Stress-induced gene expression and behavior are controlled by DNA methylation and methyl donor availability in the dentate gyrus

Animals.

Male Wistar rats (150–175 g; purchased from Harlan) were group housed (two to three animals per cage) under standard lighting (80–100 Lux; 14:10 light-dark cycle; lights on 05:00) and environmentally controlled conditions (temperature 22 ± 1 °C; relative humidity 50 ± 10%) with food and water available ad libitum. All rats were handled for at least 5 d (3 min per rat per day) before the day of the experiment to reduce nonspecific stress. All procedures were approved by the University of Bristol Ethical Committee and by the Home Office of the United Kingdom (U.K. Animal Scientific Procedures Act 1986).

Drug Treatment.

S-adenosyl methionine (SAM) (100 mg/kg; Abbott Laboratories) was dissolved in sterile pyrogen-free PBS and injected s.c. The dose of the drug was based on a previous study (18). Control animals received a vehicle injection (1 mL/kg, sterile pyrogen-free PBS). The injection was given 30 min before the start of the initial forced swim session. Unless otherwise stated, drugs and chemicals were purchased from Sigma-Aldrich.

Animal Experimentation.

Animal experimentation was conducted between 0800 and 1300. Rats were forced to swim for 15 min in individual glass beakers (height 35 cm, diameter 21.7 cm) containing water 21 ± 2 cm deep at 25 °C or were left undisturbed in their home cage. For experiments involving SAM treatment, animals were given a single s.c. injection of SAM or vehicle 30 min before forced swimming and were killed either at 1 h after the start of forced swimming or were subjected to a second forced swim (re)test 24 h later with the same test conditions. Vehicle- or SAM-injected baseline animals were killed at the corresponding time after injection. For experiments without drug treatment, animals were killed at indicated times (see figure legends) after forced swimming.

Behavioral Scoring.

Behavior in the forced swim test and retest was digitally recorded and scored in a blind fashion, as previously described (5, 11). Behavior (immobility, climbing, or swimming) was scored every 10 s during the first 5 min of the test and retest, resulting in 30 data bins.

Immunohistochemistry.

Brain tissue for immunohistochemistry was collected from rats that were deeply anesthetized with pentobarbital (1 mL/kg) and transcardially perfused with ice-cold saline [0.9% wt/vol sodium chloride (NaCl)] followed by ice-cold buffered formaldehyde solution [4% wt/vol, 0.195% wt/vol picric acid, 50 mM sodium fluoride (NaF), 1 mM sodium orthovanadate (Na₃VO₄), 0.1 M phosphate buffer (PB, pH 7.4)]. Whole brains were removed and transferred to PB [0.1 M, 50 mM NaF, 1 mM Na₃VO₄, 0.05% wt/vol sodium azide (NaN₃), pH 7.4] before being cut into 50-µm-thick coronal sections using a vibratome (VT 1000S; Leica) and stored at 4 °C until use.

Three brain sections containing the dorsal hippocampus were randomly chosen per animal (between −2.92 mm and −3.96 mm from Bregma; Paxinos and Watson rat brain atlas) (40). As described previously (5), the avidin-biotin-immuno-peroxidase (ABC) method was used to detect H3S10p-K14ac, c-Fos, and Egr-1 immunopositive neurons. All solutions contained 0.1 M PB, pH 7.4; brain sections were carefully rinsed between each step with 0.1 M PB, and all steps were performed at room temperature.

After incubation in blocking solution (2% wt/vol BSA, 50 mM glycine, 0.2% vol/vol TX-100, 0.5% wt/vol NaN₃, 50 mM NaF, 1 mM Na₃VO₄, 0.1 M PB), sections were incubated overnight with one of the following: polyclonal rabbit (pRbα) anti-H3S10p [previous work has shown that, in hippocampal neurons, the H3S10p mark is actually part of the combinatorial H3S10p-K14ac mark (10, 19); 1:300 vol/vol] and pRbα c-Fos (1:10,000 vol/vol) (both from Merck Millipore); and pRbα Egr-1 (1:1,000 vol/vol; Cell Signaling Technology). The antibody was diluted in incubating solution (0.8% wt/vol BSA, 0.1% vol/vol Triton X-100, 0.5% wt/vol NaN₃, 50 mM NaF, 1 mM Na₃VO₄, 0.1 M PB). All antibodies have been used by us and others in multiple studies (5, 11, 19, 41, 42).

