Prenatal Synthetic Glucocorticoid Treatment Changes DNA Methylation States in Male Organ Systems: Multigenerational Effects

Abstract

Prenatal synthetic glucocorticoids (sGC) are administered to pregnant women at risk of delivering preterm, approximately 10% of all pregnancies. Animal studies have demonstrated that offspring exposed to elevated glucocorticoids, either by administration of sGC or as a result of maternal stress, are at increased risk of developing behavioral, endocrine, and metabolic abnormalities. DNA methylation is a covalent modification of DNA that plays a critical role in long-lasting programming of gene expression. Here we tested the hypothesis that prenatal sGC treatment has both acute and long-term effects on DNA methylation states in the fetus and offspring and that these effects extend into a subsequent generation. Pregnant guinea pigs were treated with sGC in late gestation, and methylation analysis by luminometric methylation assay was undertaken in organs from fetuses and offspring across two generations. Expression of genes that modify the epigenetic state were measured by quantitative real-time PCR. Results indicate that there are organ-specific developmental trajectories of methylation in the fetus and newborn. Furthermore, these trajectories are substantially modified by intrauterine exposure to sGC. These sGC-induced changes in DNA methylation remain into adulthood and are evident in the next generation. Furthermore, prenatal sGC exposure alters the expression of several genes encoding proteins that modulate the epigenetic state. Several of these changes are long lasting and are also present in the next generation. These data support the hypothesis that prenatal sGC exposure leads to broad changes in critical components of the epigenetic machinery and that these effects can pass to the next generation.

Fetal plasma glucocorticoid levels increase exponentially in late gestation (1). This glucocorticoid surge promotes maturation of several organ systems, including the lungs. For this reason, synthetic glucocorticoids (sGC) are administered to pregnant women at risk of delivering preterm (2). This treatment is highly effective in decreasing the incidence of respiratory distress syndrome (3). However, preterm labor can be difficult to diagnose, and until recently the administration of multiple courses of maternal sGC had become common practice (4, 5). Animal studies have revealed that repeated exposure to sGC during pregnancy leads to modified hypothalamic-pituitary-adrenal (HPA) axis function and behavior in adult offspring of guinea pigs, rats, mice, sheep, and nonhuman primates (6–9). Furthermore, there is evidence that children prenatally exposed to increased glucocorticoids, either by synthetic administration or as a result of maternal stress, are at risk of emotional and behavioral abnormalities (10, 11).

An important question that emerges from these studies is the mechanism that mediates the long-term impact of a transient exposure to high levels of glucocorticoid during fetal life. DNA methylation is a covalent enzyme-catalyzed modification of methyl moieties to cytosines at critical positions in the genome. During embryogenesis, the DNA methylation pattern is sculpted resulting in cell type-specific distribution of methyl cytosines in the genome (12–18). DNA methylation in critical positions in promoters and regulatory sequences silences genes by several mechanisms including targeting of methylated DNA binding proteins and recruitment of chromatin modification enzymes (19) as well as direct interference with binding of transcription factors (20, 21). Recent data indicate that early experiences can modify DNA methylation resulting in reprogramming of the genome, leading to stable alterations in phenotype (22). Early life adversity results in stable changes in DNA methylation associated with altered behavior in later life in animals and humans (23–26). A recent study has demonstrated that maternal nutrient restriction has significant organ-specific and gestational age-specific effects on global methylation (27). Altered levels of maternal care can also permanently mark the genome resulting in stable life-long changes in gene expression in first-generation (F₁) offspring (28).

Recent studies have shown that altered nutrition and sGC exposure during pregnancy can lead to endocrine, metabolic, cardiovascular, and behavioral modification in a sex-specific manner across multiple generations (29–33). Maternal undernutrition during pregnancy in guinea pigs has transgenerational influences on cardiac morphology and HPA function in second-generation (F₂) offspring (31). Maternal high-fat diet in mice results in an increase in body size of female third-generation offspring, and this effect was passed on only via the paternal lineage (30). In rats, antenatal sGC treatment during the last week of pregnancy leads to glucose intolerance in F₂ offspring (32). However, the mechanisms underlying transgenerational programming in response to sGC exposure are unknown. One study has demonstrated that exposure to stress during early gestation, when epigenetic reprogramming of the male germline occurs, leads to the transmission of a stress-sensitive phenotype in adult F₂ males (29).

Programming of DNA methylation may occur at two levels: global changes in genome methylation or gene-specific changes that target discreet regulatory regions, affecting gene expression. Aberrations at both levels are evident in cancer (34, 35) and autoimmune diseases (36). These changes in DNA methylation must include wider regions than promoter and gene-regulatory sequences because the population of methylated cytosine in promoters of genes and known regulatory regions constitutes only a small fraction of global methylation. Changes in global methylation state likely affect high-level organization of genome function (37). Interestingly, global hypomethylation serves as a prognostic marker in certain cancers (38, 39). Global methylation is thus a well-established indicator of the overall state of the DNA methylation machinery and its long-range consequences on genome function and organization (40–43).

In the present study, we focused on the acute and long-term effect of antenatal sGC treatment on global DNA methylation states in fetal, juvenile and adult offspring as well as the transmission of altered DNA methylation states to the second generation. We hypothesize the following: 1) the natural glucocorticoid surge in late gestation is associated with changes in DNA methylation of the fetal epigenome; 2) that sGC treatment, although not physiologically identical with the natural glucocorticoid surge, can prematurely initiate epigenetic changes in the fetal genome; and 3) early exposure to sGC leads to permanent changes in the epigenome that manifest in F₁ and F₂ generation offspring. In this study, we used a guinea pig model of prenatal exposure to sGC, which we have extensively characterized (44–47).

