Introduction

Antenatal corticosteroids (ACS), including betamethasone or dexamethasone, are administered to pregnant persons to promote fetal lung maturation and improve perinatal survival when there is risk of preterm delivery. Despite established benefits during the perinatal period, ACS exposure influences development in multiple tissue types, leading to long-term effects on cardiometabolic, immune and neurodevelopmental outcomes in offspring [1,2,3].

In humans, ACS exposure has been associated with a heightened stress response in children (6–11 years) [4] and adolescents (14–18) [5]. Effects were greater in female as compared to male children [4]. Altered neurobehavioural outcomes, such as decreased attention, have also been reported in children (8 years) [6] and young adults (14–26 years) [7], who had been exposed to ACS. Furthermore, a recent population-based study identified that ACS exposure increased risk of mental or behavioural disorders at 5 years [8], and that mental and behavioural disorders were more prevalent among ACS exposed children (12.01%) as compared to unexposed children (6.45%) [9]. Exposure to multiple courses, as compared to a single course of ACS, resulted in a reduced head circumference at birth [10] and increased risk of neurodevelopmental and neurosensory deficits at 5 years [11, 12]. Imaging studies have identified smaller and a thinner anterior cingulate cortex in term-born children who received ACS [13]. However, investigations into the molecular mechanisms that underlie these changes in the brain are limited in humans, with studies relying on animal models or molecular insight from peripheral tissue such as blood.

In the juvenile baboon, ACS exposure has been shown to result in reduced learning ability [14], while juvenile guinea pig offspring demonstrated increased locomotor activity following ACS [15]. Notably, behavioural effects were greatest in female offspring, suggesting a higher sensitivity of females to ACS exposure. In juvenile female guinea pig offspring, ACS exposure resulted in significant changes to DNA methylation patterns in the hippocampus, in gene pathways involved in neurodevelopment. In these same animals, significant methylation changes were also identified in the whole blood, suggesting that the whole blood represent a useful surrogate tissue for examining methylation changes in the brain. Indeed, studies have shown that blood methylation status is highly correlated (r = 0.86–0.89) to both post-mortem and surgically resected brain tissue in humans, though correlations were dependent on the brain region examined [16, 17].

More recently, we reported that exposure to a single course of ACS results in significant changes to the DNA methylation pattern in the prefrontal cortex (PFC) of newborn and juvenile female guinea pig offspring. Identified genes were part of networks involved in ‘developmental cellular process’ and ‘neuronal proliferation’ in the newborn PFC, and ‘synaptic regulation’ in the PFC at post-natal day 14 [18]. Furthermore, mRNA levels of key differentially methylated genes identified in the gene networks were found to be altered in the PFC, suggesting a functional correlation of the methylation data. Interestingly, altered DNA methylation in the PFC occurred at multiple transcription factor binding sites, indicating a potential functional relevance of the methylation changes. Specifically, four key binding sites, for transcription factors PLAGL1, TFAP2C, ZNF263, and SP1 were found to be differentially methylated in the newborn and juvenile PFC, presenting a signature consistently observed at multiple developmental stages in response to ACS. It is not known whether this methylation signature associated with ACS exposure is also present in surrogate tissues such as the whole blood.

In the present study, we investigated the differential methylation patterns in the peripheral whole blood of newborn guinea pig offspring to determine whether a consistent signature is left by ACS exposure in the blood as was left in the PFC. Furthermore, we re-examined DNA methylation data from human neonatal blood following ACS exposure, where we had reported 505 differentially methylated sites following exposure to ACS [19]. This previous human study identified gene networks in blood that linked to neuronal proliferation, synaptic development, proteasome activity and intracellular protein trafficking, indicating comparable pathways to those identified in the guinea pig PFC. As such, we sought to examine the differential DNA methylation pattern in the blood of newborn guinea pigs following ACS, and whether the methylation changes in the blood of guinea pigs and humans occur at the same transcription factor binding sites that were previously identified in the PFC [18].