Next, sections were incubated with biotinylated goat αRb IgG (1:350; Vector Laboratories) and in avidin-biotinylated horseradish peroxidase complex (1:350; Vector Laboratories). After incubation with 3,3′-diaminobenzidine tetrahydrochloride (DAB) (0.8 nM, 7.5 mM ammonium chloride, 0.01% vol/vol nickel chloride, 0.1 M PB), the peroxidase reaction was initiated upon addition of hydrogen peroxide to a final concentration of 0.003% wt/vol. Sections were mounted on poly-l-lysine–coated slides, dehydrated, and coverslipped.

H3S10p-K14ac, c-Fos, and Egr-1 immunopositive granule neurons in the dentate gyrus and Egr-1 and c-Fos immunopositive pyramidal neurons in the CA1 and CA3 were counted in a blind fashion using a light microscope (Leica). For each animal, the positively stained neurons in three brain sections were counted and averaged. The immunopositive granule neurons in the dentate gyrus were counted, discriminating between the dorsal and ventral blade. Data are expressed as the mean (±SEM) number of immunopositive neurons per dentate gyrus, CA1 or CA3 in a 50-µm-thick brain section per animal.

Bisulfite Pyrosequencing.

For bisulfite pyrosequencing studies and RNA analysis, rats were rapidly anesthetized using isoflurane vapor (<15 s; Meriam Animal Health Ltd.) and decapitated. Next, the brain was removed and immediately cut into 1-mm coronal slices using an ice-cold brain matrix, and the slices were placed onto ice-cold steel boxes. Using a preparative light microscope (Leica), the dentate gyrus and the hippocampal CA regions (CA_1–3) were microdissected from the dorsal hippocampus region and collected in separate tubes on dry ice and stored at −80 °C.

Dentate gyrus and hippocampal CA regions tissue was lysed in 10 volumes of lysis buffer (0.5% wt/vol SDS, 0.1 M EDTA, 10 mM Tris⋅HCl, pH 8.0), before incubation with RNase A (0.05 µg/µL) and proteinase K (0.25 µg/µL). Genomic DNA was purified using a standard phenol-chloroform method and precipitated. Genomic DNA from dentate gyrus and hippocampal CA regions were subjected to bisulfite conversion in duplicate using the EZ DNA methylation kit (Zymo Research) following the manufacturer’s instructions. Fully methylated and unmethylated rat DNA controls (EpigenDx) were included along with negative water controls in all experimental steps.

The University of California, Santa Cruz (UCSC) Genome Browser and BLAT alignment tool (43, 44) were used to identify locations of interest within the c-Fos and Egr-1 genes. Two areas surrounding each gene were selected for primer design, a region within the CpG island located upstream of the transcriptional start site (TSS), and within the region coding for the 5′ untranslated region of the mRNA. PCR and sequencing primers were designed using PyroMarkAssayDesign2.0 (Qiagen). Sequences were as follows: c-Fos area 1, forward primer Bt-GTAGAGTTGATGATAGGGAGTT, reverse primer ACTCTATCCAATCTTCTCAATTAC, sequencing primer CTATCCAATCTTCTCAATTACTAA; c-Fos area 2, forward primer GTAGTTAGTAATTGAGAAGATTGGATAGA, reverse primer Bt-CCAAAAATAAACACTAATAAAAACTAC, sequencing primer TGAGAAGATTGGATAGAG; Egr-1 area A, forward primer GGGTTTGGGTTTTTTTAGTTTAG, reverse primer Bt-CCCTCCCCCTCCTTAATT, sequencing primer TTTGGGTTTTTTTAGTTTAGT; and Egr-1 area B, forward primer GTTAGTTTGGGGGTTTATTTATATTTT, reverse primer Bt-TCAACAACATCATCTCCTCCAATTTA, sequencing primer GTTTATTATTTAATATTAGT.

