Antenatal Steroids and Cord Blood T-cell Glucocorticoid Receptor DNA methylation and Exon 1 Splicing

Abstract

Antenatal administration of glucocorticoids such as betamethasone (BMZ) during the late preterm period improves neonatal respiratory outcomes. However, glucocorticoids may elicit programming effects on immune function and gene regulation. Here we test the hypothesis that exposure to antenatal BMZ alters cord blood immune cell composition in association with altered DNA methylation and alternatively expressed Exon 1 transcripts of the glucocorticoid receptor (GR) gene in cord blood CD4⁺ T-cells.

Study Design –

Cord blood was collected from 51 subjects in the Antenatal Late Preterm Steroids Trial: 27 BMZ, 24 placebo. Proportions of leukocytes were compared between BMZ and placebo. In CD4+ T-cells, methylation at CpG sites in the GR promoter regions and expression of GR mRNA exon 1 variants were compared between BMZ and placebo.

Result –

BMZ was associated with an increase in granulocytes (51.6% vs. 44.7% p=0.03) and a decrease in lymphocytes (36.8% vs. 43.0% p=0.04) as a percent of the leukocyte population vs. placebo. Neither GR methylation nor exon 1 transcript levels differed between groups.

Conclusion –

BMZ is associated with altered cord blood leukocyte proportions, although no associated alterations in GR methylation were observed.

INTRODUCTION

Since the mid-1990s, the standard of care for patients at risk for delivery <34 weeks gestation is a course of glucocorticoids (GCs), either betamethasone or dexamethasone [1]. GC administration decreases the risks of mortality and multiple short-term morbidities among preterm infants, including respiratory distress syndrome, intraventricular hemorrhage, and necrotizing enterocolitis [2, 3]. However, human and animal studies indicate that repeated prenatal administration of GCs can cause sustained alterations to the hypothalamic-pituitary-adrenal (HPA) axis and brain development of offspring [4–6]. Data from animal models also indicate that even single dose or short-term exposure to GCs can result in alterations to the HPA axis and brain development of offspring [7, 8]. As a result of these findings, weekly prenatal GC administration to patients at high risk for preterm birth has been abandoned. However, many clinicians administer a “rescue” course of GCs to patients who remain at high risk of preterm delivery in the next seven days, and who received their first course of GCs at least 14 days prior. The rationale for this approach is based on randomized controlled trials that support a decreased risk for respiratory distress syndrome with rescue course GCs [9, 10].

With the publication of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal Fetal Medicine Units (MFMU) Network Antenatal Late Preterm Steroids (ALPS) randomized controlled trial in 2016 [11], the administration of GCs has been expanded to the late preterm period (34.0–36.6 wks) [11, 12]. The ALPS trial found a decrease in a composite outcome of respiratory support within 72 hrs of birth among neonates born to patients who were high risk for late preterm delivery (34.0–36.6 wks) and received GCs, specifically betamethasone (BMZ), during that gestational window [11]. Thus, many tens of thousands of patients per year are now treated prenatally with GCs. Trials of a single rescue course of GCs have not identified any immediate adverse neonatal consequences or significant harm in early childhood [13]. The ALPS trial did find an increase in hypoglycemia among neonates in the BMZ group versus the placebo group, but no other adverse sequela [11]. There does, however, remain a lack of long-term neonatal follow-up data and thus the potential for fetal programming effects remains.

Fetal programming refers to long-term effects elicited in response to various exposures during development. Fetal programming frequently occurs through epigenetic mechanisms, such as DNA methylation, and results in persistent alterations in the expression of affected genes, in susceptible cell types. The glucocorticoid receptor (GR) gene has been identified as susceptible to such programming effects [14, 15]. GCs bind to GR in the cytosol of a cell, prompting GR translocation to the nucleus where GR functions as a transcription factor, binding to a myriad of target genes and influencing expression. The complement of transcripts produced from the GR gene is large and diverse, and contributes to fine-tuning of cell-type specific expression as well as regulation of GR [16]. One characteristic of the GR gene that contributes to the heterogeneity of GR transcripts is multiple alternative exon 1 mRNA variants, spliced to a common exon 2 [17]. The alternative exon 1 variants are under the control of two promoters, a proximal and distal promoter, with the proximal promoter region embedded in a CpG island, and thus susceptible to alterations in DNA methylation. Human and animal studies demonstrate that environmentally induced alterations in DNA methylation within this region of the GR gene alter tissue-specific exon 1 alternative splicing, including in the presence of GCs. [14, 18, 19].