Methods

Animal cohort development

Female Dunkin-Hartley guinea pigs (Charles River) were singly-housed and mated while food and water were available ad libitum on a 12 h light-dark cycle under approval by the Animal Care Committee at the University of Toronto. Pregnant dams were randomly assigned into treatment or control groups. Dams were subcutaneously injected with a clinically relevant single-course (two doses, provided 24 h apart) of betamethasone (1 mg/kg; n = 7) or saline (Veh: equi-volume; n = 7) on gestational days (GD) 50/51. The guinea pig glucocorticoid receptor (GR) has a four-fold lower affinity for synthetic glucocorticoids compared to the human GR [20], as such the dose used in this study (1 mg/kg) is equivalent to that used clinically in humans (12 mg intramuscularly 24 h apart) [21]. Whole blood was collected from PND1 offspring by cardiac puncture under isoflurane anaesthesia. One female offspring per litter was used for DNA methylation analysis, as we have previously shown that female offspring demonstrate greater behavioural and neuroendocrine phenotypes after antenatal corticosteroid treatment compared to males [3].

Sample preparation

Whole blood without EDTA was used to create blood spots (70 µL per spot) and dried at room temperature for 4 h using FTA Blood Stain Cards (Whatman, Sigma). Blood spot cards were stored with silica gel at −80 °C until use. Dried blood spots (DBS) provide a stable medium for DNA that can be easily transported and stored for extended periods without altering DNA methylation [22, 23]. It is not possible to get fresh whole blood from human neonates, however, blood spot cards are routinely collected for newborn screening and the Ontario Birth Study allows additional blood spots (1–5 per neonate) to be collected for research. By producing blood spot cards in the guinea pig cohort, we are assessing the same sample medium in both guinea pigs and humans. We have also previously reported that there are no significant differences in DNA methylation derived from DBS compared to frozen whole blood [23], indicating that dried blood spot cards are a reliable resource for the assessment of blood methylation patterns. To extract gDNA, DBS were cut into small pieces using sterile scissors and processed using the Quick-DNA™ Microprep Plus Kit (Zymo Research), according to the manufacturer’s instructions. gDNA was quantified with Quant-iT Picogreen dsDNA assay (ThermoFisher, Waltham, MA, USA) and quality was assessed using TapeStation (Agilent, Santa Clara, CA, USA) at the Centre for Applied Genomics (Peter Gilgan Centre for Research and Learning, Hospital for Sick Children, Toronto, Canada). Reduced Representation Bisulfite Sequencing (RRBS) libraries were prepared from 100 ng of high-quality dsDNA (DNA integrity number >6) using the Ovation RRBS Methyl-Seq System 1–16 (Tecan Genomics, Redwood City, CA, USA) and EpiTect Fast DNA Bisulfite kit (Qiagen, Germany) following the manufacturer’s protocols. RRBS libraries were sequenced using Illumina’s NextSeq500 platform at the Donnelly Sequencing Centre (University of Toronto, Canada), using single-end reads of 75 bp read lengths as per manufacturer protocol. Samples were processed in coded lots and sequenced in pooled multiplexes of 10, balanced for treatment condition to not introduce group-biased batch effects during sample processing and sequencing.

Bioinformatic analysis

Identification of differentially methylated sites and genes & pathway enrichment

Quality control of sequenced data was performed using Trim Galore (v0.6.4), a wrapper script around CutAdapt [24], to remove low-quality reads (Phred quality scores <30) and accessory sequences from adaptors. A Python script provided by NuGEN (Github: trimRRBSdiversityAdaptCustomers.py) removed reads without an MspI site signature at the 5’ end. Bismark (v0.16.0) [25] and Bowtie2 (v2.3.4.3) [26] were used to align the filtered and trimmed reads to the guinea pig genome (CavPor3.0) and Samtools (v1.9) [27] was used to sort the reads. Differentially methylated CpG sites (DMC) were identified using MethPipe [28, 29], to compare between ACS-treated and untreated samples. methpipe/methcounts was used to obtain methylation levels at individual cytosines, which was used to perform beta-binomial regression analyses in the methpipe/radmeth programme under the regression option. Identified CpG sites were adjusted (bins 1:200:1) using the adjust option in methpipe/radmeth to obtain a false discovery rate (FDR)-corrected list of significant DMCs. All bioinformatic analysis was performed blind until the final step when treatment groups had to be specified for regression analysis.