Bisulfite-treated DNA was added to a mastermix [1× reaction buffer, 25 mM MgCl₂, 0.5 units HotStar Taq (Qiagen), 200 nM of each deoxynucleoside triphosphate (dNTP) (Fisher Scientific), 0.25 µM forward and reverse primers] and amplified in duplicate using a thermo cycler (T-100; Bio-Rad). After purification, DNA methylation was quantified by pyrosequencing using the PyromarkQ24 Pyrosequencer (Qiagen) according to the standard manufacturer’s protocol. The data show the mean (±SEM) percentage of DNA methylation for each CpG site per experimental group.

Chromatin Preparation and ChIP.

For chromatin immunoprecipitation (ChIP), rats were rapidly anesthetized using isoflurane vapor (<15 s; Meriam Animal Health Ltd.) before decapitation and removal of brains for dissection and processing. To prepare tissue for ChIP, the whole hippocampus was dissected on ice-cold steel boxes. Tissue was quickly chopped into ∼1-mm³ pieces before cross-linking. All solutions used during cross-linking procedure contained the following inhibitors: 50 mM NaF, 1 mM Na₂VO₄, 5 ng/mL aprotinin, 0.1 mM phenylmethanesulfonylfluoride (PMSF), 5 mM sodium butyrate (NaBut) (Merck Millipore). Tissue was cross-linked in formaldehyde (1% wt/vol, PBS, pH 7.4) for 10 min at room temperature before quenching with glycine at a final concentration of 125 mM. After two washes with ice-cold PBS, the tissue was transferred into storage buffer [10 mM Hepes, 10 mM KCl, 1.5 mM magnesium chloride (MgCl₂), 0.1 mM ethylene glycol tetraacetic acid (EGTA), 0.5 mM DTT, pH 7.9], snap-frozen in dry ice, and stored at −80 °C.

For the preparation of chromatin, if solutions are stated to contain inhibitors, these inhibitors were as follows: 50 mM NaF, 1 mM Na₂VO₄, 5 ng/mL aprotinin, 0.1 mM PMSF, 5 mM NaBut (Merck Millipore), EDTA-free protease inhibitor mixture tablet (1 per 10 mL; Roche), and phosphatase inhibitor mixture tablet (1 per 10 mL; Roche). Buffers were kept ice-cold with the exception of the sonication buffer.

Cross-linked hippocampal tissue was homogenized in ice-cold homogenization buffer (10 mM Hepes, 10 mM KCl, 1.5 mM MgCl₂, 0.1 mM EGTA, 0.5 mM DTT, with inhibitors, pH 7.9). After centrifuging (6,000 × g, 4 °C, 5 min), the supernatant was discarded, and the pellet was resuspended in lysis buffer (0.25 M sucrose, 0.5% vol/vol Igepal, 100 mM NaCl, 6 mM MgCl₂, 0.5 mM DTT, 1 mM EGTA, 50 mM Tris, inhibitors, pH 8.0) and incubated on ice. The samples were mechanically lysed using a 25-gauge needle (Becton, Dickinson and Company). After centrifugation (2,000 × g, 4 °C, 5 min), the top two layers were discarded, leaving the pellet, which was resuspended in homogenization buffer and layered onto a sucrose cushion (1.2 M sucrose, 100 mM NaCl, 6 mM MgCl₂, 1 mM EGTA, 50 mM Tris, 0.5 mM DTT, inhibitors, pH 8.0) before centrifugation (10,000 × g, 4 °C, 30 min). The remaining sample on top of the sucrose cushion was placed on a second cushion, and the centrifugation step was repeated. The pellet was resuspended in sonication buffer (1% wt/vol SDS, 10 mM EDTA, 50 mM Tris, inhibitors, pH 8.0) and sonicated using a water-cooled (4 °C) Bioruptor (UCD-300; Diagenode) for 10 cycles (high power, 30 s on, 60 s off). Samples were removed from the Bioruptor, briefly vortexed, and sonicated for a further 10 cycles. Small aliquots of all samples were run on an agarose gel to confirm that chromatin was sheared to 1–4 nucleosomes in size (∼150–600 bp).

