Long-term effects of early-life caregiving experiences on brain-derived neurotrophic factor histone acetylation in the adult rat mPFC

Abstract

Infant–caregiver experiences are major contributing factors to neural and behavioral development. Research indicates that epigenetic mechanisms provide a way in which infant–caregiver experiences affect gene activity and other downstream processes in the brain that influence behavioral development. Our laboratory previously demonstrated in a rodent model that exposure to maltreatment alters methylation of DNA associated with the brain-derived neurotrophic factor (bdnf) and reelin genes as well as mRNA of key epigenetic regulatory genes in the medial prefrontal cortex (mPFC). In the current study, we characterized patterns of histone acetylation at bdnf and reelin gene loci after our caregiver manipulations. Using a within-litter design (n = 8–10/group from eight litters), pups were exposed to adverse (maltreatment condition: exposure to a stressed caregiver) or nurturing (cross-foster condition: exposure to a nurturing caregiver) caregiving environments outside the home cage for 30 min daily during the first postnatal week. Remaining pups in a litter were left with the biological mother during each session (providing normal care controls). We then used chromatin immunoprecipitation (ChIP) and quantitative RT-PCR to measure histone 3 lysine 9/14 acetylation associated with bdnf promoters I and IV and the reelin promoter in the adult mPFC. Maltreated females had decreased acetylation at bdnf IV, while neither males nor females exhibited histone acetylation alterations at bdnf I or reelin. These data demonstrate the ability of maltreatment to have long-term consequences on histone acetylation in the mPFC, and provide further evidence of the epigenetic susceptibility of bdnf IV to the quality of infant–caregiver experiences.

• Development
• epigenetic
• H3K9
• maltreatment
• maternal
• reelin

Introduction

Early-life stress in the form of caregiver maltreatment is a known factor in the development of psychiatric disorders (Bremner et al., 2007; Cicchetti & Toth, 2005; Heim & Nemeroff, 2001; Neigh et al., 2009). Adults who experienced childhood abuse show increased instances of depression, anxiety and post-traumatic stress disorder (PTSD) (Anda et al., 2006; Cicchetti & Toth, 2005), and rodents and non-human primates that have negative experiences with a caregiver in infancy show neurobiological and behavioral outcomes similar to those observed in humans (Ivy et al., 2008, 2010; Kaffman & Meaney, 2007; Liu et al., 1997; McGowan et al., 2009). The mechanisms that produce the neurobiological and behavioral outcomes associated with maltreatment are likely varied and complex, but alterations in gene regulation and expression appear to play a crucial role (Liu et al., 1997; Meaney, 2001; Roceri et al., 2004). One way in which gene expression (either basal levels or activity-dependent) can be altered by environmental factors is epigenetic mechanisms, including (but not limited to) DNA methylation and histone acetylation (Blaze & Roth, 2012). While DNA methylation is known to generally repress gene transcription, histone acetylation is an activator of transcription, leading to increased mRNA expression.

There is increasing evidence that early-life caregiver experiences can alter central nervous system (CNS) DNA methylation in a genome-wide (Khulan et al., 2014; Labonte et al., 2012; Naumova et al., 2012; Provençal et al., 2012) and gene-specific (Blaze et al., 2013; Champagne et al., 2006; Kundakovic et al., 2013; McClelland et al., 2011; Murgatroyd et al., 2009; Roth et al., 2009, 2014; Weaver et al., 2004) manner. Several brain regions (i.e. hippocampus, cortex, amygdala, hypothalamus) are known to exhibit alterations in methylation of DNA associated with gene loci involved in stress, plasticity and emotion, including the brain-derived neurotrophic factor (bdnf) (Blaze et al., 2013; Kundakovic et al., 2013; Roth et al., 2009, 2014), glucocorticoid receptor (GR) (Kundakovic et al., 2013; Weaver et al., 2004), arginine vasopressin (Murgatroyd et al., 2009; Murgatroyd & Spengler, 2014) and corticotropin releasing factor (Franklin et al., 2010; McClelland et al., 2011) genes. For example, our laboratory has shown that exposure to caregiver maltreatment during the first postnatal week results in DNA methylation alterations (at bdnf I, bdnf IV and reelin regulatory regions) that are present in the adult rat medial prefrontal cortex (mPFC) (Blaze et al., 2013). Notably, some of these methylation patterns differ from those in the whole PFC (Roth et al., 2009), suggesting that subregions of the PFC respond differentially to early-life stress. Human studies have also highlighted this subregion specificity, showing that changes in mPFC structure and activation after early-life stress play a role in sensitivity to stress later in life and changes in cognitive functioning and anxiety (Gorka et al., 2014; Hanson et al., 2012; van Harmelen et al., 2014).