Materials and Methods

Female guinea pigs (Hartley strain; Charles River Canada, St. Constant, Québec, Canada) were bred in our animal facility as previously described (44). This method provides accurately time-dated pregnant guinea pigs. Food (Guinea Pig Chow 5025; Ralston Purina International, Leis Pet Distributing Inc., Wellesley, Ontario, Canada) and water were available ad libitum. Animals were maintained in a 12-h light, 12-h dark cycle, with lights off at 1900 h. Room temperature was kept at 23 C. Pregnant animals were housed separately during pregnancy and the perinatal period. At all other times, animals were in visual, olfactory and auditory contact with other animals. Weaned offspring were pair housed in clear, polycarbonate cages. All studies were performed according to protocols approved by the Animal Care Committee at the University of Toronto, in accordance with the Canadian Council for Animal Care.

Fetal group

Pregnant guinea pigs were sc injected with betamethasone (BETA; phosphate-acetate mix; Betaject, Sabex, Boucherville, Québec, Canada; 1 mg kg⁻¹) on gestational days (GD) 40 and 41 (period of rapid neurogenesis), 50 and 51 (period of rapid brain growth) or received no treatment (control; no manipulation other than weighing/feeding/changing of cage). The guinea pig glucocorticoid receptor exhibits an approximately 4-fold lower affinity for sGC than the human glucocorticoid receptor (48). To compensate for this, we used a dose of BETA (1 mg kg⁻¹) that is more comparable with the dose received by pregnant women for the management of preterm labor (∼0.25 mg kg⁻¹). Pregnant guinea pigs were euthanized on either GD52 (n = 3–4/group; before the natural cortisol surge) or GD65 (n = 3–4/group; after the cortisol surge), under anesthesia (isoflurane) by decapitation. The liver, right kidney, right adrenal, and placenta were frozen on dry ice. One male fetus was taken from each litter for subsequent analysis to prevent potential litter bias.

F₁ offspring

Pregnant guinea pigs were either injected (sc) with BETA (1 mg kg⁻¹) or vehicle (saline; 0.166 ml kg⁻¹) on GD40 and 41, GD50 and 51, and GD60 and 61, as described previously (33, 49, 50). Animals delivered undisturbed. There was no effect of repeated BETA treatment on gestation length (vehicle, 69.4 ± 0.3 d; BETA, 69.6 ± 0.3 d) or birth weight (33, 50). On postnatal day (PND) 10, a group of male offspring was removed from their mothers and euthanized by decapitation. Liver, kidney, adrenal, and cerebellum were frozen on dry ice (vehicle, n = 4; BETA, n = 4). Another group of male offspring underwent behavioral and endocrine testing and were euthanized at PND90 and processed as described above (vehicle, n = 4; BETA, n = 4).

F₂ offspring

After noninvasive behavioral and endocrine tests (49, 50), adult F₁ female offspring born to mothers treated with sGC or vehicle were bred with control males. After conception, animals were left undisturbed apart from routine cage maintenance, routine weighing, and noninvasive saliva collection. F₂ generation offspring underwent the same noninvasive behavioral and endocrine testing as F₁ adults (49). F₂ adult male guinea pigs (vehicle, n = 3; BETA, n = 3) from separate litters were euthanized at PND90 and tissues were collected as described above.

DNA and RNA extraction

Homogenized tissues were incubated in DNA extraction buffer (500 μl) containing proteinase K (20 μl; 20 mg/ml; Roche, Basel, Switzerland) at 50 C for 12 h. Samples were treated with RNAase A (50 U/mg; 30 min; Roche) and phenol-chloroform (1:1). To precipitate DNA, ethanol [95% (vol/vol)] was added. The pellet was washed and redissolved in buffer [50 μl; Tris-HCL (10 mm) and EDTA (1 mm)]. DNA purity and concentration were assessed using spectrophotometric analysis. DNA integrity was verified using agarose gel [1% (wt/vol)]. Total RNA was extracted from the same tissue using TRIzol, as per the manufacturer's instructions (Invitrogen Canada Inc., Burlington, Ontario, Canada). RNA purity and concentration were assessed using spectrophotometric analysis, and integrity was verified using gel electrophoresis.

Luminometric methylation assay (LUMA)

LUMA is a high-throughput assay used to determine genomic global DNA methylation. The method used in our study is a modification of that described by Karimi et al. (51, 52). LUMA involves the digestion of genomic DNA by a methylation sensitive (HPAII) or insensitive (MSPI) restriction enzymes in combination with an internal control restriction enzyme (EcoRI). The extent of cleavage is determined by a bioluminetric polymerase extension assay based on a four-step pyrosequencing reaction.

