Main

At birth, the transition from intrauterine to extrauterine life requires major endocrine adjustments, such as the metabolic adjustments following the discontinuation of glucose supplied from cord blood and the rapid shift from producing reverse T3 to T3. Understanding these transitions is essential in the endocrine care of the perinatal patient. Evaluation of newborn adrenal function is complicated by the fact that the fetal and later infant adrenals are very different. The human fetal adrenal has two zones, fetal and definitive, in contrast to the three zones, glomerulosa, fasciculata, and reticularis, of the adult gland. Fetal adrenal steroidogenesis begins at about 7 wk gestation: steroidogenic enzymes are detectable by immunocytochemistry in the fetal zone at 50–52 d postconception, and primary cultures of the 8 wk adrenal produce cortisol and respond to adrenocorticotropic hormone (1). The fetal adrenal transiently expresses 3β-hydroxysteroid dehydrogenase, type 2 (3βHSD2, encoded by HSD3B2) at about 8–10 wk, permitting fetal adrenal cortisol synthesis at the same time when male genital development occurs, thus helping to prevent the virilization of female fetuses by suppressing fetal adrenal androgen synthesis (1). The fetal adrenal has relatively little 3βHSD2 activity after 12 wk (1,2) but has 17α-hydroxylase and robust 17,20 lyase activity (both catalyzed by cytochrome P450c17, encoded by CYP17A1), considerable sulfotransferase activity, and little steroid sulfatase activity accounting for its abundant production of dehydroepiandrosterone (DHEA) and its sulfate (DHEAS). DHEAS is secreted, 16α-hydroxylated in the fetal liver by CYP3A7 (3,4,5), and then acted on by placental 3βHSD1, 17βHSD1 (17β-hydroxysteroid dehydrogenase type 1, encoded by HSD17B1), and aromatase (P450aro, encoded by CYP19A1) to produce estriol (6,7). Fetal adrenal steroids are the source of about half of the estrone and estradiol and 90% of the estriol in the maternal circulation (8). Despite the large amounts of DHEA and DHEAS produced by the fetal adrenal and their consequent metabolism to estrogens by the placenta, evidence for an essential role for these steroids is scant, as fetuses with genetic disorders of adrenal steroidogenesis develop normally, reach term gestation, and undergo normal parturition (9). Although glucocorticoids can induce premature lung maturation, they do not appear to be needed when human gestation goes to term, as complete absence of the glucocorticoid receptor is compatible with normal term birth and pulmonary function (10).

It has long been known that the adrenal grows rapidly throughout fetal life, reaching a combined weight of 8–9 g at birth (equal to the combined weight of the adult adrenals) (11,12,13), but within weeks of birth, the fetal adrenals involute to a total weight of about 2 g (12,14). This change is mediated by apoptosis of the fetal zone, possibly in response to activin A or transforming growth factor-β (15). In parallel with the involution of the fetal zone of the adrenal, secretion of DHEA and DHEAS falls dramatically (16,17,18). The triggering mechanism for this rapid, profound change in adrenal morphology, cellular architecture, and steroidogenesis is not known. It has been suggested that the involution of the fetal adrenal is more related to gestational age than to timing after birth (19), but more recent studies indicate that parturition itself triggers fetal adrenal involution, which was interpreted as suggesting that the withdrawal of a placental factor stimulated the onset of fetal adrenal apoptosis (20). We hypothesize that the transition from intrauterine hypoxia to extrauterine normoxia is also a key event in triggering the remodeling of the fetal adrenal. Human fetal abdominal viscera receive hypoxic blood having a partial pressure of oxygen (Po2) of only ~2 kPa (1 kPa = 7.5 Torr; 1 Torr = 1 mm Hg); perinatal circulatory changes change this to adult values of ~20 kPa. As parturition itself appears to trigger the changes in fetal adrenal steroidogenesis and architecture, we considered whether the perinatal change in arterial oxygen tension participates in these changes. As a preliminary test of this hypothesis, we grew adrenal cells in long-term hypoxic conditions designed to mimic the intrauterine environment and then examined changes in gene expression upon transition to a normoxic environment that models extrauterine life.

