and glossopharyngeal nerves and the cerebellum was removed as described previously leaving an isolated preparation of the pons.²⁸ The preparation was continuously bathed (2–4 ml/min) in physiological saline containing (in mM): 100 NaCl, 6 KCl, 40 NaHCO₃, 2.6 CaCl₂, 1.6 MgCl₂, and 20 glucose, which was oxygenated with 95% O₂/5% CO₂ and maintained at room temperature (22–24°C) at pH 7.6. Evidence indicates that epigenetic modifications such as DNA methylation may be affected by environmental conditions including temperature.²⁹ Whether the procedure using hypothermia to induce torpidity in turtles has any effect on epigenetic marks is unknown. However, the preparation is allowed to warm to room temperature following surgery and the experimental conditioned preparations are compared to naïve, unstimulated preparations that serve as controls. Therefore, the epigenetic changes we report here are related specifically to conditioning-dependent modifications in the tBDNF gene. Suction electrodes were used for stimulation and recording of cranial nerves. The unconditioned stimulus (US) was a twofold threshold single shock applied to the trigeminal nerve and the conditioned stimulus (CS) was a 100 Hz, 1 s train stimulus applied to the ipsilateral auditory nerve that was below threshold amplitude required to produce activity in the abducens nerve. Neural responses were recorded from the ipsilateral abducens nerve that innervates the extraocular muscles controlling movements of the eye, nictitating membrane, and eyelid. The CS–US interval, defined as the time between the CS offset and the onset of the US, was 20 ms. The intertrial interval between the paired stimuli was 30 s. One complete training session (C1) was composed of 50 CS–US paired presentations that lasted 25 min in duration, however, shorter periods of training were also used such as 5 min (10 paired stimuli) or 15 min (30 stimuli). When two pairing sessions (C2) were applied, the sessions were separated by a 30 min rest period in which there was no stimulation. Conditioned responses were defined as abducens nerve activity that occurred during the CS and exceeded an amplitude of twofold above the baseline recording level. Pseudoconditioned (Ps) control preparations received the same number of CS and US stimuli as conditioned preparations but were explicitly unpaired using a CS–US interval randomly selected between 300 ms and 25 s. Naïve brainstems were presented with no stimuli and remained in the bath for the same time period as experimental preparations. Tissue samples for analysis were comprised of the pons (from cranial nerves IV to VIII) cut in half after the experiment to retain the stimulated side for analysis. This tissue sample contains the pontine portion of the eye blink cranial nerve circuitry.³⁰

Pharmacology

Zebularine (100 μM; Calbiochem), a cell-permeable cytidine analog and DNA methylation inhibitor that acts by covalently bonding with DNA methyltransferase (DNMT), and RG108 (200 ng/μl; Axon Medchem), a non-nucleoside DNMT inhibitor that blocks the enzyme active site, were applied to the bath for 1 h prior to and throughout the conditioning procedures. The selective cAMP analog and competitive inhibitor of PKA activation Rp-cAMPs (50 μM; Sigma) was applied for 1 h while the competitive 3-phosphoinositide-dependent kinase-1 (PDK1) antagonist BX-912 (0.3 μM; Calbiochem) was administered for 2 h prior to conditioning and continued throughout the training procedures. The selective MEK1/2 blocker PD0325901 (1 µM; SelleckChem.com) was applied 1 h prior to conditioning.

Bisulfite sequencing PCR

Methylation status of tBDNF promoter and exon regions was analyzed using bisulfite sequencing PCR (BSP). Genomic DNA was isolated from turtle brainstem tissue using the Qiagen DNeasy mini kit (Qiagen). DNA samples were treated with bisulfite reagent using the EZ DNA methylation-lightning kit (Zymo Research). Bisulfite-treated samples were amplified by PCR using primers that amplify the specific tBDNF promoter regions of all 3 previously identified noncoding exons (I-III).^13,14 Primers for BSP were designed using Methprimer software (http://www.urogene.org/methprimer) and are listed in Table 1. PCR reactions were carried out and amplified products were purified using Purelink Quick gel extraction and PCR purification Combo kit (Invitrogen) and cloned into pGEMT-easy vectors (Promega). Selected colonies from each sample were sequenced by the Iowa State University DNA Facility.

