Early life stress induces age-dependent epigenetic changes in p11 gene expression in male mice

Early life stress (ELS) causes long-lasting changes in gene expression through epigenetic mechanisms. However, little is known about the effects of ELS in adulthood, specifically across different age groups. In this study, the epigenetic modifications of p11 expression in adult mice subjected to ELS were investigated in different stages of adulthood. Pups experienced maternal separation (MS) for 3 h daily from postnatal day 1 to 21. At young and middle adulthood, behavioral test, hippocampal p11 expression levels, and levels of histone acetylation and methylation and DNA methylation at the hippocampal p11 promoter were measured. Middle-aged, but not young adult, MS mice exhibited increased immobility time in the forced swimming test. Concurrent with reduced hippocampal p11 levels, mice in both age groups showed a decrease in histone acetylation (AcH3) and permissive histone methylation (H3K4me3) at the p11 promoter, as well as an increase in repressive histone methylation (H3K27me3). Moreover, our results showed that the expression, AcH3 and H3Kme3 levels of p11 gene in response to MS were reduced with age. DNA methylation analysis of the p11 promoter revealed increased CpG methylation in middle-aged MS mice only. The results highlight the age-dependent deleterious effects of ELS on the epigenetic modifications of p11 transcription.

Children exposed to early life stress (ELS) such as neglect and abuse have a significantly increased risk of developing depression 1,2 . In human and animal studies, ELS has been reported to induce a depression-like phenotype in adulthood 3,4 . These studies are focused on the epigenetic mechanisms, for example, DNA methylation and histone modification, by which ELS may alter the expression of genes involved in the stress response, including brain-derived neurotrophic factor (BDNF), glucocorticoid receptor (GR; NR3C1), and corticotrophin-releasing factor 3,4 . The emphasis of these papers is mainly on the detrimental effects of ELS. However, little is known regarding the effects of ELS on behaviors and epigenetic mechanisms over the lifespan, especially among different adult age groups.
DNA methylation and histone modification, two representative epigenetic mechanisms, control gene transcription by affecting chromatin remodeling 5 . DNA methylation at the 5′-C-phosphate-G-3′ (CpG) dinucleotide is classically associated with transcriptional repression 6 . Regions with a high density of CpGs are known as CpG islands; they are often found in the gene regulatory regions of promoters. Methylated CpGs in a promoter can induce gene silencing by blocking transcription factor binding or by attracting proteins that cause chromatin remodeling 6 . Histone modifications affect chromatin structure via post-translational modification, for example, acetylation and methylation, of the lysine (K) residues in the histone tails 7 . Acetylation of histone H3 or H4 relaxes the interaction between DNA and histone, allowing the transcriptional machinery access to the promoter, thereby activating transcription 7 . K4 and K19 on histone H3 are commonly modified in gene transcription activation 8 . Histone methylation, in contrast, is associated with both gene activation (H3K4 and H3K36) and repression (H3K9, H3K27, and H4K20), depending on the K residue methylated and its valence state (i.e., mono-, di-, or tri-methylation) 9 .
In animals that had experienced ELS, those in young adulthood exhibit decreased H3K9 di-methylation (me2) at the BNDF IV promoter and, accordingly, increased BDNF IV expression, whereas those in middle adulthood show increased repressive H3K9me2 at the BDNF IV promoter and concurrent decreased BDNF

Results
Effects of MS on immobility in the FST and hippocampal p11 expression in young adult and middle-aged mice. Young adult and middle-aged MS animals and control animals were tested on FST according to the experimental design timeline (Fig. 1). In two-way ANOVA analysis (Table 1), age effect did not influence immobility time (F (1,42) = 1.941, p = 0.171). However, we observed the significance on MS effect (F (1,42) = 4.172, p = 0.047) and interaction effect between age and MS (F (1,42) = 6.940, p = 0.012). In post hoc test, Figure 1. Experimental design timeline. Pups were separated from their mothers for 3 h daily from postnatal day (PND) 1 to PND 21. When the pups reached young adulthood (2 months) or middle age (8 months), they were subjected to the forced swimming test (FST). Following the FST, the mice were sacrificed immediately for the p11 molecular analysis. Molecular analysis was measured within the same batch experiment in all groups.  (Table 1). Moreover, the statistical analysis showed the interaction of these factors (F (1,42) = 22.520, p < 0.001). In post hoc comparison, MS animals showed a significant reduction in p11 mRNA in young (control = 1.04 ± 0.11, MS = 0.79 ± 0.06; p = 0.049) and middle (control = 4.68 ± 0.55, MS = 1.71 ± 0.02; p < 0.001) adulthood (Fig. 2B).