Results

Human Fetal Adrenal Cells

To study the changes in the human adrenal as it transitions from the hypoxia of fetal life to the normoxia of the extrauterine newborn environment, we first incubated adrenal cells from a single 17-wk human fetus under hypoxic conditions for 1 d, followed by normoxic conditions for 1 or 2 d. Total cellular RNA from these cells was hybridized to Illumina BeadChip microarrays for gene expression analyses. Using an arbitrary cutoff of >1.5-fold change for gene activation or <0.67-fold change for gene repression, the mRNAs encoded by 107 genes were increased, and those for 114 genes were decreased when the cells were shifted from fetal hypoxic conditions to normoxic conditions for 1 d, and 179 mRNAs were increased and 296 genes were decreased after 2 d in normoxic conditions (Supplementary Tables S1 and S2 online). Of these transcripts, 54 were increased and 111 were decreased on both days ( Figure 1 ).

Figure 1
figure 1

Summary of microarray gene expression profiles from the primary culture of human fetal adrenal cells. A primary culture of fetal adrenal cells was incubated under hypoxic conditions for 1 d, followed by normoxic conditions for either 1 or 2 d; control cells were maintained under hypoxic conditions throughout the experiment. (a) Gene expression levels were calculated as the signal levels under normoxic conditions divided by the signal levels under hypoxic conditions for control cells. (b) Venn diagram showing gene expression profiles in the primary culture of fetal adrenal cells incubated under the above conditions. Arrows pointing upward and downward represent increased and decreased numbers of expressed genes.

PowerPoint slide

Full size image

While this experiment showed that the hypoxic–normoxic transition can change the abundance of many adrenal mRNAs, changes were not seen in the transcripts for any gene encoding a steroidogenic enzyme or its electron-transfer cofactor. To examine mRNAs encoding steroidogenic factors more closely, we obtained additional adrenals, incubated primary adrenal cell cultures under hypoxic and normoxic conditions, and measured the relative abundances of selected mRNAs under each condition for each adrenal by reverse transcription followed by quantitative real-time PCR. Consistent with prior observations (21), preliminary experiments showed that the relative abundances of mRNAs for P450scc and P450c17 decreased after 4 d due to the overgrowth of fibroblasts and apoptosis of fetal adrenal cells (data not shown). Thus, we used 2 d of culture for more detailed studies with adrenals from five fetuses (three male and two female; 17–23 wk gestation). We noted no changes in the morphology of the adrenal cells after transition from hypoxia to normoxia for 1–2 d. Under normoxic conditions, the abundance of the mRNAs for P450c17, steroidogenic acute regulatory protein (StAR), and 3βHSD2 increased after 2 d ( Figure 2 ). Consistent with the data from other cell types (22), glyceraldehyde-3-phosphate dehydrogenase gene expression decreased in normoxia compared with hypoxia, but the expression of mRNAs for 3βHSD2, StAR, and P450c17 increased 2.6-, 2.0-, and 1.6-fold under normoxic conditions, while expression of P450scc barely changed. However, there was a substantial variation with the fetal adrenal cells from different fetuses, so that the statistical analyses were of marginal significance.

Figure 2
figure 2

Expression of mRNAs in human fetal adrenal cells. Duplicate cultures of five fetal adrenals were grown in conditions of hypoxia (open bars) and normoxia (closed bars), and the mRNAs for P450c17, StAR, 3βHSD2, P450scc, and GAPDH were quantitated by qPCR. The mean levels in hypoxia are set at 100% for each RNA; data are mean ± SEM; *P < 0.05. 3βHSD2, 3β-hydroxysteroid dehydrogenase, type 2; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; P450c17, 17α-hydroxylase/17,20 lyase; qPCR, quantitative PCR; StAR, steroidogenic acute regulatory protein.