As DNA methylation has been the main focus in early-life stress research, there are few reports of the effects of early-life stress on histone acetylation. Groups that have looked at early stress-induced modifications to histones have used a global approach or specifically targeted histones associated with the GR gene (McGowan et al., 2011; Levine et al., 2012; Weaver et al., 2004). Furthermore, no studies (to date) have investigated histone acetylation at bdnf or reelin regulatory regions after exposure to maltreatment in infancy. Because bdnf and reelin code for proteins important in neural development and plasticity, these genes are widely implicated in various psychiatric disorders (Abdolmaleky et al., 2005; Boulle et al., 2012; Martinowich et al., 2007; Nagahara & Tuszynski, 2011; Tissir & Goffinet, 2003). BDNF is released from dendrites, axons and terminals in an activity-dependent manner to modulate synaptic strength and neuronal connectivity (Balkowiec & Katz, 2002; Martinowich et al., 2007). Likewise, cellular release of reelin contributes to synaptic strength and neuronal connectivity via its role in neuronal migration and synapse formation (D'Arcangelo, 2014). Changes in mRNA levels of these genes are a known result of early-life stress, and several groups have linked DNA methylation alterations to corresponding changes in mRNA levels (Gross et al., 2012; Kundakovic et al., 2013; Roth et al., 2009). In the current study, we measured histone 3 lysine 9/14 (H3K9/14) acetylation associated with bdnf promoters I and IV and the reelin promoter in the mPFC of adult male and female rats that were subjected to aversive or nurturing caregiving environments during infancy. Based on the susceptibility of the mPFC to early-life stress and environmentally-driven epigenetic changes, along with our previous DNA methylation results, we hypothesized that histone acetylation at bdnf and reelin promoters would be altered in the mPFC of rats subjected to caregiver maltreatment.

Methods

Animals

We obtained male and female outbred Long-Evans rats from Harlan and maintained a breeding colony in a temperature (ranging from 20 ± 1 °C) and light-controlled colony room (12 h light/12 h dark cycle with lights on at 06:00 h). Male and female rats were bred by placing one male and one female rat together in a wire-bottomed cage, and both the male and wire bottom were removed once a sperm plug was detected (indicating successful copulation). Pregnant females were singly housed in standard cages with abundant wood shavings and access to food and water ad libitum. All caregiving manipulations were performed during the light cycle. No primiparous dams were used for experimental litters or stimulus caregivers. Pup day of birth was designated as postnatal day (PN) 0 and litters were culled to 5–6 males and 5–6 females per litter on PN1 before experimental manipulations took place from PN1–7. We used brain tissue from 39 rats derived from a total of eight litters (yielding 8–10 rats per experimental group). All procedures were approved by the University of Delaware Animal Care and Use Committee.

Caregiving manipulations

Caregiver manipulations were performed as previously reported in our laboratory (Blaze & Roth, 2013; Blaze et al., 2013; Roth et al., 2014) and using a protocol adapted from others (Ivy et al., 2008; Raineki et al., 2010; Roth & Sullivan, 2005). Each experimental litter was split into three groups on PN1, with 3–4 male and female pups in each group. Pups were marked numerically each day with a non-toxic permanent marker. For 30 min per day from PN1–7, the maltreatment group was transported into a room and placed in a chamber with a lactating dam that was stressed due to insufficient bedding material and the novel environment of the testing chamber. At the same time, the cross-foster care group was placed with another lactating dam that had plenty of nesting material and was given 1 h to habituate to the chamber before receiving the pups. The remaining pups in the litter comprised the third group (normal maternal care), that were left undisturbed (except for weighing) with the biological mother during the session. After each day’s 30 min exposure session, all pups were returned to the biological mother and the stimulus dams (from the maltreatment and cross-foster conditions) were reunited with their biological litters. Exposure sessions were performed at different times every day to provide an element of unpredictability of stress exposure. Litters were left undisturbed until weaning at PN21–23 and then housed with a same-sex, same-condition littermate until PN90.