GCs profoundly affect the immune system, as demonstrated by their clinical use in treating allergy and autoimmune diseases. The HPA axis and immune system are intimately connected, with cortisol suppressing the immune system, in part, through binding to GR in leukocytes [20, 21]. In the context of a mature immune system, GC administration results in a marked neutrophilic leukocytosis and a decrease in monocyte and lymphocyte numbers. GCs alter the adaptive immune system primarily via impaired activation, proliferation, and survival of lymphocytes [22]. CD4⁺ T-cells, in particular, play an important role in establishing and maximizing the capabilities of the neonatal adaptive immune system. GCs increase apoptosis and inhibit the production of cytokines from CD4⁺ T-cells [22]. While the effects of GCs on the mature immune system are well characterized, the effects of GCs on the fetal/neonatal immune system remain poorly understood. Even the term neonate has an immature immune system. Adaptive immunity is immature at birth and must develop specificity and memory in childhood. Lymphocyte subpopulations are lower at birth as compared with adults and preterm infants have lower absolute numbers of lymphocytes compared with term infants, with a greater proportion of naïve T-cells [23].

In this ancillary study to the ALPS trial, we aimed to study whether antenatal BMZ exposure might have fetal programming effects on the immature neonatal immune system. Specifically, we utilized cord blood samples to test the hypothesis that exposure to antenatal BMZ alters cord blood immune cell composition in association with altered DNA methylation and alternatively expressed Exon 1 transcripts of the GR gene in cord blood CD4⁺ T-cells. In order to minimize noise from the methylation signatures in different immune cell types, we chose to utilize CD4⁺ T-cells for this study. Our study focused on methylation across the CpG sparse distal promoter and CpG dense proximal promoter, as well as Exon 1 variants expressed from the proximal promoter (Fig. 1). We chose to focus on the Exon 1 variants GR-1D, GR-1B, and GR-1C because these variants are expressed from the CpG rich region, and expression may be altered by DNA methylation, HPA axis disturbances, or exogenous GCs [24–26].

Fig. 1.

Glucocorticoid receptor schematic. Black circles are methylatable CpGs in distal and proximal promoter regions assessed in this study. Solid lines indicate GR splice variants assessed in this study (1D, 1B and 1C)

METHODS

Design

We performed an ancillary study to the ALPS trial at two sites of the MFMU Network. Full details of the parent study design and outcomes are previously reported [11]. Briefly, patients with a singleton pregnancy at 34 0/7 to 36 5/7 weeks of gestation at risk for late preterm delivery were randomized to BMZ or placebo. Once written informed consent was obtained from participants enrolled in the parent ALPS trial at the University of Utah, they were asked to participate in an associated tissue banking study. This ancillary study utilized samples from those participants who consented to participate in the ALPS trial and the tissue banking study. Relevant to our study, chorioamnionitis was an exclusion criterion for enrollment into the parent ALPS trial. When it was recognized that enrollment was slower than anticipated at the University of Utah, enrollment was expanded to Columbia University, also part of the MFMU Network and a participating site in the parent ALPS trial. After delivery of the neonate, 20–40 ml of cord blood was collected into EDTA tubes and delivered immediately to the University of Utah Flow Cytometry Core Facility, for storage at 4^oC. Samples collected at Columbia University were shipped on ice overnight to the University of Utah Flow Cytometry Core Facility.