Figure 2.
Changes in immobility on the FST and hippocampal p11 levels in young adult and middle-aged mice following maternal separation (MS) exposure. (A) Immobility time in the FST was measured at 2 months (young adulthood) or 8 months (middle adulthood). (B) p11 mRNA levels in the hippocampus were measured using quantitative real-time polymerase chain reaction (qRT-PCR). All analyses were performed on the hippocampus. All quantities were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Data are expressed as values relative to mean ΔCt (control group on young adulthood) using the 2 −ΔΔCt method and are presented as mean ± standard error of the mean (SEM; n = 10-12/group, young adulthood; n = 11-13/ group, middle adulthood) * p < 0.05 vs. control in young adulthood; † † p < 0.01 vs. control in middle adulthood; † † † p < 0.001 vs. control in middle adulthood.

Discussion
This paper reports that MS in early life exerts negative effects on epigenetic mechanisms associated with decreased p11 expression in adulthood, and these effects become more pronounced with age. In present study, MS had no effect on immobility in young adult mice, but increased immobility was observed in middle age. In a past study, young adult (2 months) MS mice did not affect immobility in the FST; however, MS  (H3K4me3, B), and histone H3 trimethylation at lysine residue 27 (H3K27me3, C) at the p11 promoter in the hippocampus were measured using a chromatin immunoprecipitation assay with antibodies to AcH3, H3K4me3, and H3K27me3. All analyses were performed on the hippocampus. Data were normalized to input DNA and are expressed as values relative mean ΔCt (control group on young adulthood) using the 2 −ΔΔCt . The data are presented as mean ± SEM (n = 10-12/group, young adulthood; n = 11-13/group; middle adulthood). * p < 0.05 vs. control in young adulthood; **p < 0.01 vs. control in young adulthood; ***p < 0.001 vs. control in young adulthood; † † † p < 0.001 vs. control in middle adulthood. . Alterations in DNA methylation at the hippocampal p11 promoter in young adult and middle-aged mice following MS exposure. DNA methylation levels at the hippocampal p11 promoter were measured using methylation-specific polymerase chain reaction (MSP). All analyses were performed on the hippocampus. Ct values were normalized to GAPDH. Data are expressed as percentage of methylated DNA described in "Methods" and are represented as mean ± SEM (n = 10-12/group, young adulthood; n = 11-13/group, middle adulthood). † p < 0.05 vs. control in middle adulthood; # p < 0.05 vs. MS in young adulthood.  PND 14) in rats increased immobility time at two ages; furthermore, middle-aged (10 months) rats had longer immobility times than young adult (4 months) rats 25 . Taken together, early MS significantly increased immobility in middle adulthood than in young adulthood. The present study showed that immobility time was reduced in middle aged-control animals compared to young adult controls. Similar result has been reported in young age (2 months) and middle aged (8 months)-controls of male C57BL6J mice 13 . In large-scale analysis of the FST and tail suspension test in male C57BL6J mice, immobility times in the 8-12 months old group (n = 70) were lower than those in the 4-5 months (n = 357) and 2-3 months (n = 495) old groups 23 . Overall, the results of tests in C57BL6J mice indicate that immobility decreases with age. In this study, MS is associated with a reduction of hippocampal p11 expression at two time points in adulthood. Previous studies have demonstrated that chronic stress in adults induces p11 loss, as well as depression-like behaviors 26,27 . This study is the first to show an association between ELS and p11 levels. In mice, p11 overexpression increases 5-HT 1B receptor function and is associated with reduced immobility in the tail suspension test 15 . In contrast, in p11 knockout mice, the number of 5-HT 1B receptors is reduced at the cell membrane 15 . Notably, MS (3 h daily from PDN 2 to PND 13) in a rat ELS model reduced hippocampal 5-HT 1B receptor binding when assayed through [ 125 I] cyanopindolol autoradiography 28 . The reduced binding may result from ELS-induced reductions of p11 levels.
Reduced hippocampal p11 expression in young adult and middle-aged MS animals was associated with altered histone modifications (i.e., significant decreases in H3 acetylation and H3K4me3 and a significant increase in H3K27me3) at the p11 promoter region. Moreover, p11 expression, H3 acetylation, and H3K4 trimethylation after MS were more severely perturbed in middle adulthood. Suri et al. demonstrated that MS induces opposing age-dependent effects on the expression of histone modifying enzymes such as histone deacetylase (Hdac1 to Hdac11) and histone methyltransferase (G9a and Suv391) in young adulthood (2 months) versus middle adulthood 29 . Nonetheless, the altered expression of these enzymes did not affect overall H3 acetylation, H3K9me2, and H3K9me3 at either of the ages. Although histone modifying enzymes may not globally change histone modification across the lifespan, these enzymes may differentially regulate histone modifications at the promoters of stress-related genes during aging. A loss of balance between epigenetic modifications during aging is referred to as "epigenetic drift" 30 . The global distribution of many histone acetylation and methylation modifications (i.e., H3K27me3, H3K56ac, and H4K16ac) has been found to be altered during aging in various organisms 31,32 . Accordingly, MS animals in middle age, but not young adulthood, show histone modifications associated with decreased BDNF IV expression concurrent with impairments in hippocampal-dependent cognition 10 . A similar age-dependent effect on histone acetylation and methylation at the GR exon I 7 promoter has been observed 13 . Taken together, middle-aged MS animals may exhibit more severe changes in histone modification than young adult MS animals due to age-related epigenetic drift.