Pup-directed behaviors from dams were scored by trained observers via live observations and/or video recordings. Behaviors were tallied in 5-min time bins over the 30 min sessions and averaged across the 7 days. We also recorded 40 kHz ultrasonic (Batbox III, NHBS Ltd., UK) and audible vocalizations for each condition during the exposure sessions, which were both tallied (by trained listeners) based on their presence in 1 min time bins (regardless of duration) and likewise averaged across the seven exposure sessions.

Chromatin immunoprecipitation (ChIP)

At PN90, rats were killed by decapitation at baseline conditions (i.e. taken from the home cage with minimal disturbance) under light isoflurane anesthesia. Brains were immediately removed, sliced into coronal sections on a 1 mm brain matrix, and flash frozen in 2-methylbutane on untreated slides. Chromatin immunoprecipitation (ChIP) was performed using a modified version of the EMD Millipore ChIP Assay and other published protocols (Bredy et al., 2007; Lubin et al., 2008). The mPFC (consisting of prelimbic and infralimbic cortex) was dissected on dry ice using stereotaxic coordinates from 2.52 to 5.16 mm anterior to bregma (according to Paxinos & Watson, 2007 – see Figure 1 for representation of region of interest) and tissue was immediately cross-linked for 10 min at 37 °C. The tissue was washed twice with an ice-cold protease inhibitor mixture (Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific, San Jose, CA) and 1X PBS) and homogenized in sodium dodecyl sulfate (SDS) lysis buffer (Millipore, Danvers, MA). Chromatin was sheared on ice using the EpiShear probe sonicator (Active Motif, Carlsbad, CA) at 25% amplitude for 12 min (15 s on, 30 s off). Samples were diluted 1:10 with ChIP Dilution Buffer (Millipore) and a small amount of chromatin was saved as input DNA for later analysis. Chromatin was pre-cleared for 45 min using Protein A Agarose/Salmon Sperm DNA (Millipore) and then rotated with acetyl H3K9/14 polyclonal antibody (Epigentek, Farmingdale, NY) overnight at 4 °C. A no-antibody control was also used as a negative control in later PCR reactions. Histone–DNA–antibody complexes were precipitated with Protein A Agarose/Salmon Sperm DNA and then beads were washed with low salt buffer (two washes), high salt buffer (two washes), LiCl (two washes) and 1x Tris-EDTA (TE; three washes) (Millipore). Histone–DNA complexes were eluted with 10% SDS and 1 M NaHCO₃. Samples were incubated at 65 °C overnight with 5 M NaCl and RNase A to reverse the cross-link and then DNA was purified using Qiaquick PCR Purification Kit (Qiagen, Valencia, CA) for later amplification with real-time polymerase chain reaction (RT-PCR).

Figure 1. Brain atlas (Paxinos & Watson, 2007) representation of region of interest for medial prefrontal cortex dissections (PrL = prelimbic, IL = infralimbic). Reproduced with permission.

Real-time polymerase chain reaction (RT-PCR)

To amplify immunoprecipitated DNA specific to our candidate genes, we used primers specific to the same or nearby region of the bdnf and reelin genes as we have previously used for DNA methylation studies (Blaze et al., 2013; Roth et al., 2009, 2014) (Figure 2). For bdnf, primer sequences were as follows: bdnf I: 5′-CCCCGCTGCGCTTTTCTGGT-3′ (forward) and 5′-CAATTTGCACGCCGCTCCTTTGC (reverse) and bdnf IV: 5′-ATGCAATGCCCTGGAACGGAA-3′ (forward) and 5′-TAGTGGAAATTGCATGGCGGAGGT-3′ (reverse) (Gupta-Agarwal et al., 2012; Huang et al., 2002; Lubin et al., 2008). For the reelin promoter, primers were 5′-CGCTCGGAGGCGGACGACG-3′ (forward) and 5′-CGAAGTTACTTTGGGCCGCGGGA-3′ (reverse). PCRs were performed in triplicates and product specificity was confirmed with sequencing, melt curve analysis, and gel electrophoresis. Fluorescence for each immunoprecipitated DNA amplicon was normalized to the input DNA fragment (Bredy et al., 2007; Lubin et al., 2008).