ChIP was performed using the method of Stock et al. (38) with modifications. Chromatin (20 µL) was diluted in 180 µL of water to be used as the input. For IP, 150 µg of chromatin was mixed with dilution buffer (50 mM NaCl, 1 mM EDTA, 10 mM Tris, inhibitors, pH 7.4) to a final volume of 1 mL. One of the following antibodies was added to the chromatin mixture: mRbα H3K4me3, pRbα H3K9me3, pRbα H3K27me3 [1:50 vol/vol; Merck Millipore; these antibodies were verified for specificity by Egelhofer et al. (45)], or rabbit IgG (1:50 vol/vol; CST) was added, and the samples were rolled overnight at 4 °C. Next, the chromatin/Ab mix was incubated with preblocked protein A-coated magnetic beads (30 mg/mL; Life Technologies) for 3 h at 4 °C. Beads were collected using a magnetic rack (New England Biolabs) and washed several times to reduce nonspecific binding: 2× wash buffer 1 (0.1% wt/vol SDS, 1% vol/vol Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris, pH 8.0); 2× wash buffer 2 (0.1% wt/vol SDS, 1% vol/vol Triton X-100, 2 mM EDTA, 500 mM NaCl, 20 mM Tris, pH 8.0); 1× wash buffer 3 (1 mM EDTA, 150 mM LiCl, 0.5% vol/vol Igepal, 0.5% wt/vol sodium deoxycholate, 20 mM Tris, pH 8.0); and finally 2× TE buffer (1 mM EDTA, 10 mM Tris, pH 8.0). The pellets were incubated with elution buffer 1 (1.5% wt/vol SDS, 50 mM NaCl, 10 mM Tris, pH 7.4), followed by a second incubation with elution buffer 2 (0.5% wt/vol SDS, 50 mM NaCl, 10 mM Tris, pH 7.4), after which the supernatants (i.e., the bound fractions) were pooled. NaCl was added to all samples to a final concentration of 150 mM before heating overnight at 65 °C to reverse the cross-links.

The bound and input fractions were sequentially incubated with RNase A and proteinase K. Next, the DNA was purified using the phenol-chloroform method. To precipitate the DNA, glycogen was added to a final concentration of 0.1 µg/µL as a coprecipitant before addition of 1/10th the volume of sodium acetate (3 M, pH 5.2) and 3 volumes of ice-cold ethanol (EtOH; 100%); then samples were stored at −20 °C overnight. Next, the DNA pellets from the bound and input samples were collected and washed [70% (vol/vol) EtOH] before being resuspended in nuclease-free water and quantified using a Qubit 2.0 Fluorometer (Life Technologies).

The Dnmt3a, Dnmt3b, and Tet1 ChIP were conducted using a slightly different protocol, which was published recently (7). The following ChIP-grade antibodies were used: pRbα Tet1 (1:200 vol/vol; Merck Millipore), pRbα Dnmt3a (1:200 vol/vol; Abcam), and mMsα Dnmt3b (1:200 vol/vol; Abcam).

Bound DNA and input DNA were subjected to quantitative PCR (qPCR) using a StepOnePlus machine (Life Technologies), and each sample was run in triplicate. DNA was added to a mastermix [900 nM forward and reverse primers, 200 nM probe, 1× TaqMan fast mastermix (Life Technologies), nuclease-free water]. Primers and probe sequences were as follows: c-Fos, forward primer CGCTCATGACGTAGTAAGCCATT, probe 6FAM-CGGCCAGCTGAGGCGCCTACT-TAMRA, reverse primer CTGCAATCGCGGTTGGA; Egr-1, forward primer GACCCGGAAACACCATATAAGG, probe 6FAM-AAGGATCCCCCGCCGGAACAG-TAMRA, reverse primer AAGGCGCTGCCCAAATAAG.

A standard curve was constructed for each amplicon using whole rat brain genomic DNA (100 ng/µL; BioChain), and run in parallel. The amount of target DNA in the bound and input fractions was calculated by comparing the Ct (cycle threshold) values of these samples to the standard curve. The data are expressed as the ratio of [bound fraction/(input fraction/20)], and the mean (±SEM) was calculated.

RNA Analysis.