Analysis was limited to DMCs with greater than 10 reads, showing greater than 5% methylation difference, and FDR < 0.05. DMCs were annotated using CompEpiTools [30] to known genes based on the Ensembl database. DMCs were localized to genomic features (promoter, intragenic–including both introns and exons, or intergenic regions), based on distance from a transcription start site (TSS). Promoter regions were defined as DMCs within 1000 bps up/downstream of a TSS, while DMCs further than 1000 bps were categorized as intergenic regions. Computations were performed on the Niagara supercomputer at the SciNet HPC Consortium [31].

Gene ontology analysis

DMCs were annotated to known genes to understand the functional implications of the altered methylome. Gene ontology (GO) analysis was performed on g:Profiler [32] to determine gene networks, or groups of genes with similar function, that were differentially methylated in response to ACS exposure. An unranked list of genes, independent of the number of CpG sites or the average % methylation difference, was used for analysis. Searches were performed against the guinea pig genomic database using ‘All known genes’ as the statistical domain scope, which is more extensively documented.

Transcription factor motif search

To determine if DMCs were localized within known transcription factor binding sites (TFBS), a motif enrichment search was performed using MEME [33]. Briefly, the dataset of DMCs were expanded by 100 bps (50 bps up/downstream) using BEDTools ‘slop’ [34] and converted into a. bed file to use as input into MEME Suite. Parameters were specified for motif distribution (any number of repetitions) to indicate that any number of non-overlapping motifs may be present per sequence, and as ‘1st order model of sequences’ for the background model to adjust for dimer biases (such as high GC content). Default significance threshold was set at E-value ≤ 0.05. Identified motifs were inputted to TOMTOM [35] to identify transcription factors binding sequences that align to the motifs. Significance was defined at a default threshold of E-value ≤ 10. Targeted searches of known TFBS motifs were also performed with position-weight-matrices (PWM) obtained from JASPAR [36] using FIMO [37] on Meme-Suite.

Human blood methylation data

Differentially methylated sites in the human neonatal blood following ACS exposure were assessed in our previous study [19]. Data were re-examined as part of this study for transcription factor binding site analyses. This was not undertaken in our original publication [19]. The initial dataset was collected from infants born at term and exposed to ACS who were retrospectively identified through the Ontario Birth Study (OBS) conducted at Sinai Health Systems as previously described [19]. Briefly, identified case subjects (n = 14, 8 M, 6 F) were matched to controls based on maternal age, BMI, parity, and fetal sex. Neonatal blood from heel pricks was collected at 24 h post-birth on blood spot cards, as part of routine clinical practice for diagnostic purposes. gDNA from blood spot cards was extracted using the GenSolve DNA COMPLETE Kit (GenTegra, Pleasanton, CA, USA) and prepared and sequenced using Reduced Representation Bisulfite Sequencing (RRBS) as described above. Bioinformatic analyses were performed using the above-mentioned pipeline as previously described [19]. Gene alignment was performed against the human genome (hg38). Identified DMCs were expanded by 100 bps using Bedtools > ‘slop’ and assessed using MEME [33] and TOMTOM [35] to identify transcription factors binding sites that significantly aligned with the identified motif sequences.

Results

Differentially methylated CpG sites in PND1 blood following ACS exposure

Differential methylation in the PND1 guinea pig blood was examined following ACS exposure. No variation in methylation was observed between offspring. In the PND1 guinea pig blood, 1677 DMCs were identified following ACS exposure (10x coverage, FDR < 0.05, ≥5% methylation difference), of which 956 sites (57%) were hypermethylated and 721 sites (43%) were hypomethylated (Fig. 1A and Supplementary Table 1). Categorizing DMCs to genomic features, 73 sites were localized to the promoter region, while 693 sites were found within intragenic regions (within gene body) and 911 sites were at intergenic regions (Fig. 1B). The DMCs annotated to 171 genes, 108 of which were hypermethylated (63%) and 63 (37%) of which were hypomethylated (Fig. 1C). The top 20 differentially methylated genes are presented in Table 1.