PowerPoint slide

Full size image

Human Adrenal NCI-H295A Cells

To avoid differences between individual fetal adrenals, we sought to use the immortalized human adrenal NCI-H295A cell line, in which the patterns of steroidogenesis closely resemble those of the fetal adrenal (23). Because these cells are normally cultured in normoxic conditions, we first acclimated them to the fetal environment by culturing them in hypoxic conditions for 15 d before “delivering” them to extrauterine normoxic conditions for either 1 or 2 d; control cells were maintained in hypoxic conditions throughout the experiment. We noted no changes in cellular morphology when the NCI-H295A cells were transitioned from 15 d of hypoxia to normoxia. The resulting mRNAs were analyzed by hybridization to Illumina BeadChip microarrays, thus permitting analysis of the entire transcriptome. In NCI-H295A cells, only 4 mRNAs were increased and 56 were decreased when the cells were shifted from fetal hypoxic conditions to extrauterine normoxic conditions for 1 d, and none of the altered mRNAs appeared to participate in steroidogenesis ( Table 1 and Supplementary Table S3 online). By contrast, after the NCI-H295A cells had been returned to normoxia for 2 d, 62 mRNAs were increased and 105 mRNAs were decreased ( Table 2 and Supplementary Table S3 online). Among the mRNAs that increased, three encode proteins that participate in steroidogenesis: P450c17 (CYP17A1), increased 1.65-fold; 3βHSD2 (HSD3B2), increased 1.89-fold; and StAR (STAR), increased 1.78-fold over controls. In addition, sterol isomerase (EBP), which participates in cholesterol biosynthesis, increased 1.57-fold. The degree of overlap in these gene populations is shown in Figure 3 . These gene expression profiles showed that 46 mRNAs were regulated in the same fashion after both 1 and 2 days of normoxia (1 increased and 45 decreased). In addition, the mRNAs for 3 other genes were increased and 11 were decreased after 1 d of normoxia and 61 mRNAs were increased and 60 reduced after 2 d of normoxia.

Table 1 Expression of 4 genes activated and 56 genes inhibited in NCI-H295A cells incubated under hypoxic conditions for 15 d, followed by normoxic conditions for 1 d
Full size table
Table 2 Expression of 62 genes activated and 105 genes inhibited in NCI-H295A cells incubated under hypoxic conditions for 15 d, followed by normoxic conditions for 2 d
Full size table
Figure 3
figure 3

Summary of microarray gene expression profiles from NCI-H295A cells. NCI-H295A cells were incubated under hypoxic conditions for 15 d, followed by normoxic conditions for either 1 d or 2 d, while control cells were maintained under hypoxic conditions throughout the experiment. (a) Gene expression levels were calculated as the signal levels under normoxic conditions divided by the signal levels under hypoxic conditions for control cells. A fold change cutoff of >1.5-fold or <0.67-fold over control was chosen in our study. (b) Venn diagram showing gene expression profiles in NCI-H295A cells incubated under the above conditions. Arrows pointing upward and downward represent increased and decreased gene expression.

PowerPoint slide

Full size image

Among the genes whose mRNAs increased or decreased under normoxic conditions ( Tables 1 and 2 ), ALDOA, ALDOC, BNIP3, BNIP3L, CA9, ENO1, GADPH, HK2, IGFBP2, JMJD1A, LDHA, NDRG1, PKM2, SLC2A1, SLC2A3, and TPI1 are known to be transcriptionally regulated by hypoxia (24). Only four genes, BNIP3, NDRG1, SERPINA3, and SLC2A1 were regulated in common in both NCI-H295A cells and in the primary culture of fetal adrenal cells. The complete expression profile data for all genes in NCI-H295A and in the primary cultures of fetal adrenal cells are shown in Supplementary Tables S1–S4 online.

Gene ontology analyses using Ingenuity Pathway Analysis (https://analysis.ingenuity.com) showed that many of the repressed genes in NCI-H295A incubated for 1 or 2 d in normoxic conditions participate in common pathways such as glycolysis, sucrose degradation, vitamin C transport, thyroid hormone receptor/retinoid X receptor activation, and HIF1α signaling ( Tables 3 , 4 , and 5 ). In contrast, the activated genes in NCI-H295A are involved in glutathione-mediated detoxification, dendritic-natural killer cells crosstalk, p53 signaling, and, of course, steroidogenesis.