RNA extraction from the dentate gyrus and the rest of the hippocampus tissue was performed using the guanidinium thiocyanate-phenol-chloroform (TRI) extraction method (39). Tissue was homogenized in TRI reagent and centrifuged (12,000 × g, 10 min, 4 °C) to remove insoluble material. The supernatant was removed, and samples were allowed to stand at room temperature for 15 min. Then, 1-bromo-3-chloropropane was added, and the sample was shaken vigorously. Next, the samples were centrifuged (12,000 × g, 15 min, 4 °C), and the upper phase (RNA) was removed. After adding 2-propanol and centrifuging (12,000 × g, room temperature), the pellet was washed and resuspended in nuclease-free water (Life Technologies) before quantification using a NanoPhotometer P300 (Implen). The RNA integrity was checked using a Bioanalyser (Agilent Technologies). RNA was converted to cDNA using a thermo cycler (T-100; Bio-Rad) and a reverse transcription kit (QuantiTect Reverse Transcription Kit; Qiagen) according to the manufacturer’s protocol.

For qPCR, 2 µL of cDNA was used per reaction in 18 µL of mastermix [900 nM forward and reverse primers, 200 nM probe, 1× TaqMan fast mastermix (Life Technologies), nuclease-free water] using a StepOnePlus machine (Life Technologies). Primers and probes were as follows: Dnmt1, forward primer CAAAGCAAGTGCAATCCCAAA, probe 6FAM-CAACCCGCCACAGTGCCCTGAG-TAMRA, reverse primer TCAGGTCAGGGTCATCTAGGTACTG; Dnmt3a, forward primer AAGGTCAAGGAGATCATTGATGAAC, probe 6FAM-TGCGCCAGAAGTGCCGAAACATC-TAMRA, reverse primer CATTGAGGCTCCCACATGAGAT; Dnmt3b, forward primer CCTGGCATGTAACCCAGTGA, probe 6FAM-TCGACGCCATCAAGGTTTCTGCTG-TAMRA, reverse primer GGCCTGTTCATTCCAGGTAGAT. For Tet1 mRNA, the AB TaqMan gene expression assay Rn01428192 was used.

At least three standard curves were run for each primer pair, and the average qPCR efficiency was calculated using the equation: E = (10^-1/slope) – 1) × 100 (where E is qPCR efficiency and the slope is the gradient of the standard curve). The relative mRNA expression ratio was calculated using the Pfaffl method of relative quantification (46) and standardized to the housekeeping genes Hprt1 and Ywhaz. The data were normalized to the baseline group and shown as the mean (±SEM) relative mRNA expression.

Statistical Analysis.

The data were statistically analyzed using analysis of variance (ANOVA), and, in case of a significant main effect, differences between individual groups were evaluated with appropriate post hoc tests, adjusting the level of significance according to the Bonferroni procedure to reduce the probability of a type 1 error. Some data were analyzed using Student’s t test. Results shown are mean ± SEM, with P < 0.05 considered statistically significant.

The Effect of SAM on Forced Swimming-Induced Behavior (Fig. 1).

Statistical analysis was as follows (two-way ANOVA): (A) effect of behavior, F_(2,45) = 22, P < 0.0001; effect of SAM, F_(1,45) = 0.0, P = 1.0; interaction behavior × SAM, F_(2,45) = 0.14, P = 0.86; (B) effect of behavior, F_(2,45) = 13, P < 0.0001; effect of SAM, F_(1,45) = 0.0, P = 1.0; interaction behavior × SAM, F_(2,45) = 6.8, P < 0.01. Bonferroni-corrected post hoc test with contrasts was as follows: *P < 0.05 compared with the respective vehicle-treated group; ^&P = 0.072 compared with the respective vehicle-treated group.

The Effect of SAM on Forced Swimming-Evoked c-Fos and Egr-1 Induction in the Dentate Gyrus (Fig. 2).

Statistical analysis was as follows: (A) two-way ANOVA; effect of SAM, F_(1,18) = 20, P < 0.001; effect of stress, F_(1,18) = 140, P < 0.0001; interaction SAM × stress, F_(1,18) = 37, P < 0.0001; (B) two-way ANOVA; effect of SAM, F_(1,18) = 10, P < 0.01; effect of stress, F_(1,18) = 6.1, P < 0.05; interaction SAM × stress, F_(1,18) = 6.9, P < 0.05. Bonferroni-corrected post hoc test with contrasts was as follows: *P < 0.05 compared with the respective BL group; ^$P < 0.05 compared with the respective vehicle/FS60 group; ^#P < 0.05 compared with the respective ventral blade group.