Fig. 1: Overview of differentially methylated CpG sites (DMC) in the blood of newborn guinea pigs exposed to antenatal betamethasone.
figure 1

A Scatterplot of individual CpG sites that were significantly differentially methylated in ACS-exposed animals as compared to unexposed controls. 956 sites were hypermethylated, 721 sites were hypomethylated. >10× reads; ≥5% methylation difference; FDR ≤ 0.05 (B) DMCs were localized to various genomic features, based on distance from transcription start sites. Promoter: ≤1000 bp, Intergenic: ≥1000 bp, Intragenic: within gene body, including introns and exons (C) DMCs were annotated to known genes. 171 genes were identified in total, of which 108 genes were hypermethylated and 63 genes were hypomethylated. D The top 20 terms from the gene set enrichment analysis are represented from most to least significant. All differentially methylated genes discovered in the PND1 Blood (171 genes) were used in the analysis.

Full size image
Table 1 Top 20 differentially methylated genes in the whole blood of PND1 guinea pigs exposed to ACS.
Full size table

Of the top 20 genes, nine had known function in immune regulation, representing the largest fraction of genes (Parp12, Padi2, Slco3a1, Igsf21, Stab2, Kcnh2, Il12rb1, Il1rl2, Pbxip1) [38]. Notably, all DMCs annotated to these nine genes were hypermethylated, ranging from 7.18% hypermethylated (Il1rl2) to 41.8% hypermethylated (Padi2). One other DMC was also localized in the promoter region for Kcnh2 (14.66%), while all other DMCs were within the gene body of the annotated genes. Genes involved in regulation of cytoskeletal components, cell adhesion, or cell migration were also frequently observed, comprising seven of the top 20 genes (Adamts17, Igsf21, Bcam, Stab2, Wrap73, Rreb1, Nphp4). Identified DMCs ranged from −36.65% (Wrap73) to 31.48% (Adamts17), all localized to the gene body.

An examination of all 171 genes through gene ontology (GO) analysis identified significant gene interactions and networks, demonstrating enrichment of terms such as ‘DNA binding’, ‘transcription regulation’, ‘response to stimulus’, ‘signalling’, ‘cell migration’, and ‘cell motility’ (Supplementary Table 2). The top 20 terms are represented in Fig. 1D. Secondary GO analysis using only the set of hypomethylated genes (Supplementary Table 3) identified terms related to ‘transcription’ and ‘DNA binding’. Among the set of hypermethylated genes ‘protein binding’, ‘cellular response stimulus’ and ‘regulation of cellular process’ terms were most enriched (Supplementary Table 4).

Comparison with differential methylation profiles in the guinea pig PFC

Comparing the DMCs identified in the newborn guinea pig blood to those that were previously observed in the newborn guinea pig PFC [18], there were no sites that were differentially methylated in both tissue types following ACS exposure. However, seven genes (Igsf21, Stab2, Fzr1, Dnah17, Tshz2, Prex1, Setd1b) were differentially methylated in both blood and PFC of newborn guinea pigs following exposure to ACS (Table 2). The sites at which differential methylation occur were distinct between the two tissue types. Gene set enrichment analysis of these seven genes did not identify any significant gene networks.

Table 2 Differentially methylated genes identified in both the PFC and blood of PND1 guinea pig offspring following prenatal sGC exposure.
Full size table

Differential methylation at transcription factor binding sites in PND1 blood

A key observation in the guinea pig PFC following ACS exposure was the differential methylation identified at the binding sites for transcription factors PLAGL1, TFAP2C, ZNF263, and SP1 [18]. The TFBS were differentially methylated at both PND1 and PND14, indicating that this methylation signature may be an important mediator of ACS exposure on long-term neurodevelopmental sequelae. As such, TFBS analysis was performed among the DMCs identified in the guinea pig blood derived on PND1 following ACS exposure, to determine whether the same TFBS signature is evident in peripheral tissue. A similar signature of altered methylation in the TFBS in blood and PFC would suggest that the blood could provide a surrogate biomarker reflecting TFBS methylation changes in the brain following ACS.