Cell separation, T-cell isolation, and DNA/RNA extraction

Cord blood samples were sent to University of Utah Flow Cytometry Core Facility for cell type enumeration and CD4+ T-cell isolation. For cell type enumeration, within 24 hrs of collection each whole cord blood sample underwent RBC lysis and 100µL of the sample was then transferred to a FACS tube. The following lineage marker antibodies were added: CD45 PECY7 (total immune cells), CD14 FITC (monocytes), CD3 PercpCy5.5 and CD19 APC (T-cells, B-cells), CD4 BV421 and CD8 PE (Th and T_C cells). This was followed by 30 minutes of incubation, addition of 5 mL of lysis solution, and then 10 more minutes of incubation. The cells were spun down, resuspended in PBS, and enumeration performed immediately. Doublets and debris were excluded. The lineage markers gave clear discrimination of major subsets including lymphocytes, monocytes and granulocytes. The lymphocyte population was then further differentiated into B-cell and major T-cell subsets including CD4+ and CD8+ cells. Sample gating is shown in Fig. 2A.

Fig. 2.

Sample gating strategy. A) Cell Type Enumeration. Side vs forward scatter plot showing gate around cells and excluding cell debris. Cell phenotyping by linage markers - Left to right, CD45 X side scatter for total immune cells; CD14 X side scatter - % monocytes, granulocytes, and lymphocytes; CD19 X CD3 Lymphocytes (B-cells and T-cells); CD8 X CD4 (Helper T-cells (Th) and Cytotoxic T-cells (Tc)). B) CD4+ T-cell isolation. Left to right, side vs forward scatter showing gate around cells and excluding cell debris, CD45 X side scatter for total immune cells, CD3 X CD4 for CD4⁺ T-cells.

For CD4+ T-cell isolation, the remainder of the cord blood sample was diluted with PBS for a final 1:1 dilution. Cells were resuspended in PBS/BSA staining buffer and the following antibodies added in order to identify and isolate the CD4⁺ T-cells: CD45, CD3, and CD4. Cells were incubated for 30 min. Doublets and debris were excluded. Sample gating for the CD4+ T-cell isolation is shown in Fig. 2B. Data acquisition was accomplished on a BD FACSCanto II and cell sorting experiments utilized a BD FACSAria II Cell Sorter.

CD4⁺ T-cells were snap frozen in liquid nitrogen and stored at −80C for later DNA/RNA extraction. DNA was extracted from the CD4⁺ T-cells by the University of Utah Center for Clinical & Translational Science. DNA isolation was performed using the Gentra Puregene Blood Kit (Qiagen). Concentration and purity were measured using the NanoDrop spectrophotometer 1000. RNA isolation was performed using the NuceloSpin® RNA (MACHEREY-NAGEL) according to manufacturer’s instructions. RNA was quantified using the Qubit RNA quantification assay and quality assessed via 260/280 nm ratio.

Methylation Analysis

Methylation at 32 CpGs across the GR distal and proximal promoter regions (Fig. 1 and Supplemental Table) were assessed using the Illumina HumanMethylation450 BeadChip. For each sample, 500 ng T-cell DNA was bisulfite treated using the EZ DNA Methylation Kit (Zymo, Irvine, CA), with protocol modifications recommended by Illumina (95 C for 30 sec, 50 C for 60 min x 16 cycles). Microarray processing of the Illumina HumanMethylation 450K BeadChip was performed by the University of Utah Genomics Core Facility according to the Illumina protocol. Bisulfite converted samples were hybridized to the BeadChip, washed and stained according to Illumina protocols, and imaged using an Illumina iScan system.

BeadChip data were analyzed using the Methylation Module v1.8 of the GenomeStudio software package (Illumina). Data were background corrected and negative control normalized. Output average beta values (the relative quantity of methylation at an individual CpG locus ranging from 0 to 1 (unmethylated to completely methylated)) were obtained. Because samples were run on three separate BeadChips, three samples were run in duplicate to check batch effects. No batch effects were observed.

GR Alternatively Spliced mRNA Transcript Levels

Real-time reverse transcriptase (RT) PCR was used to measure mRNA transcript levels of GR exons 1D, 1B, and 1C mRNA transcripts in cord blood CD4⁺ T-cells. Real-time RT PCR was performed as previously described [27], with the following Assay-on-demand primer/probe sets: GR Exon 1D 03666144_m1 (GenBank: AJ877166.1); GR Exon 1B Hs01005211_m1 (RefSeq NM_001018076.1); Gr Exon 1C Hs01010755_m1 (RefSeq NM_000176.2); and Total GR Hs00230813_m1 (RefSeq NM_000176.2) (Applied Biosystems, Thermo Fisher Scientific, Foster City, CA). For analysis we used the comparative CT method comparing the exon set of interest to total GR [28].