DNA (2 μg) was equally divided for two parallel digestion reactions with either HpaII (10 U; New England Biolabs, Ipswich, MA) + EcoRI (10 U; New England Biolabs) or MspI (10 U; New England Biolabs) + EcoRI (10 U; New England Biolabs). Samples were incubated (37 C, 4 h) and then heat inactivated (80 C, 20 min). DNA (15 μl) was mixed with pyrosequencing annealing buffer (15 μl; QIAGEN, Toronto, Ontario, Canada). Samples were transferred to 24-well pyrosequencing plates for sequencing (PyroMark 24; Biotage, Uppsala, Sweden). The nucleotide dispensation order used was based on that reported by Pilsner et al. (53). The peak heights of nucleotide incorporation for C and A represent the HpaII/MspI cuts (methylation) and EcoRI (input DNA), respectively. The formula to calculate percentage genomic methylation is: 1 − [(HpaII/EcoRI)/(MspI/EcoRI)] × 100. All samples were run in triplicate and analysis of each tissue sample was carried out in three independent LUMA analyses.

Quantitative real-time PCR

Reverse transcription was performed using RNA (3 μg) and AMV reverse transcriptase (20 U; Roche), as per the manufacturer's instructions. cDNA (2 μl) was used in a reaction (20 μl) with SYBR green mix (Roche) with forward and reverse primers (0.5 μm). The reaction was performed in a Roche LightCycler LC480 using the following conditions: 95 C for 10 min followed by 45 repeats of 95 C for 10 sec, annealing at the appropriate temperature for 10 sec and extension at 72 C for 10 sec. After the 45th cycle, a final extension step at 72 C for 10 min was performed. The data were quantified and normalized using Roche LightCycler 480 software. The expression of the following mRNA encoding methylation-related proteins were analyzed: DNA methyltransferase 1 (Dnmt1), DNA methyltransferase 3a (Dnmt3a), DNA methyltransferase 3b (Dnmt3b), methyl-CpG binding domain protein 2 (Mbd2), methyl-CpG binding domain protein 3 (Mbd3), methyl CpG binding protein 2 (Mecp2), growth arrest and DNA-damage-inducible-α (Gadd45a), cAMP response element-binding protein binding protein (Crebbp), and tet oncogene 1 (Tet1).

Statistical analysis

All data were expressed as mean ± sem. For all the tests, significance was set at P < 0.05. The statistical analysis was undertaken using Prism (GraphPad Software Inc., San Diego, CA). All data were analyzed using a two-way ANOVA followed by a Bonferroni post hoc test.

Results

Developmental and sGC-induced changes in global DNA methylation states: fetal tissues

Developmental changes in global DNA methylation in the fetal liver, adrenal, kidney, and placenta were assessed before (GD52) and after (GD65) the normal glucocorticoid surge (Fig. 1) (54). The impact of sGC exposure before the endogenous cortisol surge was assessed at GD52 (24 h after final treatment) and GD65 (14 d after final treatment). Two-way ANOVA revealed a significant interaction between treatment and age for fetal liver (P < 0.001; Fig. 1A), adrenal glands (P < 0.0001; Fig. 1B), kidney (P < 0.001; Fig. 1C), and placenta (P < 0.0001; Fig. 1D). Post hoc analysis revealed a significant effect of age on global methylation. There was a significant decrease in global methylation in the adrenal (P < 0.01; Fig. 1B) and placenta (P < 0.001; Fig. 1D) at GD65 compared with GD52. In contrast, there was a significant increase in global methylation in the kidney with advancing gestation (P < 0.01; Fig. 1C). With respect to sGC treatment, post hoc analysis revealed an effect of sGC treatment on global methylation. At GD52, fetal liver (P < 0.05; Fig. 1A), adrenal (P < 0.01; Fig. 1B), and placenta (P < 0.01; Fig. 1D) exhibited a significant decrease in methylation following antenatal sGC treatment. However, sGC treatment resulted in a significant increase in global methylation in the fetal kidney (P < 0.001; Fig. 1C). Analysis at GD65 revealed that global methylation was substantially reduced in the sGC treated placenta (P < 0.001; Fig. 1D) when compared with controls. In contrast, sGC exposure before the natural cortisol surge resulted in significantly increased levels of global methylation in the liver (P < 0.01; Fig. 1A) and adrenal (P < 0.01; Fig. 1B) compared with control at GD65. At GD65, there were no differences in methylation in the kidney between control and sGC exposed fetuses (Fig. 1C).

Fig. 1.

The effect of sGC exposure on global states of DNA methylation in the fetus. Genomic DNA methylation was determined in liver (A), adrenal glands (B), kidney (C), and placenta (D) derived from fetuses that had been exposed to sGC (1 mg kg⁻¹; n = 3; solid bars) on d 40, 41, 50, and 51 of gestation and from controls (n = 3; open bars). The level of global methylation was determined at two time points during gestation (GD52 and GD65) using LUMA. Data are presented as mean ± sem. A significant interaction between treatment and age is represented as ###, P < 0.001. A significant difference between control animals at different ages is represented as ++, P < 0.01 and +++, P < 0.001. A significant difference between control and sGC-treated groups at equivalent age is represented as *, P < 0.05, **, P < 0.01, and ***, P < 0.001.

