|Home | About | Journals | Submit | Contact Us | Français|
The consolidation of conditioned fear involves upregulation of genes necessary for long-term memory formation. An important question remains as to whether this results in part from epigenetic regulation and chromatin modulation. We examined whether homer1a, which is required for memory formation, is necessary for Pavlovian cued fear conditioning, whether it is downstream of BDNF - TrkB activation, and whether this pathway utilizes histone modifications for activity-dependent transcriptional regulation. We initially found that Homer1a ko mice exhibited deficits in cued fear conditioning (5 tone-shock presentations with 70 dB, 6kHz tones and 0.5s, 0.6mA footshocks). We then demonstrate that homer1a mRNA 1) increases after fear conditioning in vivo within both amygdala and hippocampus of wild type mice, 2) increases after BDNF application to primary hippocampal and amygdala cultures in vitro, and 3) these increases are dependent on transcription and MAPK signaling. Furthermore, using chromatin immunoprecipitation we found that both in vitro and in vivo manipulations result in decreases in homer1 promoter H3K9 methylation in amygdala cells but increases in homer1 promoter H3 acetylation in hippocampal cells. However no changes were observed in H4 acetylation or H3K27 dimethylation. Inhibition of H3 acetylation by sodium butyrate enhanced contextual but not cued fear conditioning and enhanced homer1 H3 acetylation in the hippocampus. These data provide evidence for dynamic epigenetic regulation of homer1a following BDNF-induced plasticity and during a BDNF-dependent learning process. Furthermore, upregulation of this gene may be regulated through distinct epigenetic modifications in the hippocampus and amygdala.
Understanding the molecular mechanisms of fear is crucial for progress in fear-related disorders such as posttraumatic stress disorder. Many genes are dynamically regulated during the consolidation of conditioned fear, including immediate-early genes (Lonergan et al., 2010; Hall et al., 2001; Radwanska et al., 2002; Strekalova et al., 2003), kinases (Ploski et al., 2010), neurotrophic factors (Ressler et al., 2002; Ou and Gean, 2007, Von Hertzen and Gies, 2005), and synaptic structural proteins(Lonergan et al., 2010). Gene expression is dynamically regulated by modification of chromatin structure, which regulates accessibility of transcriptional machinery to promoter regions (Day and Sweatt, 2010; Renthal and Nestler, 2008; Barrett and Wood, 2008). Modifications include acetylation, methylation and phosphorylation of histone tails, which can in turn affect the chromatin structure and promoter accessibility (Krishnan et al., 2011). Histone modifications occur after contextual fear conditioningin the hippocampus (Chwang et al. 2006), and drugs that increase histone acetylation such as histone deacetylase (HDAC) inhibitors enhance long term memory for contextual fear (Levenson et al., 2004). This suggests that epigenetic mechanisms are linked extensively to associative learning and gene transcription (Sweatt, 2004, 2009; Peleg, 2010; Miller et al., 2008; Lattal et al., 2007; Chwang et al., 2006).
Here we examine epigenetic regulation of the gene variant, homer1a (also known as vesl-1S). Homer1a belongs to a family of scaffolding proteins that localize at the postsynaptic density (PSD) (Foa and Gasperini, 2009; Shirashi-Yamaguchi and Furuich, 2007), which play a role in intracellular calcium homeostasis, receptor trafficking, gene transcription, and signal transduction (Foa and Gasperini, 2009; Shirashi-Yamaguchi and Furuich, 2007). The long forms of Homer (homer 1b-f) are constitutively expressed and contain of two functional domains: the EVH1 binding domain which targets Shank, mGluR1/5 and ryanodine receptors among others, and a coiled-coil structure that aggregates with other Homer molecules and PSD proteins (Shirashi-Yamaguchi and Furuich, 2007).
The shorter variant of the Homer1 family (homer1a) lacks the coiled-coil domain and is expressed in an activity dependent manner (Shirashi-Yamaguchi and Furuich, 2007) (Figure 1a). This suggests that homer1a might act to disrupt homer-protein clusters by competitively binding to target proteins (Shirashi-Yamaguchi and Furuich, 2007). Furthermore it has been shown that homer1a can disrupt the physical link between mGluR and NMDA, altering mGluR1/5’s ability to modulate NMDA functioning (Bertaso et al., 2010).