Table 3 Pathway analysis of genes downregulated in NCI-H295A cells incubated in hypoxic conditions for 15 d, followed by normoxic conditions for 1 d
Full size table
Table 4 Pathway analysis of genes upregulated in NCI-H295A cells incubated in hypoxic conditions for 15 d, followed by normoxic conditions for 2 d
Full size table
Table 5 Pathway analysis of genes downregulated in NCI-H295A cells incubated in hypoxic conditions for 15 d, followed by normoxic conditions for 2 d
Full size table

Discussion

Little information is available concerning the potential effects of environmental oxygenation on fetal adrenal function, and most such reports have investigated animal models of high-altitude stress. Thus, when pregnant rats were transitioned to reduced air pressure of ~380 Torr (~50 kPa; Po2 ~10 kPa; designed to correspond to 18,000 feet above sea level), the adrenals of fetuses were larger, possibly due to increased adrenocorticotropic hormone secretion (25). Long-term maintenance of pregnant sheep at 3,820 m above sea level (Po2 ~102 Torr; 13.6 kPa) reduced expression of mRNAs and proteins for P450scc, P450c17, and MC2R (adrenocorticotropic hormone receptor) but did not alter P450c21 (21-hydroxylase, encoded by CYP21A2), StAR (encoded by STAR), 3βHSD2, or DAX-1 (26). At birth, the fetal zone of the human adrenal cortex involutes rapidly, as evidenced by rapidly declining serum concentrations of DHEA and DHEAS. Furthermore, this rapid decline in DHEA/S is seen in both premature and term infants (27). Therefore, we and others have hypothesized that the involution of the fetal adrenal is not “programmed” but “triggered.” A current view is that the trigger is the loss of placental hormones and growth factors (20). Such a trigger could also be secondary to the profound environmental change that accompanies birth. Such changes initiate the transition from fetal to postnatal circulatory patterns, including closure of the foramen ovale and the ductus arteriosus. Closure of the ductus is directly triggered by increased oxygen tension (via prostaglandins), and many other events in the newborn are triggered by the transition to normoxia (28). Thus, we hypothesized that the transition from the intrauterine hypoxic environment to the extrauterine normoxic environment might participate in initiating the rapid changes in adrenal steroidogenesis that follow birth. The fetal adrenal and other organs served by the fetal abdominal aorta are bathed in oxygen-poor blood having a partial pressure of oxygen of about 20–22 Torr (2.6–2.9 kPa) (29). Therefore, to model the changes in the adrenal environment that accompany birth, we incubated human fetal adrenal cells and human adrenal NCI-H295A cells in 2% oxygen (hypoxia, Po2 ~2.0 kPa) followed by incubation in atmospheric oxygen (normoxia, ~20 kPa) for 1 and 2 d.

Results with human fetal adrenals suggested that the hypoxic–normoxic transition increased the mRNAs for StAR, 3βHSD2, and P450c17 and decreased the mRNA for P450scc, but only the data with StAR reached nominal significance of P < 0.05. There was substantial variation among adrenals from different donors, with no pattern attributable to donor sex or gestational age in the 17–23-wk period used. Therefore, we turned to human adrenocortical carcinoma NCI-H295A cells, which possess features typical of fetal, rather than adult, adrenal cells (e.g., expression of IGF-2 and P450aro) (23). The cells were propagated in hypoxic conditions for 15 d to acclimate them to this model intrauterine environment before “delivering” them to normoxia. This transition induced changes in the abundance of many mRNAs, with many more changes after 2 d than after 1 d. Not surprisingly, one of the induced genes was HIF1A, which encodes a hypoxia-induced transcription factor (30). Most notably, after 2 d of normoxia, the abundances of the mRNAs for the steroidogenic factors 3βHSD2, StAR, and P450c17 increased >1.6-fold. This will change the pattern of steroidogenesis from Δ5 to Δ4 steroids, as is seen following birth.

Our study emphasizes analysis of mRNAs and did not measure the abundances of steroidogenic enzyme proteins or the steroid products of our adrenal cell systems; such measurements will be of interest in future studies. In addition, our experimental design only examined events in the first 2 d following the transition to normoxia, yet the involution of the fetal adrenal and the transition from fetal to newborn pattern of steroidogenesis takes several weeks, so that future studies may also examine a broader time frame. However, it seems likely that the rapid changes in oxygenation at delivery would constitute an acute trigger to change the adrenal’s transcriptional programming and that such acute changes would subsequently affect adrenal morphology and steroid secretory patterns over the first weeks of life, so we would expect that the changes in mRNA abundances that we have measured would precede changes in adrenal morphology and steroid secretion. In this context, it may be important to add tropic activators of the protein kinase A pathway (adrenocorticotropic hormone to adrenal cells and 8-Br-cAMP to NCI-H295A cells), to mimic conditions in vivo.