Impact of development and sGC on expression of DNA methylation-related genes in fetal kidney and placenta

In the fetal kidney, two-way ANOVA revealed a significant interaction between treatment and age for Mbd2 (P < 0.0001; Fig. 2D) and Crebbp (P < 0.05; Fig. 2H). Post hoc analysis revealed a significant effect of age on a number of methylation-related genes. Dnmt3b mRNA (P < 0.001; Fig. 2C) and Tet1 mRNA (P < 0.01; Fig. 2I) expression significantly decreased at GD65 compared with GD52. In contrast, Mbd2 mRNA (P < 0.001; Fig. 2D) significantly increased with advancing gestation. With respect to sGC treatment, post hoc analysis revealed a significant effect of treatment on a number of methylation-related genes. At GD52, sGC treatment significantly increased levels of Mbd2 (P < 0.001; Fig. 2D), Gadd45a mRNA (P < 0.05; Fig. 2G), and Crebbp mRNA (P < 0.05; Fig. 2H) compared with control fetuses, whereas levels of Dnmt3b (P < 0.001; Fig. 2C) and Tetl mRNA (P < 0.01; Fig. 2I) were reduced by sGC exposure at GD52. By GD65, sGC treatment resulted in a reduction in Dnmt3b mRNA (P < 0.01; Fig. 2C) and Mbd2 mRNA (P < 0.01; Fig. 2D) compared with controls.

Fig. 2.

Levels of mRNA for genes involved in regulation of epigenetic states in the fetal kidney after sGC treatment (1 mg kg⁻¹; n = 3; solid bars) and nontreated controls (n = 3; open bars) at GD52 and GD65. Dnmt1 (A), Dnmt3a (B), Dnmt3b (C), Mbd2 (D), Mbd3 (E), Mecp2 (F), Gadd45a (G), Crebbp (H), and Tet1 mRNA (I) are shown. All mRNA expression is relative to Gapdh reference gene. Data are presented as mean ± sem. A significant interaction between treatment and age is represented as #, P < 0.05, and ###, P < 0.001. A significant difference between control animals at different ages is represented as ++, P < 0.01, and +++, P < 0.001. A significant difference between control and sGC treated groups at equivalent age is represented as *, P < 0.05, **, P < 0.01, and ***, P < 0.001.

In the placenta (Fig. 3), two-way ANOVA revealed a significant interaction between treatment and age for Dnmt3a (P < 0.01; Fig. 3B), Dnmt3b (P < 0.01; Fig. 3C), Mbd2 (P < 0.001; Fig. 3D), Mecp2 (P < 0.01; Fig. 3F), Crebbp (P < 0.001; Fig. 3H), and Tet1 mRNA (P < 0.05; Fig. 3I). Post hoc analysis revealed a significant effect of age on expression of methylation related genes. There was a significant decrease in Dnmt3a mRNA (P < 0.01; Fig. 3B) but an increase in Dnmt3b (P < 0.001; Fig. 3C), Mbd3 (P < 0.001 Fig. 3E), and Crebbp mRNA (P < 0.05; Fig. 3H) at GD65 compared with GD52. With respect to sGC treatment, levels of Dnmt1 (P < 0.001; Fig. 3A) and Mecp2 (P < 0.01; Fig. 3F) mRNA were significantly increased at GD52. By GD65, antenatal sGC exposure resulted in an increase in Dnmt1 mRNA (P < 0.01; Fig. 3A) and Mbd2 mRNA (P < 0.001; Fig 3D;). In contrast, sGC treatment resulted in significantly decreased mRNA levels for Dnmt3b (P < 0.01; Fig. 3C;) and Crebbp (P < 0.05; Fig. 3H) when compared with controls.

Fig. 3.

Levels of mRNA for genes involved in regulation of epigenetic states in the placenta following sGC treatment (1 mg kg⁻¹; n = 3; solid bars) and nontreated control (n = 3; open bars) at GD52 and GD65. Dnmt1 (A), Dnmt3a (B), Dnmt3b (C), Mbd2 (D), Mbd3 (E), Mecp2 (F), Gadd45a (G), Crebbp (H), and Tet1 mRNA (I) are shown. All mRNA expression is relative to Gapdh reference gene. Data are presented as mean ± sem. A significant interaction between treatment and age is represented as #, P < 0.05, ##, P < 0.01, and ###, P < 0.001. A significant difference between control animals at different ages is represented as +, P < 0.05, ++, P < 0.01, and +++, P < 0.001. A significant difference between control and sGC-treated groups at equivalent age is represented as *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Effect of sGC exposure on global DNA methylation in F₁ and F₂ offspring

The long-term impact of sGC treatment on global methylation was investigated in F₁ juvenile (PND10) and adult males (F₁) and F₂ adult male offspring (Fig. 4). Two-way ANOVA revealed a significant interaction between treatment and age (F₁, PND10 vs. F₁, adult) for the kidney (P < 0.0001; Fig. 4C). Post hoc analysis of juvenile F₁ kidney revealed a significant increase in global methylation after antenatal sGC exposure (P < 0.05; Fig. 4C). Although there was no significant interaction observed in the adrenal and cerebellum, there were significant independent effects of treatment and age observed in both the adrenal (P < 0.01; P < 0.05; Fig. 4B) and cerebellum (P < 0.01; P < 0.001; Fig. 4D) and a significant effect of treatment in the liver (P < 0.01; Fig. 4A). In adult F₁ offspring, there was a significant reduction in global methylation in the liver (P < 0.01; Fig. 4A), adrenal (P < 0.001; Fig. 4B), kidney (P < 0.001; Fig. 4C), and cerebellum (P < 0.01; Fig. 4D) in animals exposed to antenatal sGC treatment compared with controls.

Fig. 4.