Most of the current behavioral epigenetic literature focuses on global histone modifications. In contrast, here we examine mechanisms of transcription during fear conditioning and BDNF-induced plasticity of a specific gene, homer1a, which may regulate synaptic plasticity during memory consolidation. The BDNF-TrkB pathway has been heavily implicated in amygdala and hippocampus during fear conditioning, with BDNF mRNA and protein upregulation during memory consolidation (Rattiner et al., 2004a). Moreover inhibiting BDNF signaling impairs consolidation of conditioned fear (Rattiner et al., 2004b). Thus examining BDNF induced plasticity in vivo as well as in primary cell culture may provide mechanistic insight into regulation of homer1a in Pavlovian fear conditioning.
All experiments were performed on adult (6–10 weeks old) wild-type strain C57BL/6J male mice from Jackson Laboratory (Bar Harbor, ME.). All procedures were approved by the Institutional Animal Care and Use Committee of Emory University and were in compliance with National Institutes of Health guidelines. Separate cohorts of animals were used for each experiment. The Erk inducible knockout mice contain loxP sites flanking exon 2 of the Erk2 gene (Samuels et al., 2008). These mice have an intact Erk1 gene, but deletion of the Erk2 exon 2 prevents translation and protein production of the Erk2 protein product in cells expressing Cre recombinase (Samuels et al., 2008).
Homer1a KO breeder mice (Hu et al., 2010) were a gift from Paul Worley. Hu and colleagues generated the Homer1a targeting construct by fusing 2.7 kb of genomic DNA, including intron 4 and exon 5 of the Homer1 gene, with part of the rat Homer1c cDNA (2 kb), containing exons 6–10. The resulting targeted recombinant mouse results in normal levels of Homer1b/c, 2, and 3 protein expression but not Homer1a (Hu et al., 2010). We crossed heterozygous Homer1a KO mice to generate wildtype and KO littermates (as seen in Figure 2a). 6–10 week old male and female Homer1a wildtype and KO littermates were fear conditioned and tested as below.
Fear conditioning was conducted in nonrestrictive acrylic cylinders (SR-LAB startle response system, San Diego Instruments) located in a ventilated, sound-attenuated chamber. The foot shock (unconditioned stimulus) was delivered through a stainless steel grid floor. Shock reactivity was defined as the peak activity (measured with a piezoelectric accelerometer) that occurred during the 200 msec after the onset of the unconditioned stimulus. The tone-conditioned stimulus was generated by a Tektronix function generator audio oscillator and delivered through a high frequency speaker. One day prior to training, mice were preexposed to the tone through a 5 ‘tone-alone’ presentation program to both habituate them to handling and to the tone, but to get baseline fear responses to the tone presentation. Preexposure was done in a separate context. During cued fear conditioning, mice received five trials of a conditioned stimulus tone (30 seconds, 6 kHz, 70 dB) coterminating with an unconditioned stimulus foot shock (500 msec, 0.6 – 1 mA) with a variable inter-trial-interval between 60 and 180 seconds. The expression of fear was assessed 24 hours after fear conditioning and consisted of 3 minutes in the same context for which the training occurred (contextual memory) with no stimulus and 5 conditioned stimulus tone presentations of 30 second each with a 1.5-minute inter-trial interval in a different context (see figure 1 for schematic). Stimulus presentation and data acquisition were controlled and digitized by, and stored in, an interfacing desktop computer using SR-LAB and analyzed with the Freeze View software program (Coulbourn Instruments, Whitehall, Pa.).
Primary cultures of postnatal hippocampal neurons were described previously (Brewer, 1997) with modifications, C57BL/6J mice (21 days postnatal) were decapitated and the hippocampus and amygdala were removed and immersed in ice-cold dissection buffer consisting of Hibernate-A medium (BrainBits, Springfield, IL, USA), B27 supplement and gentamycin (Invitrogen) (12g/ml) for the preparation of separate hippocampal and amygdala neuronal cell cultures. The hippocampus and amygdala tissues were sliced and then enzymatically digested with papain (Worthington, Lakewood, NJ, USA) in Hibernate-A medium at 32°C for 30 minutes. Cells were dissociated by triturating with Pasteur pipets fired on the tips to narrow openings. Neurons were purified in a density gradient media including Hibernate-A and OptiPrep (Sigma, St. Louis, MO, USA) by centrifugation. The density gradient media consisted of four layers. The first was 1 mL dissection buffer containing 35% OptiPrep; the second 1 mL dissection buffer contained 25% Optiprep and the third 1 ml dissection buffer contained 20% OptiPrep and the fourth 1 mL dissection buffer contained 15% OptiPrep. They were added on the top of each other carefully, resulting in clear layer separation. Then cells were added on the top of the density gradient media. After centrifugation, the densest layer with a cream color, located at the middle of the tube, could be seen. This layer of neurons was taken out by using a sterile transfer pipette and put into a new tube. After washing with dissection buffer, neuronal cells were plated onto Poly-D-Lysine (Sigma) coated plates or glass coverslips at the density of 2.5 x 105 cells/cm2 in culture media consisting of Neurobasal A medium (Invitrogen) with 2% B27 supplement, 2 mM glutamax and gentamycin (5 g/ml). Thereafter, the cultures were kept in a humidified incubator at 37°C and 5% CO2 and media were changed every 5 days until used for experiments. After 2–3 weeks in vitro, the cells were used for the experiments reported in the present study.