Increased DNA methylation in p11 promoter regions following MS were only present in middle-aged mice, not those in young adulthood. The increased DNA methylation, along with the altered histone modifications, could account for the reduced p11 expression in middle-aged MS animals. In accordance, a negative correlation between p11 gene transcription and CpG DNA methylation at the p11 promoter has been reported 33 . Moreover, reduced p11 levels in a genetic rodent model of depression were associated with higher DNA methylation, while escitalopram treatment elevated p11 levels and reduced DNA methylation of the p11 promoter.
In rodents, the absence of maternal care such as licking, grooming, and arched-back nursing, is associated with hypothalamic-pituitary-adrenal (HPA) axis dysregulation as identified by increased glucocorticoid levels 34 . The lack of maternal care modifies DNA methylation, resulting in decreased hippocampal GR expression 35 . Accordingly, adults with histories of childhood maltreatment show increased DNA methylation of the GR exon I F promoter and, consequently, increased HPA activity 36,37 . GR acts as a ligand-activated transcription factor. When activated by glucocorticoids, GR translocates to the nucleus and binds to GR binding sites within the promoters of target genes, thereby activating or repressing their expression 38 . p11 is a target gene for GRs. Putative GR binding sites were identified using the web-based tools described in "Methods" section; this site includes one CpG site (Fig. 5). Zhang et al. reported that GR increases p11 promoter activity via the interaction of glucocorticoid-bound GR with GR binding sites 39 . Thus, elevated GR levels may activate p11 transcription. Hippocampal GR expression levels have been found to be reduced in young adult and middle-aged MS animals and, furthermore, the reduction in middle adulthood was remarkably higher than in young adulthood 13 . Thus, the differential reduction in p11 levels in middle-aged MS animals could be due to CpG methylation of the GR binding site, which prevents GR from easily accessing the p11 promoter, resulting in reduced p11 transcriptional activity. Nevertheless, further experiments are needed to address whether MS causes an age-dependent decrease in nuclear GR levels.
Research on the long-term effects of ELS suggests that ELS can be beneficial or detrimental depending on factors such as social support, enriched environment, and additional stress. Clinical study has shown that subjects with ELS experience reduced their vulnerability to stress by reducing stress and negative effects in daily life though work-related social support 40 . In maternally separated rats, exposure to an enriched environment during adolescence enhanced cognitive function and BDNF signaling in the hippocampus 41 . On the other hand, the combination of early-life MS and adult social defeat stress reveled significant dysregulation of the histone methylation, BDNF, β-catenin, and GR signaling, indicating that ELS increases vulnerability to stress 42 .
There are several limitations to the present study. Firstly, we performed FST to assess the depression-like behavior. It is not sufficient to use only FST. Additional behavioral tests are needed, such as the tail suspension test to measure behavioral despair state or the sucrose preference test to measure anhedonia. Also, behavioral tests that measure reward and arousal may be more appropriate. In particular, the effects of ELS on learning www.nature.com/scientificreports/ and memory are well established, and there is evidence that these effects worsen with age 43 . The Morris water maze test, which measures hippocampal dependent learning and memory, may also be appropriate. Secondly, the weakness of the current study is that it has only been tested in male mice. Epidemiological studies have shown that depression is more prevalent in women than men 44 . For depression-related studies, some researchers preferred male mice to female ones due to variability in the estrus cycles [45][46][47][48][49][50] . Others, on the other hand, used both male and female animals during their studies, and the baseline variation in females was not greater than that of males [51][52][53][54] . In further study, it is desirable to investigate both male and female mice. Finally, we did not control the effects of litter, but it's important to take this into account. Variation between dams/litters can affect behavioral and molecular outcomes. Therefore, if more than one animal is used from each litter, they should be averaged to an n = 1 to control for litter effects. The effect of MS on the immobility of FST is also age dependent. However, we cannot conclude that the results of FST in the current study reflect only epigenetic modification of the p11 gene. This is because epigenetic mechanisms of other stress-related genes may also be involved. Changes in p11 expression and epigenetic regulation with age may partially mediate the behavioral change associated with the state of despair seen in FST. In conclusion, this study demonstrates that MS has age-dependent effects on the expression of p11 gene that is associated with changes in histone modification and DNA methylation. Furthermore, the deleterious effects are more severe with age.