The effect of sGC exposure in the F₀ pregnancy on global DNA methylation in F₁ juvenile (n = 4) and adult (n = 4) offspring and F₂ adult offspring (n = 3). Genomic DNA methylation was determined in liver (A), adrenal glands (B), kidney (C), and placenta (D) derived from offspring (F₁ and F₂) of F₀ animals that had been exposed on d 40, 41, 50, 51, 60, and 61 of pregnancy to sGC (1 mg kg⁻¹; solid bars) or vehicle (saline; open bars). Data are presented as mean ± sem. A significant interaction is represented as ##, P < 0.01, and ###, P < 0.001. A significant difference between control and sGC-treated groups at equivalent age is represented as *, P < 0.05, **, P < 0.01, and ***, P < 0.001. A significant difference between F₁ PND10 and F₁ adult control animals is represented as +, P < 0.05, and +++, P < 0.001. A significant difference between F₁ adult and F₂ adult control animals at equivalent age is represented as §, P < 0.05, and §§§, P < 0.001.

Intergenerational effects

Two-way ANOVA also revealed a significant interaction between treatment and generation (F₁ adult vs. F₂ adult) for the liver (P < 0.001; Fig. 4A), adrenal (P < 0.001; Fig. 4B), and kidney (P < 0.01; Fig. 4C). There was no significant interaction between treatment and generation observed in cerebellum; however, there was a significant independent effect of treatment (P < 0.01; Fig. 4D). In F₂ male offspring, post hoc analysis revealed that antenatal sGC exposure significantly decreased global methylation in all organs analyzed (P < 0.01; Fig. 4). Further analysis revealed small but significant increases in DNA methylation in the liver (P < 0.05; Fig. 4A) and adrenal (P < 0.001; Fig. 4B) of control F₂ adult animals compared with F₁ offspring.

Antenatal sGC treatment and expression of DNA methylation-related genes in F₁ and F₂ offspring: cerebellum and kidney

Kidney

Two-way ANOVA revealed a significant interaction between treatment and age (F₁ PND10 vs. F₁ adult) for Dnmt1 (P < 0.01; Fig. 5A), Dnmt3b (P < 0.0001; Fig. 5C), Mbd2 (P < 0.0001; Fig. 5D), Mbd3 (P < 0.05; Fig. 5E), Mecp2 (P < 0.0001; Fig. 5F), Gadd45a (P < 0.01; Fig. 5G), Crebbp (P < 0.0001; Fig. 5H), and Tet1 mRNA (P < 0.01; Fig. 5I). In juvenile animals (PND10), prenatal sGC treatment resulted in a significant decrease in Dnmt1 (P < 0.01; Fig. 5A), Mbd2 (P < 0.001; Fig. 5D), Mbd3 (P < 0.01; Fig. 5E), and Tet1 mRNA (P < 0.05; Fig. 5I) in the kidney compared with controls. In contrast, prenatal sGC exposure led to significant increases in Dnmt3b (P < 0.001; Fig. 5C) and Mecp2 mRNA levels (P < 0.001; Fig. 5F). In adult F₁ offspring, antenatal sGC treatment resulted in significantly reduced Dnmt3a (P < 0.05 Fig. 5B) and Dnmt3b mRNA levels (P < 0.01; Fig. 5C) but elevated Mbd2 (P < 0.001; Fig. 5D), Gadd45a (P < 0.01; Fig. 5G), and Crebbp mRNA levels (P < 0.001; Fig. 5H), compared with vehicle-treated controls. With respect to age, levels of Dmnt1 (P < 0.01; Fig. 5A) and Mbd2 mRNA (P < 0.01; Fig. 5D) were lower, whereas Dmnt3a (P < 0.05; Fig. 5B) and Dmnt3b mRNA levels (P < 0.01; Fig. 5C) were higher with advancing age.

Fig. 5.

The effect of sGC exposure in the F₀ pregnancy on mRNA expression of genes involved in regulation of epigenetic states in the kidney of F₁ juvenile (n = 4) and adult (n = 4) offspring and F₂ adult offspring (n = 3). Dnmt1 (A), Dnmt3a (B), Dnmt3b (C), Mbd2 (D), Mbd3 (E), Mecp2 (F), Gadd45a (G), Crebbp (H), and Tet1 mRNA (I) were measured in offspring (F₁ and F₂) of F₀ animals that had been exposed to sGC (1 mg kg⁻¹; solid bars) or vehicle (saline; open bars) on d 40, 41, 50, 51, 60, and 61 of pregnancy. All mRNA expression is relative to Gapdh reference gene. Data are presented as mean ± sem. A significant interaction between treatment and age/generation is represented as #, P < 0.05, ##, P < 0.01, and ###, P < 0.001. A significant difference between control and sGC-treated groups at equivalent age is represented as *, P < 0.05, **, P < 0.01, and ***, P < 0.001. A significant difference between F₁ PND10 and F₁ adult control animals is represented as +, P < 0.05, and ++, P < 0.01). A significant difference between F₁ adult and F₂ adult control animals at equivalent age is represented as §, P < 0.05.

Intergenerational effects

Two-way ANOVA revealed a significant interaction between sGC treatment and generations (F₁ adult vs. F₂ adult) for Mbd3 (P < 0.005; Fig. 5E), Mecp2 (P < 0.05; Fig. 5F), Crebbp (P < 0.0001; Fig. 5H), and Tet1 mRNA (P < 0.05; Fig. 5I). In the adult F₂ generation, sGC treatment reduced Dnmt3a (P < 0.01; Fig. 5B) and Dnmt3b mRNA levels (P < 0.05; Fig. 5C) but increased Mbd2 (P < 0.001; Fig. 5D), Gadd45a (P < 0.01; Fig. 5G), and Tet1 mRNA (P < 0.01; Fig. 5I) levels. Further post hoc analysis revealed that Dnmt3b mRNA levels were lower in the F₂ generation compared with the F₁ generation (P < 0.05; Fig. 5C).