Recombinant human BDNF was purchased from Cell Sciences (Canton, MA, USA) and reconstituted in sterile PBS as 100 mg/ml stock. The aliquots of stock were stored at −30°C and final drugs and concentrations for cell culture experiments were as following: BDNF (100 ng/ml), 7, 8-DHF (500 nM), U0126 (Tocris biosciences Ellisvile, MO, USA, 10 uM), ActD (25 uM Sigma, St Louis, MO, USA). Mice received intraperatoneal injections of 1.2 g/kg sodium butyrate (NaB, Sigma Aldrich, B5887) dissolved in distilled water or an equal volume of distilled water alone (vehicle). This dose has been shown previously to enhance contextual fear memories (Levenson et al., 2004). The injections occurred immediately after training. For in vitro studies a concentration of 10 uM dissolved in distilled water was used.
Tissue samples were treated using an EpiQuik tissue ChIP kit (Epigentek Group Inc. Brooklyn, NY). Brains were extracted using rapid decapitation 2 hours after training. Amygdala and hippocampal tissue was rapidly dissected under a dissecting scope with a scalpel in ice-cold PBS and then frozen immediately on dry ice and stored at −80°C until ready to use. Cells/tissues were harvested and mixed with formaldehyde at a final concentration of 1.0% for 10 min at 37°C to cross-link protein to DNA. Cells/tissue then were suspended in 0.2 mL of SDS lysis buffer and allowed to settle on ice for 10 minutes. DNA cross-linked with protein was then sonicated into fragments of 200–1000 bp. One-tenth of the sample was set aside as an input control, and the rest was then immunoprecipitated 1.5 h at room temperature with 5 g of primary antibody in the ChIP kit strip wells. As a control samples were immunoprecipitated with 5g nonimmune rabbit IgG. After immunoprecipitaiton, the DNA-protein complex was eluted and the proteins were digested with DNA release buffer and proteinase K. DNA was dissociated at 65°C for 1.5 hours under reverse buffer. The DNA, associated with antibody of interest (pan-H3-acetylated, pan-H4 acetylated, H3K9 dimethylation, and H3K27 dimethylation; antibodies were obtained from Epigentek) was extracted with binding buffer, precipitated with 70% and 90% ethanol and finally elutes DNA by elution buffer. Quantitative real-time PCR was performed with primers specific to the Homer1 promoter and for the GAPDH promoter regions.
Total RNA was prepared from frozen amygdala and hippocampal dissections in mice. Brains were extracted using rapid decapitation 2 hours after training. Amygdala and hippocampal tissue was rapidly dissected in ice-cold PBS and then frozen immediately on dry ice and stored at −80 C until ready to use. Briefly, tissue samples were homogenized and centrifuged at 13,000g for 3 minutes. RNA was washed with 70% ETOH and purified using RNeasy columns (Qiagen). RNA amount and quality were determined using a nanodrop spectrophotometer.
140 micrograms of total RNA were reverse transcribed using the RT2-First Strand Kit (C-03, SA Biosciences). Quantitative PCR was performed using the Applied Biosystems 7500 Fast. Online detection of reaction products was carried out using the SybrGreen Gene Assay with custome made primers for homer1a, homer1c and GAPDH. SybrGreen mastermix was obtained from SA biosciences, and manufacturer’s instructions were followed. Calculated values are presented as mean +/− SEM to indicate accuracy of measurement. Homer1a and Homer1c values were normalized for measurements of GADPH. PCR conditions were 2 minutes at 50°C, 10 minutes at 95°C and 40 cycles with 15s 95°C, 60 s 60°C.