Using MEME Suite v5.5.3 [33], we performed motif enrichment analysis to identify fixed-length patterns of sequences which are 6–50 bp long, that are repeatedly observed amongst the provided input (significance threshold E-value ≤ 0.05). TOMTOM [35] was subsequently used to compare the identified sequences against a database of known transcription factor binding motifs to highlight the top matches.

In the blood of guinea pigs on PND1, significant changes in methylation were identified at motifs that correspond to the binding sites for transcription factors PLAGL1, TFAP2C, EGR1, SP1, and ZNF263 following ACS exposure (Fig. 2 and Table 3). Binding sequences for PLAGL1 were found within 454 sites (27% of all differentially methylated sites; Supplementary Table 5), annotating to 80 genes enriched in ‘cellular response to stimulus’, ‘signal transduction’, ‘response to chemical’, and ‘cell communication’. TFAP2C binding sites were identified at 356 DMCs (22%; Supplementary Table 6), which annotated to 71 genes involved in ‘cell differentiation’, ‘cell proliferation’, ‘transcription regulation’, and ‘cell motility’, while EGR1 binding sites were identified among 313 sites (19%), in 63 genes that were involved in ‘transcription’, ‘cellular biosynthetic process’, ‘signalling’, ‘response to stimulus’, and ‘cell communication’ (Supplementary Table 7). SP1 binding sites were identified at 84 sites (5%), which annotated to 55 genes enriched in ‘regulation of cell population and proliferation’, ‘cell differentiation’, ‘cell migration’, ‘cell motility’, and ‘signal transduction’ (Supplementary Table 8). ZNF263 binding sites were identified among 213 DMCs (13%) annotating to 50 genes enriched for ‘transcription regulation activity’, ‘protein kinase binding’, ‘calcium-dependent protein serine/threonine kinase activity’, ‘cell motility’, ‘cell migration’, and ‘cellular response to stimulus’ (Supplementary Table 9).

Fig. 2: Transcription factor binding site motifs significantly differentially methylated in newborn guinea pig blood following antenatal betamethasone exposure.
figure 2

Binding motifs for PLAGL1, TFAP2C, EGR1, ZNF263, and SP1 were identified to be significantly differentially methylated in the newborn guinea pig blood following ACS exposure. Significant motif sequences (E-value ≤ 0.05) that were identified surrounding the DMCs identified in the PND1 guinea pig blood following ACS are indicated in the bottom row. Transcription factor binding sites that significantly aligned (E-value ≤ 10) to the identified motifs are indicated in the top row. Each nucleotide is represented in a different colour, and the size of each letter represents the probability of the nucleotide at the position.

Full size image
Table 3 Transcription factor binding sites (TFBS) that were significantly differentially methylated in the whole blood of guinea pig and human neonates following ACS.
Full size table