Cerebellum

Two-way ANOVA revealed a significant interaction between treatment and age (F₁ PND10 vs. F₁ adult) for Dnmt1 (P < 0.01; Fig. 6A), Mbd2 (P < 0.0001; Fig. 6D), Mbd3 (P < 0.05; Fig. 6E), and Mecp2 mRNA (P < 0.001; Fig. 6F). In juvenile animals, prenatal sGC treatment resulted in a significant increase in Dnmt3a (P < 0.001; Fig. 6B) and Dnmt3b (P < 0.01; Fig. 6C) in the cerebellum. In adult F₁ animals, sGC treatment resulted in a significant increase in Dnmt1 (P < 0.01; Fig. 6A), Dnmt3a (P < 0.001; Fig. 6b), Mbd2 (P < 0.001; Fig. 6D), Mecp2 (P < 0.01; Fig. 6F), Crebbp (P < 0.05; Fig. 6H), and Tet1 mRNA levels (P < 0.01; Fig. 6I). With respect to age, Dmnt3a (P < 0.05; Fig. 6B), Mbd2 (P < 0.01; Fig. 6D), Mbd3 (P < 0.01; Fig. 6E), Mecp2 (P < 0.001; Fig. 6F), and Gadd45a mRNA levels (P < 0.01; Fig. 6G), decreased with age.

Fig. 6.

The effect of sGC exposure in the F₀ pregnancy on mRNA expression of genes involved in regulation of epigenetic states in the cerebellum of F₁ juvenile (n = 4) and adult (n = 4) offspring and F₂ adult offspring (n = 3). Dnmt1 (A), Dnmt3a (B), Dnmt3b (C), Mbd2 (D), Mbd3 (E), Mecp2 (F), Gadd45a (G), Crebbp (H), and Tet1 mRNA (I) were measured in offspring (F₁ and F₂) of F₀ animals that had been exposed to sGC (1 mg kg⁻¹; solid bars) or vehicle (saline; open bars) on days 40, 41, 50, 51, 60, and 61 of pregnancy. All mRNA expression is relative to Gapdh reference gene. Data are presented as mean ± sem. A significant interaction between treatment and age/generation is represented as #, P < 0.05, ##, P < 0.01, and ###, P < 0.001. A significant difference between control and sGC-treated groups at equivalent age is represented as *, P < 0.05, **, P < 0.01, and ***, P < 0.001. A significant difference between F₁ PND10 and F₁ adult control animals is represented as +, P < 0.05, ++, P < 0.01, and +++, P < 0.001. A significant difference between F₁ adult and F₂ adult control animals at equivalent age is represented as §§§, P < 0.001.

Intergenerational effects

We then compared the effects of sGC treatment on expression of methylation-related genes in the cerebellum between adult F₁ and F₂ offspring. Two-way ANOVA revealed a significant interaction between treatment and generation (F₁ adult vs. F₂ adult) for Dnmt3a (P < 0.0001; Fig. 6B), Mbd2 (P < 0.0001; Fig. 6D), and Mecp2 mRNA (P < 0.0001; Fig. 6F). In the adult F₂ generation, sGC treatment resulted in increases in the mRNA levels of Dnmt1 (P < 0.01; Fig. 6A), Dnmt3a (P < 0.05; Fig. 6B), Mbd2 (P < 0.001; Fig. 6D), Mecp2 (P < 0.001; Fig. 6F), and Tet1 (P < 0.01; Fig. 6I). Further post hoc analysis revealed that Mecp2 (P < 0.001; Fig. 6F) mRNA levels were higher in the control F₂ offspring compared with the F₁ offspring.

Correlation of global methylation and expression of genes involved in epigenetic regulation

A correlation analysis was performed using data from juvenile and adult (F₁ and F₂) kidney and cerebellum LUMA (DNA methylation) and quantitative RT-PCR (mRNA expression) to determine which of the changes in mRNA expression correlated with changes in global methylation. In the kidney, there was a significant correlation between global methylation and the expression of Dnmt3b (R = 0.5644, P < 0.01; Fig. 7C), Mbd2 (R = −0.8774, P < 0.001; Fig. 7D), Mecp2 (R = 0.6188, P < 0.01; Fig. 7F), Gadd45a (R = −0.8193, P < 0.001; Fig. 7G), and Crebbp (R = −0.7958, P < 0.001; Fig. 7H). In the cerebellum, global methylation is correlated with mRNA expression of Dnmt1 (R = −0.5194, P < 0.01; Fig. 8A), Dnmt3a (R = −0.4502, P < 0.05: Fig. 8B), Mbd2 (R = −0.5985, P < 0.01; Fig 8D), Mecp2 (R = −0.5018, P < 0.05; Fig. 8F), Crebbp (R = −0.4993, P < 0.05; Fig. 8H), and Tet1 (R = −0.6352, P < 0.01; Fig 8I). Mbd2 is the only mRNAs analyzed whose expression shows high correlation with reduced methylation across all samples and all ages. This supports an important role for MBD2 in the loss of DNA methylation.