Primers were designed and confirmed by Primer blast. There sequence is as follows: Homer1a – FWD – 5′-GAAGTCGCAGGAGAAGATG-3′; Homer1a – REV – 5′-TGATTGCTGAATTGAATGTGTACC-3′; Homer 1c – FWD – 5′-ACACCCGATGTGACACAGAACT-3; Homer 1c – REV - 5′-TCAACCTCCCAGTGGTTGCT-3′; Homer1 Promoter FWD – 5′-GGTGACGTATGTGCGGAGAGGA-3′; Homer1 Promoter – REV – 5′-GGTCCGTCGGTCCGTCCCTTT-3′; Primers for GAPDH and GAPHD promoter region were obtained from SA Biosciences.
Statistically significant differences were determined by Student’s t-test or by between subjects two-way ANOVA. The resulsts were presented as means +/− s.e.m. For all chromatin immunoprecipitation and mRNA data, fold chages relative to control were determined using the ΔΔCt method; a mean fold change value along with an s.e.m. value were determined; the ΔΔCt values from each data set were used in two-tailed paired t-tests (which were adjusted for multiple comparisons) to determine statistical significance (*P < 0.05). All values included in the figure legends represent mean +/−s.e.m. The RTPCR ChIP data was analyzed identically to the mRNA data using the ΔΔCt method, except that ChIP data were normalized to ‘input’ rather than GAPHD.
Heterozygous Homer1a KO mice were crossed to generate wildtype and KO littermates and genotyped as previously described (Hu et al., 2010; Figure 2a). 6–10 week old male and female Homer1a wildtype and KO littermates were then fear conditioned and tested to determine if Homer1a was required for normal fear learning. When we examined shock reactivity, we found that there was no difference in activity response to a 0.6mA footshock, suggesting that both mice of both genetic backgrounds have similar pain sensitivity and unconditioned footshock responses. Following a single session of 5 trials of 0.6mA footshock paired with 6kHz tone, with a pre-training saline injection stress, mice were tested for fear expression to the tone 24 and 48 hours later. During fear testing, we found a significant decrease (F(1,16)=4.96, P<.05, N=9/group) in the expression of cued fear across both testing sessions (Figure 2b). These data suggest that Homer1a expression, even in the presence of normal Homer1b,c, 2, and 3, is required for normal cued fear learning.
Homer1a was dynamically regulated during consolidation of Pavlovian fear conditioning. Homer1a contains a unique stop site at the end of exon 5 that makes its sequence unique to the longer gene variants of homer such as homer1c. Primers for RTPCR were designed based on this sequence to differentiate between homer1a and homer1c expression (Figure 1b). For all of the experiments described amygdala and hippocampal tissue was extracted during consolidation of fear, 2 hours after training (Figure 3a). As illustrated in figure 3a, all animals were measured for baseline freezing to presentation of a tone in one context (Context A). One day later, animals were presented with 5 tone-shock pairings (30 sec 6kHz tone co-terminating with 0.5sec, 1mA shock) or 5 tones without any shock in a novel context (Context B). On the third day, animals were tested in Context B without any tones or shocks for 3 minutes as an assessment of contextual fear conditioning and then immediately placed into context A where freezing in response to 5 tone-alone trials is assessed. Using this procedure, we found that one training session achieved both retention of contextual and cued fear conditioning (Figures 3b and c) (b: T (18) = 4.20; p < 0.05, n = 10/group; c: T(18)= 2.69, p < 0.05, n = 10). In a separate cohort of animals quantitative PCR demonstrated an increase in homer1a mRNA in the hippocampus (Figures 4a and b) (T(18) = 3.35, p < 0.05, n = 10) and in the amygdala (Figures 4c and d) (T(18) = 2.39, p < 0.05, n = 10) 2 hours after fear conditioning. RNA for homer1c (a longer gene variant of the Homer1 gene family) was not increased during Pavlovian fear conditioning in either brain region (Figures 4a and c) (hippocampus; amygdala). No changes in homer1a mRNA levels were seen in the striatum.