The Effect of SAM on H3S10p-K14ac Formation in the Dentate Gyrus After Forced Swimming (Fig. 3).

Statistical analysis was as follows: two-way ANOVA; effect of SAM, F_(1,16) = 0.24, P = 0.63; effect of stress, F_(1,16) = 20, P < 0.001; interaction SAM × stress, F_(1,16) = 0.0034, P = 0.95. Bonferroni-corrected post hoc test with contrasts was follows: *P < 0.05 compared with the respective BL group; ^#P < 0.05 compared with the respective ventral blade group.

Forced Swimming-Induced CpG-Specific DNA Methylation Changes in the c-Fos Promoter Region (Fig. 4).

Statistical analysis was as follows: two-way ANOVA with repeated measures; area 1, effect of CpG number, F_(6,36) = 81, P < 0.0001; effect of stress, F_(1,36) = 0.26, P = 0.63; interaction CpG number × stress, F_(6,36) = 0.65, P = 0.69; area 2, effect of CpG number, F_(6,48) = 39, P < 0.0001; effect of stress, F_(1,48) = 3.1, P = 0.12; interaction CpG number × stress, F_(6,48) = 2.2, P = 0.065. Student’s t test was as follows: *P < 0.05; ^&P < 0.1, compared with the respective BL group.

Forced Swimming-Induced CpG-Specific DNA Methylation Changes in the Egr-1 Promoter Region (Fig. 5).

Statistical analysis was as follows: two-way ANOVA with repeated measures; (B) area A, effect of CpG number, F_(16,160) = 48, P < 0.0001; effect of stress, F_(1,160) = 4.1, P = 0.070; interaction CpG number × stress, F_(16,160) = 3.1, P < 0.001; area B, effect of CpG number, F_(7,63) = 370, P < 0.0001; effect of stress, F_(1,63) = 0.37, P = 0.56; interaction CpG number × stress, F_(7,63) = 1.9, P = 0.090. Student’s t test was as follows: *P < 0.05; ^&P < 0.1, compared with the respective BL group.

The Effect of SAM Treatment on Forced Swimming-Induced DNA Methylation Changes at CpGs Within the c-Fos UTR in the Dentate Gyrus (Fig. 6).

Statistical analysis was as follows: three-way ANOVA; effect of SAM, F_(1,133) = 1.9, P = 0.17; effect of stress, F_(1,133) = 4.5, P = 0.035; effect of CpG number, F_(6,133) = 99, P < 0.0001; interaction SAM × stress, F_(1,133) = 8.9, P < 0.01; interaction SAM × CpG number, F_(5,133) = 1.3, P = 0.28; interaction stress × CpG number, F_(6,133) = 0.91, P = 0.49; interaction SAM × stress × CpG number, F_(5,133) = 1.2, P = 0.30. Bonferroni-corrected post hoc test with contrasts was as follows: *P < 0.05 compared with the respective vehicle/FS60 group; ^$P < 0.05 compared with the respective SAM/BL group; ⁺P = 0.076 compared with the respective SAM/BL group; ^&P = 0.076 compared with the respective vehicle/FS60 group.

The Effect of SAM Treatment on Forced Swimming-Induced DNA Methylation Changes at CpGs Within the Egr-1 Gene Promoter in the Dentate Gyrus (Fig. 7).

Statistical analysis was as follows (three-way ANOVA): effect of SAM, F_(1,255) = 51, P < 0.0001; effect of stress, F_(1,255) = 5.4, P < 0.05; effect of CpG number, F_(16,255) = 29, P < 0.0001; interaction SAM × stress, F_(1,255) = 17, P < 0.0001; interaction SAM × CpG number, F_(11,255) = 1.9, P < 0.05; interaction stress × CpG number, F_(16,255) = 2.1, P < 0.05; interaction SAM × stress × CpG number, F_(11,255) = 1.9, P = 0.051. Bonferroni-corrected post hoc test with contrasts was as follows: *P < 0.05 compared with the respective vehicle/FS60 group; ^$P < 0.05 compared with the respective SAM/BL group.