Fig. 7.

Correlation analysis of global DNA methylation with mRNA expression of genes involved in epigenetic regulation in the kidney. A correlation analysis was performed between global DNA methylation and mRNA expression of genes that are implicated in the regulation of epigenetic state. The correlation analysis was determined using global methylation and mRNA expression data from the F₁ PND10 animals (n = 4), F₁ adult animals (n = 4), and F₂ adult animals (n = 3). Correlation is shown for Dnmt1 (A), Dnmt3a (B), Dnmt3b (c), Mbd2 (D), Mbd3 (E), Mecp2 (F), Gadd45a (G), Crebbp (H), and Tet1 mRNA (I). The R value and P value for each gene analyzed is presented in the figure.

Fig. 8.

Correlation analysis of global DNA methylation with mRNA expression of genes involved in epigenetic regulation in the cerebellum. A correlation analysis was performed between global DNA methylation and mRNA expression of genes that are implicated in the regulation of epigenetic state. The correlation analysis was determined using global methylation and mRNA expression data from the F₁ PND10 animals (n = 4), F₁ adult animals (n = 4), and F₂ adult animals (n = 3). Correlation is shown for Dnmt1 (A), Dnmt3a (B), Dnmt3b (C), Mbd2 (D), Mbd3 (E), Mecp2 (F), Gadd45a (G), Crebbp (H), and Tet1 mRNA (I). The R value and P value for each gene analyzed is presented in the figure.

Discussion

Fetal plasma glucocorticoid levels increase exponentially in late gestation (54). In the present study, we have shown, for the first time, that this glucocorticoid surge is associated with very substantial changes in global methylation in a number of organ systems. The trajectory of these developmental changes in DNA methylation states varies in different tissues and is associated with substantial changes in the expression of genes involved in the methylation process. We have also shown that prenatal exposure to sGC before the natural glucocorticoid surge has a major impact on DNA methylation in the fetus and neonate, and again this is associated with altered expression of genes involved in methylation. Of critical importance, the effects of prenatal sGC persist into adulthood and are present in the next generation. Together these data provide strong evidence that glucocorticoids are an important developmental trigger and that their effects are mediated, at least in part, by the level of DNA methylation in the genome.

Global methylation is higher in the adrenal and placenta before the natural glucocorticoid surge (GD52) compared with after the surge (GD65). Premature fetal exposure to glucocorticoid through maternal sGC treatment leads to a reduction in DNA methylation in these organs, together with the liver. In contrast, DNA methylation in the kidney is increased between GD52 and GD65, and this increase in DNA methylation can be induced prematurely by maternal sGC treatment. This would suggest that glucocorticoids provide an important signal or trigger in maturation of the fetal epigenome and that these effects are highly tissue specific. This is consistent with the fact that the glucocorticoid surge is essential for normal organ maturation (1). However, the situation in the adrenal is more difficult to interpret because sGC exposure inhibits production of physiological factors that are important in driving normal adrenal development. Clearly, further studies are required to determine the mechanism by which sGC modulate epigenetic changes in specific organs. To our knowledge, this is the first study to consider global methylation in the guinea pig. Lower levels of methylation were present in the guinea pig placenta (∼50%) compared with other somatic tissues, and this difference was greatest at the end of gestation. A previous report has indicated a similar level of global DNA methylation in the mouse (55). However, overall, the level of global methylation in other somatic tissues in the guinea pig appeared to be lower than has been reported in the mouse, suggesting species-specific differences in global methylation patterns.

To our knowledge, only one previous study has considered the effect of fetal exposure to glucocorticoid on specific gene methylation profiles in the fetus, and no studies have considered their impact on organ-specific changes in global methylation in the fetus. This previous study showed glucocorticoids to cause permanent demethylation of a key enhancer of the rat liver-specific tyrosine aminotransferase (Tat) gene (56). This demethylation resulted in enhanced transcription factor binding, which was maintained after glucocorticoid withdrawal, indicating stability of the effect (56). This study does not identify the specific regions of DNA in which methylation has changed, and it is therefore not possible to determine whether the nature of the changes in global methylation identified before and after the natural cortisol surge are the same as those induced by fetal exposure to sGC. In the fetal liver and adrenal, there is an interesting switch in the acute (GD52) and longer-term (GD65) effects of sGC on global methylation such that sGC treatment (before GD52) results in elevated DNA methylation in both organs at GD65 compared with nontreated animals. The mechanisms that underlie this longer-term rebound in global methylation remain to be determined.

There were very profound long-term effects of antenatal sGC exposure on global methylation in offspring. In juvenile offspring (PND10), prenatal sGC exposure resulted in a reduction in global methylation in the liver and cerebellum with a similar strong trend in the adrenal. In the liver and adrenal, this pattern of reduced DNA methylation was identical with the acute effects we had identified in the fetus. These reductions were maintained at a significant level into adulthood. Interestingly, in the juvenile kidney, we observed hypermethylation that was similar to the acute effects in the fetus. However, by adulthood, the effect had reversed and there was almost a 20% reduction in global methylation. These data indicate that prenatal exposure to sGC has life-long effects on organ methylation resulting in profound demethylation in adulthood. It has been shown that prenatal exposure to sGC can lead to life-long changes in hepatic, renal, and adrenal function as well as behaviors (49, 57–60). A number of these studies have identified striking differences in gene expression within these organ systems (49, 57), and limited studies have linked altered gene expression to epigenetic modification of specific gene promoters (56). The present study indicates that these changes in function (and gene expression) might be driven by global changes in organ DNA methylation rather than discrete modifications at specific gene promoters.