To assess differential regulation of Homer1a through TrkB signaling, primary hippocampal and amygdala cell cultures were used. Much like in fear conditioning, BDNF-induced plasticity increased homer1a mRNA levels in both hippocampal and amygdala cell culture (figure 5a and d) (hippocampus T (10) = 3.25, p < 0.05, n = 6/group; amygdala T(10) = 2.67, p < 0.05, n = 6) but not homer1c levels. The trkB-specific agonist (7, 8-DHF) upregulated homer1a in cell culture (hippocampus T(10) = 2.25, p < 0.05, n = 6/group, amygdala T(10)= 4.86, p < 0.05, n = 6). Figures 5b and e are representative pictures of immunostainting for CamKII in both hippocampal (b) and amygdala (e) primary neuronal cultures. Figure 5c and f are representative pictures of immunostaining for parvalbumin in both hippocampal (c) and amgydala (f) primary neuronal cell cultures.
Blocking transcription with Actinomysin D (ActD) inhibited BDNF-induced upregulation of homer1a in both hippocampus and amygdala cell cultures (Fig. 6a and d) (a: F(3,20) = 27.34, p < 0.05, n = 6/group; d: F(3,20)=258.90, p < 0.05, n = 6/group). In addition MEK inhibition by U0126 blocked BDNF induced increases in homer1a in both hippocampus and amygdala cells (Fig. 6b and e) (b: F(3,20)= 7.039, p < 0.05, n = 6/group, e: F(3,20) = 14.57, p < 0.05, n = 6/group). We next utilized primary cell culture from floxed-ERK knockout mice, in which we transfected cells with a Cre Recombinase expressing lentivirus to delete the ERK gene. We found that genetically deleting ERK impaired BDNF induced upregulation of homer1a in both hippocampus and amygdala cells as well (Fig. 6c and f) (c: F(3,20) = 23.42, p < 0.01, n = 6/group, f: F(3,20) = 89.61, p < 0.01, n = 6/group). Thus, BDNF appears to upregulate homer1a in a transcriptionally dependent manner, and through MEK and ERK signaling mechanisms. None of these manipulations had any effect on homer1c mRNA levels (table 1). Genetic deletion of ERK was demonstrated through QT-PCR in amygdala (T(10) = 2.28, p < 0.05, n = 6/group) and hippocampal cells (T(10) = 2.37; p < 0.05, n = 6).
In order to determine the epigenetic role of BDNF signaling on homer1a expression we examined histone modifications around the homer1a promoter region after BDNF induced plasticity. Chromatin immunoprecipitation (ChIP) assays were performed to measure the levels of several histone modifications around the Homer1 promoter after BDNF-induced plasticity. Levels of promoter enrichment were quantified by QT-PCR. We found that BDNF application had distinct effects in hippocampal and amygdala primary cell culture. In the hippocampal cell cultures, there was a significant increase in H3 acetylation (T(10) = 6.80, p < 0.05, n = 6/group) following BDNF application, but no changes were apparent in H4 acetylation, H3K9 methylation or H3K27 methylation (Figure 7a). In amygdala cell cultures, however, there appears to be a decrease in H3K9 methylation (T(10) = 2.44, p < 0.05, n = 6/group) following BDNF application, but no changes in H3 acetylation, H4 acetylation of H3K27 methylation (Figure 7c). Significant changes in acetylation or methylation were not detected at the GAPDH promoter region.
We next examined the effect of Pavlovian fear conditioning on histone modifications around the Homer1 promoter region. In the hippocampus there was a significant increase in H3 acetylation (T(18) = 2.37, p < 0.05, n = 10/group), but no difference in H4 acetylation, H3K9 methylation or H3K27 methylation (Figure 7b). In the amygdala, however, there was a significant decrease in H3K9 methylation (T(18) = 3.14, p < 0.05, n = 10/group), but no changes in H3 acetylation, H4 acetylation, or H3K27 methylation (Figure 7d). Significant changes in acetylation or methylation were not detected at the GAPDH promoter region. Notably, these in vivo results parallel the histone modification-specific findings seen in amygdala and hippocampal primary cell culture.
HDAC inhibitors have been shown to enhance contextual fear conditioning(Levenson et al., 2004). In this experiment, we examined the effect of the HDAC inhibitor, sodium butyrate (NaB) on fear conditioning. Note that the overall level of fear retrieval in controls (Figure 8a) was less than in our initial study (Figure 3c), which is likely due to injection stress. We showed that IP administration of NaB can induce increases in contextual fear memories (T(18) = 2.10, p < 0.05, n = 10/group) but did not appear to cause an increase in cued fear conditioning (Figure 8a). NaB also appeared to enhance hippocampal homer1a mRNA expression (Figure 8b; F(3,36) = 5.01, p < 0.05, n = 10/group), but seemed to reverse the mRNA increase in amygdala tissue (Figure 8c, F(3,36) = 5.45, p < 0.05, n = 10/group). Sodium butyrate enhanced H3 acetylation in hippocampal tissue (Figure 8d, F(3,36) = 9.54, p < 0.01, n = 10/group) but reversed fear conditioned induced decreases in H3K9 methylation in amygdala tissue (Figure 8e, F(3,36) = 4.58, p < 0.05, n = 10/group).