Statistics

Calculation of sample size was limited by the lack of available data on specific CpG methylation changes in CD4⁺ T-cells in response to clinical exposures. We thus based our sample size calculation on methylation estimates from controls in a study by Oberlander et al., which evaluated the methylation of CpG sites in the proximal promoter region of the GR gene in PBMCs in association with maternal depression [29]. The mean and standard deviation (SD) of percent methylation varied across CpG sites in the Oberlander study. Using the largest SD (3%), we calculated that a total sample size of 60 subjects (N=30 BMZ, N=30 placebo) would detect a 2.2% difference in percent methylation of CpG sites in our GR promoter regions of interest (with 80% power and a type 1 error of 0.05). If the average SD (1.08%) in methylation across CpG sites was used, then a 0.8% difference in methylation could be detected. Oberlander et al. found GR methylation differences in umbilical cord PBMCs of approximately 2% with high vs. low scores on measures of maternal depression in the third trimester²⁵. We anticipated finding differences at least as great, particularly as we utilized a specific cell type rather than all PBMCs.

Baseline clinical and demographic characteristics were compared between the BMZ and placebo groups using Fisher’s exact test for categorical variables or Wilcoxon rank-sums for continuous variables. A Wilcoxon rank-sum test was used to compare cell-type percentages, methylation, and PCR between the BMZ and placebo groups.

RESULTS

A total of 51 participants were included in the study, 40 from the University of Utah and 11 from Columbia. We were unable to reach our planned 60 patients prior to closure of the parent ALPS trial. Demographic and clinical characteristics of the 51 participants were similar between groups (27 BMZ, 24 placebo, Table 1).

Table 1:

Participant Demographics

Characteristic	Betamethasone n=27	Placebo n=24	P value
Pre-pregnancy body mass index, kg/m2	28.9 ± 6.6	26.6 ± 6.8	0.191
Indication for trial entry			0.78
Preterm labor with intact membranes	6 (22.2)	6 (25.0)
Ruptured membranes	9 (33.3)	5 (20.8)
Expected delivery for gestational hypertension/preeclampsia	8 (29.6)	9 (37.5)
Expected delivery for fetal growth restriction	0 (0.0)	0 (0.0)
Expected delivery for oligohydramnios	1 (3.7)	0 (0.0)
Expected delivery for other indications	3 (11.1)	4 (16.7)
Gestational age at trial entry			1.00
≤ 34 weeks	7 (25.9)	6 (25.0)
35 weeks	14 (51.9)	12 (50.0)
≥36 weeks	6 (22.2)	6 (25.0)
Maternal age, years	30.9 ± 6.5	27.6 ± 6.8	0.121
Race			1.00
Black or African American	2 ( 7.4)	2 ( 8.3)
White	23 (85.2)	21 (87.5)
Other/unknown/more than one race	2 ( 7.4)	1 ( 4.2)
Hispanic/Latina ethnicity	7 (25.9)	9 (37.5)	0.55
Nulliparous	7 (25.9)	9 (37.5)	0.55
Smoking during current pregnancy	2 (7.4)	0 (0.0)	0.49
Preeclampsia or gestational hypertension	10 (37.0)	9 (37.5)	1.00
Gestational diabetes	1 (3.7)	1 (4.2)	1.00
Gestational age at delivery, wks	35.6 ± 1.0	35.8 ± 1.3	0.841
Cesarean delivery	11 (40.7)	6 (25.0)	0.37
Post-randomization Chorioamnionitis	0 (0.0)	1 (4.2)

Data are mean ± SD or n (%) unless otherwise noted

¹Wilcoxon rank-sum test, all other tests are Fisher’s exact tests

Of the 51 cord blood samples collected, 48 were available for flow cytometry (25 BMZ, 23 placebo). The other three samples clotted and could not be utilized. We measured granulocytes, lymphocytes and monocytes as a percentage of total leukocytes, B-cells and T-cells as a percentage of total lymphocytes, and CD4+ and CD8+ T-cells as a percentage of total T-cells. BMZ was associated with an increase in granulocytes (51.6% vs. 44.7% p=0.03) and a decrease in lymphocytes (36.8% vs. 43.0% p=0.04) as a percent of the leukocyte population vs. placebo (Table 2). Percentages of monocytes, B-cells, total T-cells, CD4⁺ T-cells, and CD8⁺ T-cells did not differ between BMZ and placebo (Table 2).