Figure 2. Diagrams of gene targets in the study. Primer pair positions are indicated by the red text and left and right arrows. bdnf, accession NC_005102.4 (primer sequences from Gupta-Agarwal et al., 2012; Lubin et al., 2008); reelin, accession NC_005103.4.

Methylation-specific real-time PCR (MSP)

To measure DNA methylation associated with acetylated H3K9 in maltreated females, we performed a bisulfite treatment on chromatin-immunoprecipitated DNA and used methylated and unmethylated primer sets to amplify bdnf IV (as previously published, Blaze et al., 2013; Lubin et al., 2008; Roth et al., 2009). Tubulin was used as a reference gene and reactions were run in triplicate, with product specificity confirmed by melt curve analysis and gel electrophoresis.

Statistical analyses

Specific types of caregiving behaviors (i.e. step on, roughly handle, pup licking) and pup vocalizations (ultrasonic and audible) were each analyzed with one-way ANOVAs, and a two-way ANOVA was used to compare overall nurturing and adverse maternal behaviors across groups. Bonferonni or unpaired t-tests were used for between group comparisons. For RT-PCR data, we used the comparative Ct method (Livak & Schmittgen, 2001) to obtain the relative fold change of our experimental (maltreatment or cross-foster care) groups compared to the controls (normal maternal care). Differences in histone acetylation and DNA methylation were analyzed using one- or two-tailed one-sample t-tests (a mean value of 1 would indicate no change in levels in comparison to the normal care control group) and unpaired t-tests (to compare the foster care and maltreatment groups). Male and female samples were run on different 96-well PCR plates that contained a representative sample of each caregiving condition. To help minimize differences between plates within a sex the same lot of SYBR green supermix, RNase free water, and primers were used for all assays. Given that plates did not contain a standard and sexes were run independently, we did not statistically contrast sexes. Outliers that fell more than two standard deviations above or below the mean were excluded (ChIP bdnf I, n = 2, bdnf IV, n = 3; DNA methylation n = 2). We denoted significance at p < 0.05.

Results

Maternal behavior

To alter the caregiving environment of infant rat pups, we used a within-litter design to subject pups to either adverse or nurturing caregiving environments inside (normal care group) or outside (maltreatment or foster care group) of the home cage. Caregiving behaviors that were classified as aversive were stepping on, dragging, dropping, actively avoiding and roughly handling pups. Nurturing behaviors included pup licking, anogenital licking, hovering over pups and nursing. The 30-min exposures were scored in five-minute bins in which behaviors were tallied based upon their presence or not. After scoring, we calculated the percent occurrence of each behavior throughout each exposure session (determined by the number of five-minute bins in which a behavior was observed divided by total number of bins) and then averaged the values across all seven exposure days for each litter (averages are shown in Table 1). A two-way ANOVA confirmed that there was a main effect of type of caregiving behavior (F(1, 42) = 144.8, p < 0.001) and a type of caregiving behavior × infant condition interaction (F(2, 42) = 25.5, p < 0.001). The occurrence of aversive behaviors was significantly greater in the maltreatment group (p < 0.001 versus cross-foster care, p < 0.01 versus normal maternal care). Maltreated pups also experienced significantly less nurturing behavior than pups in the other caregiving conditions (p < 0.001 versus cross-foster care, p < 0.01 versus normal maternal care). Moreover, pups that experienced nurturing care inside or outside of the home cage had similar caregiving experiences (all p values for statistical comparisons >0.05). Percent occurrence of specific aversive and nurturing behaviors is listed in Table 1, along with statistical comparisons of each behavior across the three groups.

Table 1. Average percent occurrences of individual adverse and nurturing pup-directed maternal behaviors and pup vocalizations across the seven 30-min exposure sessions.