In this paper we demonstrate that the homer1a knockout mice have a deficit in cued fear conditioning, that homer1a mRNA is increased during the consolidation of Pavlovian cued fear conditioning in the hippocampus and amygdala, as well as during BDNF-induced plasticity in amygdala and hippocampal primary cell culture. As mentioned previously, BDNF has been shown to play a critical role in fear conditioning in both the hippocampus and amygdala. We chose to use a primary cell culture model as a way to more directly address the molecular mechanisms of homer1a gene transcription and potentially tease apart any differences in hippocampal and amgydala neurons separate from their functional connectivity seen in the brain. We fail to rule out the possibility that other receptor-ligand systems play a role in homer1a transcription. There are several potential signaling pathways that could result in upregulation of homer1a during Pavlovian fear conditioning, including NMDA activation (Ango et al., 2000; Sato et al., 2001). However, we have demonstrated that BDNF induced plasticity in cell culture is one mechanism for Homer1a upregulation. Given the role of BDNF in the amygdala and in the hippocampus during consolidation of fear conditioning (Rattiner et al., 2004b; Rattiner et al., 2004a; Heldt et al., 2007; Musumeci et al., 2009), it is plausible that an increase in BDNF in the hippocampus and amygdala might result in an increase in homer1a signaling during the consolidation of Pavlovian fear conditioning.
We also demonstrate that Pavlovian fear conditioning results in an increase in H3 acetylation around the Homer1 promoter in the hippocampus, and a decrease in H3K9 methylation around the Homer1 promoter in the amygdala. However, we saw no differences in H4 acetylation or H3K27 methylation. Given that there is no existing evidence for histone posttranslational modifications regulating homer1a and very limited data for specific modifications regulating the expression of other genes during Pavlovian fear conditioning (Fuchikami et al., 2010; Gupta et al., 2010) we began this work using a representative sample of well characterized histone-specific antibodies. There is evidence for H3K9 global methylation and H3 and H4 global acetylation during Pavlovian fear conditioning (Gupta et al., 2010). In addition there are data demonstrating an increase in H3 acetylation specifically around the promoter region of BDNF (Fuchikami et al., 2010; Gupta et al., 2010; Takei et al., 2011) though no study to date has examined histone modifications around the Homer1 promoter or any other genes during Pavlovian fear conditioning. For future studies, it may be informative to look at a wider array of posttranslational modifications involved in homer1a during Pavlovian fear conditioning. However, with our very limited selection, we were able to find distinct epigenetic regulation of homer1a in hippocampal and amygdala tissue during Pavlovian fear conditioning and cells during BDNF induced plasticity.
Interestingly, during BDNF induced plasticity, we also demonstrate an increase in H3 acetylation around the Homer1 promoter in hippocampal primary cell culture and a decrease in H3K9 methylation in amygdala primary cell culture. Despite the fact that homer1a mRNA was upregulated in both the hippocampus and amygdala, histone modifications around the Homer1 promoter were distinct between the two brain regions. The hippocampus primarily exhibited increases in histone H3 acetylation, which is associated with enhanced gene transcription, within the Homer1 promoter region. The amygdala, however, primarily exhibited decreases in histone H3K9 methylation, a repressive marker of transcription, in the Homer1 promoter region. Notably, although the specific histone regulation was different in these two brain regions, and in primary cell culture from these regions, both histone tail modifications would result in enhanced transcription. These differences were seen in vivo after fear conditioning and in vitro after BDNF – induced plasticity suggesting that these differences are not due to the unique functional connectivity of the hippocampus and amygdala but due to intrinsic molecular properties of the neurons themselves.
It is important to note that we have not directly demonstrated, in vivo, the connection between BDNF-dependent ERK activation and epigenetic regulation of the Homer1 promoter. Rather our data combines convergent studies demonstrating correlational relationships in vivo, and correlational, as we as some causal relationships in vitro. Future studies should inhibit BDNF at the time of learning, perhaps via BDNF knockdown approaches (e.g. Choi et al., 2010) or through blocking TrkB activation with genetic approaches (e.g. Chhatwal et al., 2006).