Table 2:

Flow cytometry data for cell-type enumeration (% of total parent cell type)

	BMZ (n=25)					Placebo (n=23)
Cells	Median	25th percentile	75th percentile	Minimum	Maximum	Median	25th percentile	75th percentile	Minimum	Maximum	P value
Granulocytes1	51.6	44.6	57.5	35.2	77.3	44.7	34.7	51.7	22.4	67.8	0.03
Lymphocytes1	36.8	32.4	45.3	17.6	51.0	43.0	36.4	51.6	20.3	66.6	0.04
Monocytes1	8.9	8.1	11.5	3.2	13.9	9.3	8.1	12.3	5.6	14.6	0.67
B-cells2	15.9	14.0	21.4	5.9	37.9	16.6	13.3	19.1	7.5	30.6	0.77
T-cells2	60.9	53.9	67.6	40.6	74.7	57.8	53.5	64.0	47.6	77.4	0.75
CD4+ T-cells3	70.4	64.9	74.0	57.2	80.5	72.0	63.8	76.9	59.9	81.7	0.67
CD8+ T-cells3	26.9	22.0	31.3	18.0	36.9	24.7	20.5	31.0	16.5	35.9	0.49

¹percent of leukocytes;

²percent of lymphocytes;

³percent of T-cells.

Methylation of the distal and proximal region of the GR gene was measured in 45 cord blood CD4⁺ T-cell samples, six samples had insufficient DNA to perform the bisulfite conversion. One sample was deleted due to low DNA concentration, leaving 24 in the BMZ group and 20 in the placebo group used in the analysis. In general, CpGs at the region of the distal promoter (CpG’s 1–2), in the start of the proximal promoter region (CpG 5–6), and the CpG’s immediately preceding Exon 2 (CpG 30–32) tended to be methylated. While those in the proximal promoter region (CpG 7–29) tended to be unmethylated (Fig. 3). DNA methylation did not vary between BMZ and placebo at any of these 32 individual CpG sites (Fig. 3).

Fig. 3.

Box-plot showing CpG Methylation as average beta values (the relative quantity of methylation at an individual CpG locus ranging from 0 to 1 (unmethylated to completely methylated)) in CD4+ T-cells isolated from cord blood. Region covered includes the distal and proximal promoters through Exon 2 of the GR gene. Error bars are minimum and maximum percentage methylation, box is the upper and lower quartiles, and the line within the box is median. Grey box is BMZ (n=24), white box is placebo (n=20). No differences were observed between BMZ and placebo for % methylation in any region examined

We measured mRNA levels of transcripts GR-1D, GR-1B, and GR-1C relative to total GR in 51 cord blood T-cell samples. The RT-PCR analysis of GR-1D, GR-1B, or GR-1C relative to total GR included 24 in the BMZ group and 23 in the placebo group, with 4 samples omitted due to low cDNA yield and undetectable signal. mRNA transcript levels of GR-1D, GR-1B, or GR-1C relative to total GR were not different between BMZ and placebo (Fig. 4).

Fig. 4.

Box-plot showing levels of GR 1D, 1B, and 1C mRNA transcripts relative to total GR in cord blood CD4+ T-cells. Error bars are minimum and maximum transcript level, box is the upper and lower quartiles, and the line within the box is median. Grey box is BMZ (n=24), white box is placebo (n=23). No differences were observed between BMZ and placebo for any GR transcript levels examined

DISCUSSION

Our ancillary study to the ALPS trial demonstrated that exposure to antenatal BMZ alters cord blood immune cell composition. However, we did not find associated changes in GR DNA methylation, or levels of the alternatively spliced GR Exon 1 transcripts GR-1D, GR-1B, or GR-1C in cord blood CD4⁺ T-cells. In our study, we did not identify molecular evidence of programming effects of antenatal BMZ on CpGs in CD4⁺ T-cells, possibly due to not reaching our target enrollment. Therefore, the possibility for programming effects remains.