Adverse behaviors (% occurrence)							Nurturing behaviors			Pup vocalizations
	Step on F(2, 23) = 18.14 p < 0.001	Drag F(2, 23) = 7.10 p < 0.01	Drop F(2, 23) = 3.58 p < 0.05	Actively avoid F(2, 23) = 1.72 p > 0.05	Roughly handle F(2, 23) = 7.90 p < 0.01	Overall F(2, 23) = 12.75 p < 0.001	Pup licking F(2, 23) = 1.07 p > 0.05	Hover/Nurse F(2, 23) = 3.87 p < 0.05	Overall F(2, 23) = 12.75 p < 0.001	Audible F(2, 23) = 3.09 p = 0.07	Ultrasonic F(2, 23) = 11.21 p < 0.01
NMC	9.8***	4.5**	2.6	1.7	6.9*	28.7**	33.4	41.1	71.3**	32.5	59.2**
CFC	9.8***	6.3	0.2	1.4	4.8**	25.6***	33.6	43.9	74.1***	36.4	53.7***
MAL	18.5	8.7	2.3	3.9	12.3	47.3	26.1	28.2	52.7	50.9	84.4

NMC – normal maternal care, CFC – cross-foster care, MAL – maltreatment. Two-tailed unpaired t-tests revealed NMC versus CFC comparisons were not significant.

*p < 0.05.

**p < 0.01.

***p < 0.001 versus MAL.

Infant vocalizations

Infant audible and ultrasonic (40 kHz) vocalizations that occurred within the exposure chambers (the maltreatment and cross-foster care groups) or home cage (the normal care group when littermates were with foster dams) were tallied for each 1 min time bin during the 30-min exposure sessions and averaged across the seven days (Table 1). For audible vocalizations, a one-way ANOVA revealed no significant effect of infant condition (F(2, 21) = 3.1, p = 0.07), although mean values indicated that maltreated rats vocalized much more than the other groups. For 40 kHz ultrasonic vocalizations, there was a significant main effect of infant condition (F(2, 21) = 11.2, p < 0.001) such that the maltreated rats emitted significantly more vocalizations than pups in both the normal maternal care (t(14) = 3.5, p < 0.01) and cross-foster care (t(14) = 5.8, p < 0.001) conditions. Literature suggests that 40 kHz ultrasonic vocalizations may serve as distress called from pups (Portfors, 2007), indicating they are responding to the stressful environment.

bdnf and reelin H3 acetylation

We used ChIP to measure H3K9/14 acetylation in the mPFC of adult (PN90) male and female rats that experienced nurturing care or maltreatment in infancy. We first examined both promoters I and IV of the bdnf gene because these are regions where we have previously found early-life caregiving-induced changes in DNA methylation within the mPFC (Blaze et al., 2013). We did not detect any significant changes in bdnf I acetylation in maltreated, or cross-fostered, male or female rats (Figure 3A and B; all p values for statistical comparisons >0.05). At bdnf IV, there was no significant change in males (Figure 3C), but a one-sample t-test revealed that females who experienced maltreatment during infancy had significantly less acetylation of bdnf IV DNA in comparison to normal care controls (Figure 3D; t(8) = 2.5, p < 0.05).

Figure 3. Adult (PN90) H3K9/14 acetylation at brain-derived neurotrophic factor (bdnf) promoters I and IV in the medial prefrontal cortex (mPFC). One-sample (to compare to normal care controls) and two-tailed unpaired t-tests indicate that neither (A) males nor (B) females showed any differences in H3K9/14 acetylation at bdnf I. At bdnf IV, (C) males showed no significant changes, but a one-sample t-test revealed that (D) females that experienced maltreatment had less acetylation in comparison to normal care control females (*p < 0.05). Error bars represent SEM, n = 8–10/group.

Using the same methods, we measured H3K9/14 acetylation at the reelin promoter in the adult rat mPFC since we have also found changes in reelin methylation in rats from our caregiver manipulation paradigm (Blaze et al., 2013). Interestingly, there were no significant changes in acetylation of DNA associated with the reelin promoter in the mPFC of males or females exposed to either the maltreatment or cross-foster care (Figure 4A and B; all p values >0.05).

Figure 4. Adult (PN90) H3K9/14 acetylation at the reelin promoter in the medial prefrontal cortex (mPFC). One-sample and two-tailed unpaired t-tests indicate that neither (A) males nor (B) females showed any differences in H3K9/14 acetylation. Error bars represent SEM, n = 8–10/group.

bdnf DNA methylation

We have previously shown increased bdnf IV DNA methylation in maltreated females (Blaze et al., 2013). To replicate this observation and confirm the inverse relationship between histone acetylation and DNA methylation, we bisulfite-converted H3K9 ChIP DNA from the maltreated females and measured DNA methylation at bdnf IV. A one-tailed one-sample t-test confirmed that maltreated females had greater bdnf IV methylation in comparison to normal care controls (mean of 1.954 fold-change, ±0.46 SEM; t(6) = 2.1, p < 0.05).