Moreover, histone deacetylase (HDAC) inhibition with sodium butyrate (NaB) had differential effects on hippocampal and amygdala tissues. As expected, NaB enhanced hippocampal H3 acetylation around the Homer1 promoter in fear-conditioned mice but not control mice. NaB did not induce global increases in acetylation in that the GAPDH promoter showed no enhancement in H3 or H4 acetylation, nor did NaB induce an increase in histone H4 acetylation around the Homer1 promoter in the hippocampus. This result is straightforward in that NaB prevents the removal of acetyl groups from histone tails. In order for there to be a resulting increase in acetylation there would already have to be an initial addition of acetyl groups, which only would occur around already activated genes. If histone H3 but not H4 is increasingly acetylated around the Homer1 promoter, then NaB should only enhance H3 acetylation. This may also explain why NaB did not enhance acetylation of H3/H4 tails around the Homer1 promoter in amygdala tissues. NaB did however reduce H3K9 methylation around the Homer1 promoter in control mice and reverse the decrease in H3K9 methylation seen in fear conditioned mice. While it has been demonstrated that HDAC inhibition can decrease H3K9 methylation (Gupta et al., 2010), the mechanism is less clear and the mechanism underlying NaB effects on H3K9 methylation in the amygdala is not so obvious.
Understanding the molecular mechanisms of Pavlovian fear conditioning will give us an understanding of the mechanism and potential treatments for PTSD and other fear-related disorders. Histone modifying drugs have been shown to enhance fear conditioning as well as fear extinction (Kaplan and Moore 2011). Our data suggest a distinct epigenetic signature for Homer1a gene expression in hippocampal and amygdala cells/tissues separate from the two regions’ functional connectivity. These differences may explain why HDAC inhibitors such as sodium butyrate only enhance hippocampal dependent contextual memories and mainly enhance hippocampal histone modifications.
At first glance, an increase in Homer1a may seem contradictory relative to other findings with LTP and plasticity in fear conditioning. For example, it has been shown that Homer1a is upregulated during increases in network activity which scales down the expression of synaptic AMPA receptors(Hu et al., 2010; Inoue et al., 2007; Sala et al., 2003). On the other hand, fear conditioning results in LTP of synaptic transmission from auditory thalamus and cortex to the LA and increased synaptic GluR1 subunit of AMPA receptors (Clugnet and LeDoux, 1990; Rogan et al., 1997; McKernan and Shinnick-Gallagher, 1997). One possible explanation is that Homer1a increases may be occurring only transiently following fear conditioning. We have recently found that β-catenin (Maguschak and Ressler, 2008, 2011a), possibly through Wnt-mediated signaling(Maguschak and Ressler, 2011b), is associated with transient decreases followed by increases in synaptic stability during fear consolidation. Similarly, Homer1a increases may serve to transiently destabilize synapses, thus allowing for strengthening of the most currently plastic/active synapses as new learning occurs. Further studies should help to clarify its functional role and how this function is differentially modulated via transcriptional regulation.
While we know that fear learning requires long term potentiation and synaptic plasticity, very little direct evidence links the increases in gene transcription with increases in LTP seen in fear conditioning. Homer1a, a know regulator of both Pavlovian fear conditioning and physiological plasticity, is epigenetically regulated after learning, and may provide a useful connection between gene expression and physiological synaptic plasticity. Furthermore, studying the role of homer1a transcription and activity in the consolidation of fear may help better understand the connection between epigenetics, gene transcription, LTP and the consolidation of memory formation. In addition, it may provide insight into future drug targets for the enhancement of extinction learning as well as behavioral therapy for PTSD and other fear-related disorders.
Support was provided by NIH (DA019624, P30 NS055077), the Burroughs Wellcome Fund, and Center for Behavioral Neuroscience STC Center: NSF Agmt #IBN-9876754, National Primate Research Center base grant #RR-00165, Animal Resource Program at NIH.
Author ContributionsThe study was conceived by. A.L.M and K.J.R.. The experiments were designed by A.L.M., L.M., and K.J.R., and carried out by A.L.M, L.M., and N.S. J.H.H. and P.W. generated the Homer1a knockout mice. The manuscript was written by A.L.M. and K.J.R.