Methylation at transcription factor binding site in human blood

We previously reported extensive changes in DNA methylation, where 505 differentially methylated sites were identified in newborn human blood following ACS treatment [19]. No variation in methylation values for observed between offspring. Initial comparison of DNA methylation in guinea pig and human blood following ACS revealed no overlap of altered methylation at specific genes between the species. However, following our finding that key transcription factor binding sites are differentially methylated in both the guinea pig blood and PFC following ACS treatment, we re-examined our human neonatal blood methylation data to perform transcription factor binding site enrichment analysis. Similar to the guinea pig, significant differential methylation was identified at binding sites for PLAGL1, TFAP2C, EGR1, ZNF263 and SP1 following ACS (Fig. 3 and Table 3). Binding sites for PLAGL1 were identified among 104 DMCs (20%) annotating to 26 genes (Supplementary Table 10) involved in regulating ‘gene transcription’, while binding sites for TFAP2C were annotated to 49 DMCs (10%) and 17 genes (Supplementary Table 11). EGR1 binding sites was observed within 136 DMCs (27%) and 26 genes following ACS (Supplementary Table 12), and ZNF263 binding sites were identified at 84 DMCs (17%) which annotated to 26 genes (Supplementary Table 13). None of the annotated genes for TFAP2C, EGR1, nor ZNF263 were significantly enriched in any known gene pathways. SP1 binding sites were found within 82 DMCs (16%), annotating to 23 genes (Supplementary Table 14) that were enriched in ‘RNA polymerase II-specific DNA-binding transcription factor binding’, ‘NuRD complex’, ‘positive regulation of proteolysis’, ‘myeloid cell differentiation’, and ‘membrane-bounded organelle’. This suggests that though individual genes may not be overlapping targets of differential DNA methylation following ACS treatment between guinea pigs and humans, there are five core transcription factors and their downstream pathways which are affected in common in both species in response to ACS exposure.

Fig. 3: Transcription factor binding site motifs significantly differentially methylated in human neonatal blood following ACS exposure.
figure 3

Binding motifs for PLAGL1, TFAP2C, EGR1, ZNF263, and SP1 were identified to be significantly differentially methylated in human neonatal blood following ACS exposure. Significant motif sequences (E-value ≤ 0.05) represented on the bottom row, were identified among the full dataset of DMCs identified in the human blood following ACS. Transcription factor binding sequences that aligned significantly (E-value ≤ 10) to the identified motifs are represented in the row above. Each nucleotide is represented in a separate colour, and the size of each letter represents the probability of the nucleotide at the position.

Full size image

Discussion

ACS exposure is associated with increased risk of developing neurodevelopmental disorders in humans, yet mechanistic studies have been limited due to the inaccessibility of brain tissue for examination. Studies have relied on animal models for brain-specific molecular investigation. Previously, we have reported genome-wide DNA methylation changes in multiple brain regions (prefrontal cortex, hippocampus) at various stages of development (newborn, juvenile) following ACS exposure in guinea pigs [18, 39, 40]. However, whether animal-derived mechanistic data can be translated to humans remained unclear.

We recently identified a conserved methylation signature in the PFC of newborn and juvenile guinea pigs where the TFBS of PLAGL1, TFAP2C, ZNF263 and SP1 were significantly differentially methylated in response to ACS exposure [18]. In the current study, we identified that the same ACS-induced TFBS signature is also present in the newborn blood of both guinea pigs and humans. Across our two studies, we have identified an epigenetic signature that is not only consistently observed in the brain and blood of guinea pigs but also in human blood, indicating potential for clinical translatability.

Exposure to ACS results in genome-wide DNA methylation differences in the peripheral blood of PND1 guinea pig offspring. Affected pathways included RNA transcription regulation or cellular response to stimulus, including terms such as ‘DNA binding’, ‘regulation of transcription by RNA polymerase II’, ‘response to stimulus’, ‘signalling’, ‘cell migration’, ‘cell motility’, and ‘endothelial adhesion’. These terms may represent pathways involved in immune regulation, as might be expected for glucocorticoid-mediated effects on leukocyte function [41]. Comparison of the methylation differences found in the blood to those previously found in the PFC indicated no common DMCs. This was unexpected as previous studies have identified significant correlations in DNA methylation between the hippocampus and peripheral blood of PND14 guinea pig offspring following multiple courses of ACS [40]. This may be due to key differences in the experimental designs, as the previous study administered multiple courses of ACS, with treatment administered across a wider phase of fetal development [40]. The present study also examined the newborn (PND1) blood, unlike the previous study which focused on PND14 offspring [40]. While no specific DMCs were shared between the PFC and blood, differential methylation at TFBS for ZNF263, EGR1 and SP1 was observed in both studies, indicating a glucocorticoid-specific biomarker that is present, independent of the treatment dose, offspring age, and brain area.