Discussion

While DNA methylation alterations have been consistently linked to caregiving conditions, histone acetylation has not been well-characterized. The current study was designed to assess histone acetylation in the mPFC of adult rats exposed to various caregiving conditions. Our main finding here is that female rats that experienced maltreatment had less histone H3K9/14 acetylation associated with bdnf IV DNA in the mPFC, which corresponded to greater bdnf IV methylation. As early-life stress is known to induce changes in mPFC structure and activation, conferring sensitivity to stress later in life and changes in cognitive function and anxiety (Gorka et al., 2014; Hanson et al., 2012; van Harmelen et al., 2014), our data suggest both histone and DNA methylation alterations could be mechanisms underlying these long-term effects. DNA methylation involves the addition of methyl groups to cytosines, and usually represses transcription via the binding of methyl-CpG binding protein (MeCP2) and other co-repressors or by interfering with transcription factor binding. Conversely, histone acetylation is characterized by the addition of acetyl groups to lysine residues on histone tails, a reaction that is catalyzed by histone acetyltransferases (HATs) and can be reversed by histone deacetylases (HDACs) (Strahl & Allis, 2000). Histone acetylation loosens the chromatin complex and makes it more accessible for transcription factor binding, thus increasing gene expression.

We previously reported altered methylation of DNA associated with exons I and IV of the bdnf gene in the mPFC of male and female rats that experienced early-life maltreatment (Blaze et al., 2013). Specifically, in comparison to controls, females that experienced maltreatment had greater methylation of bdnf IV and less methylation of bdnf I, while males had less methylation of bdnf I. Based upon these findings we predicted that histone acetylation associated with the bdnf gene would likewise be altered in rats exposed to maltreatment. Female data for bdnf IV from the present study were consistent with our prediction, showing less histone H3K9/14 acetylation associated with bdnf IV DNA in the mPFC. Notably, this corresponds to our observation of higher methylation in this group in our previous (Blaze et al., 2013) and current study, supporting the typically observed negative correlation between DNA methylation and histone acetylation (Cedar & Bergman, 2009; Eden et al., 1998; Irvine et al., 2002; Weaver et al., 2004). Other studies have found experience-induced changes in histone acetylation associated with this same locus of the bdnf gene. For example, Fuchikami et al. (2009) revealed decreased H3 acetylation at bdnf IV in the adult rat hippocampus after a single bout of immobilization stress (Fuchikami et al., 2009). Conversely, prenatal cocaine exposure increased H3 acetylation at bdnf IV (Kabir et al., 2014). Communal nesting in infancy and fear conditioning in adulthood have too been shown to increase bdnf IV H3 (Branchi et al., 2011, Bredy et al., 2007; Lubin et al., 2008; Takei et al., 2011) and H4 (Bredy et al., 2007; Takei et al., 2011) acetylation.

In a previous study, we also saw significant changes in reelin promoter methylation in our cross-foster care group (males showed an increase, females showed a decrease), with a trending increase in maltreated males (Blaze et al., 2013). Additionally, our laboratory has previously reported decreased reelin mRNA in the mPFC of males and females that were maltreated (Blaze et al., 2013). Other groups have shown that early-life maternal separation produces alterations in adult reelin mRNA levels in the hippocampus (Gross et al., 2012; Zhang et al., 2013). We thus predicted that we would see changes to reelin promoter histone acetylation. In the current study, however, we did not find any changes in reelin H3 acetylation in males or females from either group.

While H3 acetylation data for bdnf IV are consistent with our previous and current methylation findings in maltreated female rats, we did not find any changes in acetylation at bdnf I or reelin in males or females in either of our caregiving conditions. Multiple environmental factors, including drug administration and stress in adulthood, are known to alter H3 acetylation at these same bdnf (Fuchikami et al., 2009; Lubin et al., 2008; Schmidt et al., 2012) and reelin loci (Kundakovic et al., 2007; Mitchell et al., 2005; Sui et al., 2012). It is possible that stress early in life may exert differential effects on bdnf I and reelin histone acetylation. It is also possible that histone acetylation is focused at other lysine residues on H3 or perhaps on H4, which we did not examine here.