GCs have potent clinical effects. Clinicians are well familiar with the altered proportions of maternal immune cell types observed after administration of antenatal GCs, including an increase in total leukocyte and granulocyte counts and a decrease in lymphocyte counts [30]. In our study, we observed the same changes in cell type distribution in cord blood after antenatal GC exposure, which is consistent with BMZ crossing the placenta. Other studies have also documented an increase in total leukocytes and granulocytes and a reduction in lymphocytes in premature infants after exposure to antenatal GCs [31, 32]. In addition, Chabra et al, found a decrease not only in total lymphocytes but also in CD4⁺ T-cells in cord blood collected from infants exposed to antenatal GCs [32].

Our study did not demonstrate a decrease in CD4⁺ T-cells with BMZ. However, our sample size may have been too limited to detect a difference in a specific subtype of lymphocytes. In other studies, antenatal corticosteroids have been reported to decrease proliferative responses in CD4 and CD8+ T-cells, presumably because of interference with functional capacity of the T-cells [33]. This could be associated with inappropriate cellular immune responses that could be potentially harmful to the infant. The parent study from which these samples and data were derived showed that antenatal corticosteroids improved neonatal respiratory function but also increased the risk of neonatal hypoglycemia with no significant between group differences in the incidence of chorioamnionitis or neonatal sepsis [11]. Long-term follow-up studies of pulmonary and neuropsychological function of the children born in the parent study are currently underway and will provide important outcome data on late preterm infants exposed to antenatal corticosteroids. An important additional consideration is the effect of intra-amniotic inflammation on the fetal hematologic profile. Fetal inflammatory response syndrome resulting from intra-amniotic inflammation/infection is associated with an increased total white blood cell count [32]. In our study, as chorioamnionitis was an exclusion criterion for entry into the parent ALPS trial, and post-randomization chorioamnionitis occurred in only one patient in the placebo group, the observed changes in granulocyte and lymphocyte percentages are more likely to be the result of BMZ as opposed to intra-amniotic inflammation/infection.

Although human and animal studies indicate that prenatal GC exposure can have fetal programming effects, little is known about the epigenetic effects of prenatal GC exposure on GR alternative splicing in a single cell type. The study of prenatal exposures and epigenetics is challenging in humans, in part because epigenetic regulation is cell-type specific. Tissues such as liver, brain, and adipose, which are frequently used in animal models, are not generally accessible in human studies; rather sample acquisition is most often limited to blood. Additionally, previous human studies of prenatal exposures and epigenetic modifications have been limited by the use of DNA isolated from whole blood or peripheral blood mononuclear cells (PBMCs). Whole blood contains many different cell types, including CD4⁺ and CD8⁺ T-cells, B- cells, granulocytes, and monocytes. Importantly, epigenetic differences in these immune cell types have been identified that relate to their cell-type specific functions [34, 35]. Because each cell population varies in developmental timing and location, we speculated that different immune cell types might be differentially affected by prenatal exposures.