Methods
Animals.

Maternal separation (MS).
Each dam and its litters were kept together in each cage. Sixteen dams/their litters were divided randomly into control and MS groups. Pups in the MS group were separated from their mothers for 3 h daily starting on postnatal day (PND) 1 to PND 21 as described previously 56 . Pups in the control group were left undisturbed except during routine animal facility handling. Litters were kept with their dams until PND 35, then separated by sex and randomly housed to 5-7 animals per cage. Only male mice were used. Male control and MS mice were assayed at either 2 months old (young adulthood) or at 8 months old (middle adulthood; Fig. 1). These ages were defined according to the comparison between human and C57BL/6J mouse ages: young adult (2-6 months old) and middle age (8-14 months old) 23,57 .. Measurement of mRNA levels using quantitative real-time polymerase chain reaction (qRT-PCR). Following the FST, whole brains were extracted (n = 10-12/group, young adulthood; n = 11-13/ group, middle adulthood). The hippocampus was dissected from the brain; RNA isolation, cDNA synthesis, and qRT-PCR were performed on the hippocampal tissue as described previously 56 . Gene-specific primers ( Table 2)  Chromatin immunoprecipitation (ChIP) assays. Chromatin was extracted from the isolated hippocampus using a standard protocol (SimpleChIP ® Plus Enzymatic Chromatic IP Kit; Cell Signaling, Beverly, MA, USA) as described previously (n = 10-12/group, young adulthood; n = 11-13/group, middle adulthood) 56 . Primers were designed around a putative p11 promoter region (Table 2 and Fig. 5). Because the p11 promoter is not well characterized in mice, the region of interest was created based on previous data on epigenetic alterations in the rat p11 promoter 33,58 . The putative proximal promoter of the p11 gene (~ 700 base-pair upstream region) includes transcription factor binding sites generated by PROMO 59,60 and MotifMap 61,62 , two freely available webbased tools that identify presumptive transcription factor binding sites (Fig. 5).
Methylation-specific polymerase chain reaction (MSP) analysis. Genomic DNA was extracted from the hippocampus using a QIAGEN DNA prep kit (51036; Valencia, CA, USA) and treated with bisulfite using an EpiTect ® Bisulfite kit (59104; QIAGEN). To determine the DNA methylation status of CpG in the p11 promoter region, a qRT-PCR was performed on the same amount of bisulfite-treated DNA using an EpiScope ®  Forward TGC TCA TGG AAA GGG AGT TC   Reverse CCC CGC CAC TAT TGA TAG AA   GAPDH mRNA 65b  Forward AAC AGC AAC TCC CAT TCT TC   Reverse TGG TCC AGG GTT TCT TAC  www.nature.com/scientificreports/ MSP kit (#R100A; TaKaRa, Otsu, Japan) containing SYBR green (TaKaRa). Specific primers for methylated or unmethylated p11 promoters were designed using MethPrimer 63 . Primers include either a 150 base-pair (methylation-specific) or 155 base-pair (unmethylation-specific) region with 5 CpGs sites in the p11 promoter region (Fig. 5). Primer sequences are listed in Table 2. The p11 promoter region was confirmed to be amplified with methylation-and unmethylation-specific primers ( Figure S2). The qPCR reaction conditions were as follows: initial denaturation at 95 °C for 30 s, followed by 40 cycles of denaturation at 98 °C for 5 s, annealing at 53 °C for 30 s, and extension at 72 °C for 1 min. Ct values were normalized to GAPDH, and differences in methylation and unmethylation between control and MS groups (n = 10-12/group, young adulthood; n = 11-13/group, middle adulthood) were calculated using the ΔCt. Levels of methylated DNA (%) were calculated according to the following formula: methylated rate (%) = ΔCt methylated DNA /(ΔCt methylated DNA + ΔCt unmethylated DNA ) × 100.
Statistical analysis. GraphPad Prim 8.0 (La Jolla, USA) was used for statistical analysis. To determine the main and interaction effects of age and MS, two-way ANOVA was performed. Turkey's multiple-comparison tests were used for post hoc comparisons. P-values < 0.05 were considered to indicate statistical significance, and all data are presented as means ± standard error of the mean (SEM).

Data availability
The data are availability from the corresponding author upon request.