Finally, many studies in the field of behavioral epigenetics have been conducted under the framework that environmentally-driven increases in DNA methylation and decreases in histone acetylation will correlate with sustained decreases in basal levels of gene expression. Due to methodological limitations (i.e. needing sufficient tissue for ChIP), we were unable to dedicate any tissue for RNA extractions and thus did not measure bdnf or reelin gene expression in the current study. We have previously reported lower levels of bdnf mRNA (bdnf IV-containing transcripts) in the PFC (as a whole) of maltreated rats (Roth et al., 2009). In our prior study focused on the mPFC we did not find a change in total bdnf mRNA levels, however we did not examine just bdnf IV-containing transcripts (Blaze et al., 2013). In that same study we did find a significant decrease in reelin gene expression in maltreated females. Though concomitant changes in gene expression are one way to envision functional significance of epigenetic changes, an emerging view in the field is that discordance between epigenetic changes and basal gene expression might reflect a form of CNS plasticity that serves to direct gene transcription upon subsequent stimulation (Baker-Andresen et al., 2013).

Future directions

Other studies have shown global changes in H3 and H4 acetylation (i.e. not looking at a specific gene locus) following either early- or later-life stress. For example, maternal separation in infancy alters H4 acetylation in the forebrain neocortex (Levine et al., 2012). Moreover, rats that were exposed to social defeat in adulthood have histone acetylation changes in several stress-related brain regions (Hinwood et al., 2011; Kenworthy et al., 2014). Further, our laboratory has shown that maltreated males (and females, though not statistically significant) also had lower levels of hdac1 mRNA levels compared to control males (Blaze & Roth, 2013). Because epigenetic changes to histones are present on a global level in other models, it may be useful to take a global approach to investigate histone acetylation in our model. Conversely, it may also be useful to look more specifically at other lysine residues on histones 3 and 4 as well as HDACs bound to acetylated histones at the various bdnf regulatory regions or reelin promoter.

Other histone modifications are also known to be affected by stress. In adults, both social defeat (Tsankova et al., 2006) and acute or chronic restraint stress (Hunter et al., 2009) alter histone methylation, both globally and at bdnf IV. Further, aberrant histone methylation has been associated with multiple psychiatric disorders (Akbarian & Huang, 2009; Huang et al., 2007). Less is known regarding histone methylation after early-life stress (Kao et al., 2012), but it is likely that our model elicits changes in histone methylation. It will be important in future studies to investigate repressive chromatin marks, such as H3K9 methylation, to see if these are likewise affected.

Finally, an emerging theme in the behavioral epigenetics literature is that of cell-type specificity. It would be useful to know the phenotype of cells affected by early-life maltreatment, since current methodologies (including ours here in this study) often use a heterogeneous population of cells including neurons and glia. Because methylation patterns have been shown to differ between these cell types (Li et al., 2014), future studies should distinguish epigenetic patterns between cell-populations in brain samples.

Conclusions

Although it is widely known that early-life stress exerts deleterious effects on brain function and mental health, the pursuit of mechanisms for these effects is ongoing. Using a rodent model, we have characterized changes in histone acetylation associated with bdnf and reelin DNA following exposure to caregiver maltreatment during the first week of life. Results indicate that adult females who experienced maltreatment in infancy have less bdnf IV histone acetylation, while there were no changes in bdnf I or reelin histone acetylation in either males or females. Together with our previous work, maltreated females appear to be especially vulnerable to epigenetic changes at bdnf IV. Additional research is needed to characterize how early-life stress, including maltreatment, affects histone acetylation at different loci and how these changes are reflected in downstream processes, from gene expression to behavior. Investigating epigenetic modifications to histones will also provide insights for the utility of pharmaceuticals with epigenetic targets (i.e. HDAC inhibitors) to reverse the negative effects of early-life stress (Abel & Zukin, 2008; Szyf, 2009).

Acknowledgements

We thank Hannah Evans, Samantha Jones, Hillary Porter, Brittany Rider, Lisa Scheuing, Angela Maggio, Sarah Pingar and Kristyn Borrelli for their help in generating animals, behavior coding and gel electrophoresis.

Declaration of interest

The authors report no conflicts of interest. This work was supported by a grant from the National Institute of General Medical Sciences (1P20GM103653).