Reverse transcriptase PCR

To examine expression of tBDNF mRNA transcripts, RT-PCR was performed for all experimental groups followed by semi-quantitative analysis. Real-time PCR could not be used for quantification because the transcript sequences are overlapping and the required PCR product would be too long for optimal amplification.  Total RNA was extracted from brainstem samples and an equal concentration (2.0 µg/sample) was reverse transcribed using a 3′ RACE adapter and M-MLV Reverse Transcriptase at 42°C for 1 h. To analyze the expression of the tBDNF transcripts, RT-PCR was performed in 2 steps. cDNA (1 µl) was amplified in a total volume of 25 µl using the Accuprime Pfx polymerase system. The primary RT-PCR was carried out by using exon specific outer primers and 3′ outer primers. We optimized the number of cycles ranging from 22 to 36 and also the different levels of input sample for 2 rounds to determine the exponential phase of the RT-PCR. The optimal number of cycles was recorded and used to carry out the experiments. Primer sequences were previously published.¹³ Conditions for the RT-PCR reaction were: initial denaturation at 94°C for 2 min, 25 cycles at 94°C for 30 sec, 62°C for 30 sec (exon II) or 60°C for 30 sec (exon III), 68°C for 2 min and final extension at 68°C for 10 min. The secondary RT-PCR was carried out by using each primary RT-PCR product as a template with exon specific inner primers and 3′ inner primers. Conditions for the RT-PCR reaction were: initial denaturation at 94°C for 2 min, 30 cycles at 94°C for 30 sec, 60°C for 30 sec, 68°C for 2 min and followed by final extension at 68 °C for 10 min. Samples were confirmed to be free of DNA contamination by performing reactions without reverse transcriptase. Each RT-PCR product was electrophoresed onto 2.0% agarose gels and stained with ethidium bromide. Images of the amplified products were acquired and the density of each band with background subtraction was measured using the InGenius Bio Imaging System (Syngene). Levels of mRNA expression were normalized with respect to β-actin bands for each sample.

Chromatin immunoprecipitation (ChIP) assays

ChIP assays were performed as previously described by Weinmann et al.³¹ with minor modifications. Briefly, brainstem samples taken from the pons ipsilateral to the stimulation were minced on ice immediately after the conditioning procedures and incubated in 1% formaldehyde at 37°C for 10 min. The tissue was washed in ice-cold PBS containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 µg/ml protease inhibitor mixture) and homogenized with SDS lysis buffer. Chromatin was sheared by sonication and lysates were centrifuged at 13,000 rpm for 10 min at 4°C. A portion of the total chromatin sample was set aside for input controls. The supernatant containing the sheared chromatin was diluted (1:10) with ChIP buffer and was pre-cleared with a 50% suspension of salmon sperm DNA-saturated protein A/G agarose beads (Millipore) for 1 h at 4°C on a rotating platform. The agarose beads were pelleted by centrifugation and the supernatant immunoprecipitated using primary antibodies to: CREB (Cell Signaling), Tet1 (ThermoFisher), BHLHB2 (DEC1; Santa Cruz Biotechnology), MeCP2 (Santa Cruz), ERK1/2 (Cell Signaling), H3K4me3 and H3K27me3 (Cell Signaling), RNAPIISer5 (Abcam) and RNAPIISer2 (Millipore), or normal rabbit IgG (Santa Cruz) as a control, at 4°C overnight with rotation. The antibodies were confirmed for their specificity by Western blots of turtle and rat brain tissue (Fig. 9). Antibodies for CREB,¹⁸ and ERK1/2¹⁷ were tested and are similar to those shown previously. After overnight incubation, immune complexes were collected by incubating with 60 µl of 50% suspension of salmon sperm DNA-saturated protein A/G agarose beads for one hour at 4°C with rotation. Agarose beads were pelleted by centrifugation and the immune complexes were washed once with low salt buffer, high-salt buffer, lithium chloride wash buffer, and twice with 1X TE buffer. Immunoprecipitated chromatin was eluted twice, each time with 250 μl of freshly prepared elution buffer (1% SDS, 0.1 M NaHCO₃) and pooled. 5 M NaCl was added to the elutes along with input controls and protein-DNA crosslinks were reversed by incubating the sample at 65°C overnight. The samples were further incubated with proteinase K (10 mg/ml) for 1 h at 50°C. DNA was isolated using phenol/chloroform/isoamyl alcohol followed by precipitation with ethanol, dried, and dissolved in water. ChIP and input samples were amplified with tBDNF primers specific for transcription factor binding sites of each promoter using quantitative real-time PCR (qPCR). See Table 2 for a list of tBDNF primers used for ChIP-qPCR. The qPCR was carried out by using SYBR green PCR master mix (Applied Biosystems). The fluorescence intensity of each amplicon was measured by the Step-One Plus real-time PCR system (Applied Biosystems). qPCR conditions were 95°C for 10 min followed by 40 cycles at 95°C for 15 sec, and 60°C for 1 min. ChIP data were normalized to input DNA from each sample and the fold differences were calculated using the 2^(-ΔΔCt) values.^32,33 All reactions were performed in duplicates.