In the present study, the changes in DNA methylation are in the range of 5–10% of CCGG, and this represents 1 million to 2 million CG in the genome. This far exceeds the number of CG present within glucocorticoid responsive regulatory sequences. Global changes in DNA methylation likely act at a different level of organization than site-specific methylation events. However, given the fact that the natural glucocorticoid surge near term leads to robust and selective changes in the expression of many genes (1), it is likely that glucocorticoids initiate effects both at the level of global methylation and selectively at the level of specific gene promoters. It has been proposed that changes in global methylation result in long-range effects on genome function (61). Interestingly, the changes in DNA methylation states seen in offspring exposed to sGC; global hypomethylation and deregulated expression of mbd2 and dnmt mRNA are hallmarks of cancer (62). Because this state is maintained in adult organs and is transmitted to the next generation, future studies should be undertaken to elucidate whether sGC exposure is associated with increased risk of tumorigenesis.

The fact that antenatal sGC treatment results in altered DNA methylation in adult F₂ offspring strongly supports the hypothesis of transgenerational transmission of epigenetic states. In this connection, emerging studies are demonstrating that both phenotypic and DNA methylation differences can be transmitted across generations. For example, exposure of pregnant rats to the endocrine disruptors vinclozolin or methoxychlor caused decreased spermatogenic capacity in the F₁ generation that was transferred through the male germline to the fourth generation (17). In another study, maternal high-fat diet was shown to affect third-generation female body size via the paternal lineage, again suggesting the possibility of epigenetic inheritance (30). Together these data suggest that some DNA methylation changes, induced by paternal or maternal exposures, exhibit meiotic stability and can be transmitted across generations (17, 30). In this study, we also identified a small but significant increase in global methylation in the liver and adrenal in the F₂ compared with the F₁ controls; this did not occur in the kidney or cerebellum. F₁ and F₂ animals were reared and housed under identical conditions in our animal facility; however, F₀ animals were obtained as adults from our supplier. The reason for the upward drift in methylation across generations is unclear. One possibility is that it may result from differences between husbandry in our facility compared with that of the supplier's colony. Furthermore, studies are required to investigate such a possibility.

Antenatal sGC exposure affects the general state of DNA methylation, suggesting that sGC have an effect on the machinery that either determines and/or maintains the DNA methylation pattern. In this regard, sGC exposure alters the expression of several genes that encode proteins that are involved in defining epigenetic states. Remarkably, changes in Mbd2, Crebbp, and Dnmt3a expression in the kidney and cerebellum are maintained in adult F₁ offspring and are present in the F₂ generation. Interestingly, a strong inverse correlation existed between the level of global DNA methylation and level of expression of Mbd2 and Crebbp in both kidney and cerebellum, suggesting that expression of these two genes is strongly associated with reduced levels of DNA methylation. It is tempting to speculate that the modification of these genes by prenatal sGC exposure of F₀ mothers mediates the long-term effects on DNA methylation states that cross the generations. In this regard, Mbd2 was previously shown to be associated with DNA demethylation in prostate and breast cancer cell lines (63, 64). Although it is difficult to associate specific changes in expression of Dnmt with changes in DNA methylation states, the data further support the hypothesis that sGC have broad impact on DNA methylation states. Further studies are clearly warranted to delineate the specific genes and gene networks that are affected by sGC and whether these changes are observed and maintained into the next generation.

The present data indicate that the impact of sGC exposure (F₀ pregnancy) crosses maternally from the F₁ to the F₂ generation. However, these data are not sufficient to prove a germline, trans-generational epigenetic inheritance. First, the F₂ generation developed from an F₁ germline that was itself exposed in utero to sGC. Second, development of the F₂ fetus took place in the F₁ uterine environment that may have been impacted by the antenatal sGC exposure. Furthermore, adult female guinea pigs that were exposed to sGC as fetuses exhibit altered HPA function as adults, which will likely lead to altered fetal exposure to cortisol during pregnancy (49). Given the strong influences of glucocorticoids on the epigenome, it is likely that this provides an additional mechanism for transgenerational epigenetic effects. However, our data raise the provocative possibility that the memory of sGC exposure in DNA methylation states can be transmitted through the germline. Further studies are required to investigate this possibility.

In conclusion, we have shown, for the first time, that the natural glucocorticoid surge is associated with very substantial changes in global methylation. The trajectory of these developmental changes in DNA methylation states varies in different tissues and is associated with substantial changes in the expression of methylation-related genes. Prenatal exposure to sGC before the natural glucocorticoid surge has a major impact on global DNA methylation and associated methylation-related machinery in the fetus and neonate, and, of critical importance, the effects persist into adulthood and are passed to the next generation. Together these data provide strong evidence that glucocorticoids represent an important developmental trigger and that their effects are mediated at the level of DNA methylation. Furthermore, changes in DNA methylation resulting from prenatal sGC exposure may underlie a number of metabolic, endocrine, and behavioral phenotypes that have been identified after such exposure in F₁, F₂, and third-generation generation offspring. Given that approximately 10% of all pregnant women receive sGC in late gestation to mature the fetal lung, the present study has substantial relevance to clinical practice.