PLAGL1 is a transcription factor which is normally imprinted and only paternally expressed, though it shows biallelic expression in the blood [42]. PLAGL1 has been shown to regulate anti-proliferative effects, especially in inducing differentiation-associated cell cycle arrest pathways [43], by activating cell cycle and apoptosis pathways through transcriptional activation of p53 [44]. In blood, PLAGL1 has also been shown to downregulate cytokine production through SOCS3 (suppressor of cytokine signalling) activation [45, 46]. In the PND1 guinea pig blood, SOCS3 was hypermethylated in its promoter region at a putative binding site for TFAP2C, thus presenting a possible genomic network by which ACS treatment targets inflammatory suppression [47, 48].

TFAP2C, which encodes for the AP2-gamma protein, is a member of the AP2 family of transcription factors that is known for its ability to inhibit cell migration/invasion [49, 50] while promoting stemness [51] and a state of naïve pluripotency [52]. These downstream functions of TFAP2C align well with known effects of glucocorticoids to impede leukocyte migration and extravasation [53], and direct T-cell migration away from sites of inflammation [54, 55]. In our gene set enrichment analysis, target genes of TFAP2C were enriched in terms such as ‘cell differentiation’, ‘cell motility’, ‘cell migration’, and ‘plasma-membrane bound cell projection’, consistent with known effects of glucocorticoids and TFAP2C [49, 50].

While little is known concerning the effects of glucocorticoid exposure on PLAGL1 and TFAP2C activity, relationships between glucocorticoids and the transcription factors EGR1, ZNF263, and SP1 have been previously reported. EGR1, also known as NGFI-A, is a zinc finger transcription factor that is a downstream target of glucocorticoid signalling [56] known to regulate cellular proliferation, differentiation, and apoptosis [57]. We have previously reported an increase in EGR1 expression following ACS exposure in the hippocampus 24 h after treatment [58], while others have reported hypomethylation at the binding sites for EGR1, ZNF263, and SP1 in the blood of 5-year-old children who also showed high cortisol concentrations in their hair [59], indicating a strong association between these transcription factor binding sites and exposure to endogenous and synthetic glucocorticoids.

In this project, we have demonstrated significant differential methylation at key transcription factor binding sites in peripheral tissue following ACS exposure. Importantly, the same methylation signature was identified in multiple tissue types (PFC vs blood), at multiple developmental timepoints (PND1 vs PND14 PFCs), and in multiple species (guinea pigs vs humans). That the signature was also evident in association with high levels of endogenous glucocorticoids in 5-year-old children highlights the potential strength of this signature as a biomarker for fetal exposure to high levels of exogenous and endogenous glucocorticoids in early life (fetal and neonatal). The latter occurs in cases of high fetal, neonatal or maternal stress. However, we acknowledge that the data reported here were analysed using bioinformatic approaches and may not directly reflect the activity levels of the transcription factor [60]. DNA-protein interactions are also dependent on various additional environmental triggers [61]. As such, future studies to examine such DNA-protein interactions will be required to delineate which of the gene pathways we identified are differentially regulated at a functional level. Furthermore, studies to perform comprehensive DNA methylation and transcriptomic association analyses will support the functional relevance of the differential methylation at TFBS and downstream molecular changes. Nonetheless, our identification of the same signatures in human neonatal blood is highly valuable, emphasizing the potential for the clinical translation of molecular epigenetic data. We are currently assessing neurocognitive outcomes in the OBS cohort as children reach 5-years of age. In future studies, we will be positioned to relate DNA methylation changes to functional outcomes. Further prospective longitudinal studies with larger sample sizes will be important in establishing whether the methylation signature identified in this study can be used for predictive or diagnostic purposes. The identification of biomarkers at early timepoints is critical as the brain is most malleable in these early stages of development. Recognizing epigenetic differences at this early stage provides a deeper understanding of the molecular pathways during a sensitive window of development. Supported by future studies, these data can enable the advancement of individualized intervention strategies, designed to identify infants and children most at risk of developing poor phenotypes following prenatal adversities and provide therapeutic solutions to reshape the molecular landscape and ameliorate developmental outcomes.