Figure 9.

Specificity of the antibodies used in the present study. Antibodies were tested on naïve brain tissue from turtle (T) and rat (R). Bands appeared at the appropriate molecular weights in both species.

Co-immunoprecipitation and Western blot

Frozen brainstem samples were homogenized in lysis buffer with a protease and phosphatase inhibitor cocktail. Protein samples were precleared with protein A/G-agarose and supernatants were incubated in the primary antibodies or nonspecific rabbit or mouse IgG as a control at 4°C for 2 h. Protein A/G-agarose was added to the protein samples and incubated at 4°C overnight. Immunoprecipitated samples or IgG control samples were washed with ice-cold lysis buffer and dissociated by heating for 5 min in the loading buffer and then subjected to SDS-PAGE. For all Western blots and co-immunoprecipitation experiments both input protein and IgG controls were loaded at the same time. The same primary antibodies used for ChIP assays were also used for co-immunoprecipitation and/or Western blotting. Additional antibodies used for Westerns were Tet2 and Tet3 (Santa Cruz). Proteins were detected by the ECL Plus chemiluminescence system (Amersham) or the Odyssey infrared imaging system (Li-Cor Biosciences) and quantified by computer-assisted densitometry.

Tet1 siRNA design and application

A Tet1 siRNA was designed based on conserved regions of human Tet1 mRNA (NM_030625.2) and the predicted sequence from the pond turtle Chrysemys picta bellii (XM_005297302.1). The sense sequence from turtle was 5'-TGGAACTGTGACAGATAATGAACAC-3' and corresponds to human Tet1 2685 – 2709 nt present in exon 4. Exon 4 is exclusive to Tet1 only. The duplex siRNA was synthesized and purified (Integrated DNA Technologies) and used at a final concentration of 150 nM mixed with Lipofectamine RNAiMax reagent (Invitrogen) in physiological saline and bath applied to preparations for 24 h. A Silencer Negative Control no. 1 siRNA was also used as a control (Ambion). After the elapsed time, preparations underwent the conditioning procedure and were processed for further analysis.

Statistical analysis

Data were analyzed with StatView software using a one-way ANOVA followed by a Fisher's post-hoc test. Values are presented as means ± SEM. N's represent the number of brainstem preparations. P values are determined relative to the naïve group except where noted. Significant differences were considered to be P < 0.05.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

Supported by the National Institutes of Health grant NS051187 (JK).