Potential fetal programming effects on the HPA axis and the GR gene have been identified in previous studies. Intrauterine growth restriction (IUGR) is associated with hyper-reactivity of the HPA axis and the later development of obesity, diabetes, metabolic syndrome, as well as sustained alterations in innate and adaptive immunity [36–38]. Turner et al. examined the methylation of GR promoter regions in PBMCs obtained from 26 healthy adult volunteers. The levels and positions of methylation were highly variable among study subjects and the authors speculated that this variation in methylation might relate to epigenetic programming in response to early-life events [39]. In another human study, 12 victims of childhood abuse who committed suicide had altered methylation of the GR-1F promoter region in the brain as compared with 12 controls. These methylation differences correlated with changes in the expression of the GR-1F mRNA transcript. While we did not examine levels of the GR-1F variant in our study, a number of human studies have focused on adverse early life events and methylation of the 1F exon of the GR gene [40–42]. Overall, studies of prenatal GC exposure and GR gene methylation have largely focused on endogenous GC exposure, or maternal psychosocial stress [43, 44]. One meta-analysis combining evidence from 977 individuals, showed that methylation levels at single CpG site within the GR gene was significantly correlated to prenatal stress [43]. However, it is important to note that most of these studies utilize cord blood, whole blood collected in childhood, saliva, or buccal swabs for their methylation analyses. In our study, we did not detect changes in methylation of the GR gene promoters or in the expression of GR-1D, GR-1B or GR-1C mRNA transcripts with antenatal GCs. However, our analysis did not evaluate all potential GR Exon1 variants, and our patient population had significant within group variability (e.g. indication for preterm birth risk). Thus, the potential for fetal programming effects in the setting of antenatal GCs remain.

The HPA axis is involved in regulation of the immune system, but very little data exist regarding prenatal GC exposure (either endogenous or exogenous) and programming effects on the immune system. Animal studies have found that antenatal GC administration can affect postnatal febrile and cytokine responses to stimuli [45, 46]. Although a Cochrane systematic review found that antenatal GCs reduce the risk of early neonatal sepsis [47], one study reported an association between multiple courses of prenatal betamethasone treatment and an increase in neonatal sepsis [48]. The benefits of antenatal GCs are significant when administered to the highest risk patients who deliver <34 weeks gestation. However, effects are more modest in the late preterm period. BMZ was not associated with an increase in proven neonatal sepsis in the parent ALPS trial [11]. But given limited long-term follow-up data and less benefit with administration at later gestational ages, treatment should be limited to those who would benefit, namely those who would deliver in this period. This is particularly true when “treatment creep,” or the administration of GCs outside of scenarios supported by randomized controlled trials and the recommendations of professional societies [49], is a common clinical occurrence.

Our study was not without limitations. While the use of a single cell-type is a strength of our study, we were limited by the lack of relevant studies to provide a strong basis for our sample size calculation. An important consideration is that we were unable to reach our planned 60 patients prior to closure of the parent ALPS trial. This may have rendered our study underpowered to detect meaningful differences in DNA methylation and Exon1 splice variants of the GR gene. We were also limited by sample volume, and thus only able to assess a limited number of Exon 1 transcripts. The possibility exists that antenatal GC administration may have an effect on cord blood CD4⁺ cell GR Exon 1 transcripts not examined in this study. Similarly, the regulation of expression of GR is complex and involves additional alternative splicing events beyond those of Exon 1. These were also not examined in this study. Lastly, due to sample volume limitations, we were not able to measure the mediator of effects, specifically GR protein levels. A further limitation lies in potential within-group heterogeneity of our study population. While the clinical/demographic characteristics were not significantly different between study groups (Table 1), the etiologies leading to late preterm birth differed across the study population. As such, within group differences that were not assessed might impact GR gene methylation and Exon 1 transcript expression. Furthermore, although patients were categorized into broad clinical indications for ALPS trial enrollment, the range of severity of conditions such as preeclampsia and fetal growth restriction might also impact GR methylation. Newborn gender is also likely to be a key factor in how prenatal stress affects the epigenome [50]. Finally, some of the pregnant individuals in this study likely received antibiotics for group B Streptococcus prophylaxis, although this information was not specifically collected in the parent trial and decisions regarding antibiotic administration were left to the provider. Although we are not aware of data suggesting that antibiotics solely for group B Streptococcus prophylaxis influence neonatal immune cell counts, there is the potential for unmeasured confounding of the results by antibiotic administration.

In conclusion, our study did not detect differences in methylation of the GR promoter regions in CD4⁺ T-cells with prenatal BMZ exposure. However, future work should continue to focus on utilizing specific immune cell types as well as greater long-term follow-up of immune outcomes. Finally, further study is also needed into the interaction between antenatal GC exposure (and timing), GR gene methylation, prenatal clinical phenotypes such as IUGR and chorioamnionitis, maternal stress, and postnatal experiences such as neonatal infection and adverse early life events.

Availability of data and material (data transparency):

Not applicable