Our data are consistent with our hypothesis that the change in oxygenation that follows birth is a key factor in determining the change in the patterns of adrenal steroidogenesis that follow birth. However, no aspect of our data is inconsistent with the hypothesis that withdrawal of placental factors also plays a role, especially in the rapid involution of adrenal size, as known adrenal growth factors (IGF-2, EGF, FGF) (31) were not among the factors dramatically changed by the hypoxic–normoxic transition in our studies. Thus, we propose that both the hypoxic–normoxic transition and the potential withdrawal of placental factors are required to initiate the anatomic remodeling of the fetal adrenal and its change in steroidogenic patterns as the fetus transitions to extrauterine life. While there may be differences among expression levels of mRNAs, their encoded proteins, and downstream steroids, our preliminary data suggest that the hypoxic/normoxic transition at birth is likely to be an important component of the perinatal changes in adrenal architecture and steroidogenesis.

Methods

Cells

We used two cell systems. First, we used primary cultures from human fetal adrenals obtained with written consent from women undergoing elective procedures at San Francisco General Hospital. This research was performed with Institutional Review Board approval from the University of California San Francisco’s Committee on Human Research. All specimens were anonymous. The gestational age of the fetal specimens was estimated based on foot length. Second, we used the NCI-H295A human adrenocortical cell line (32) that expresses all adrenal steroidogenic enzymes (23) and has been selected to grow in monolayer (33). All cells were grown in Roswell Park Memorial Institute (RPMI) medium (UCSF cell culture facility) with 2% fetal bovine serum. All experiments were performed in triplicate. Hypoxic conditions consisted of an atmosphere of 2% oxygen, 93% nitrogen, and 5% CO2 in the XVIVO hypoxia tissue culture hood from BioSpherix (Lacona, NY). Constant oxygen levels were maintained in the hypoxia chamber throughout the experimental procedure and incubations. All plastic ware, pipette aids, tissue culture media, and buffers were equilibrated in the hypoxic conditions before use.

Incubations

Under an institutional review board–approved human experimentation protocol, fetal adrenal tissues was transported in a full tube of phosphate-buffered saline that had been degassed so as to minimize exposure to oxygen before arriving in the laboratory. All manipulations were done under hypoxic conditions. Fetal adrenals were de-encapsulated, the two adrenals were combined, minced into small pieces, rinsed twice with Ca/Mg-free Hank’s balanced salt solution, digested with 44 mg dispase (Life Technologies, Carlsbad, CA), 20 mg collagenase type I (Worthington Biochemical, Lakewood, NJ) at 37°C for 40 min, filtered through 100 µm nylon mesh, layered onto 5 ml ficoll-paque plus (GE Healthcare, Piscataway, NJ), and centrifuged at 600g for 30 min at room temperature on an IEC centraGP8R centrifuge (Thermo Fisher Scientific, Waltham, MA). The cells from the resulting interphase was collected and washed, resuspended in RPMI medium, and incubated in the hypoxic environment. Cells from each fetus were incubated separately in duplicate cultures grown in hypoxic or normoxic conditions; subsequent RNA preparations and analyses were done separately.

NCI-H295A cells were incubated under hypoxic conditions for 15 d and were split twice before the start of the experiment. After 15 d, six 10-cm plates of cells were moved from hypoxic conditions to an incubator with room air (normoxia). RNA was isolated from three plates after 1 d in normoxia and from the other three plates after 2 d in normoxia. RNA was also isolated from three control plates maintained in hypoxia.

RNA Analysis

Total RNA was isolated using TRIzol (Life Technologies) according to the manufacturer’s recommended protocol. For cells kept in hypoxia, homogenization with TRIzol was done under hypoxic conditions. RNA was quantitated using an ND-1000 NanoDrop spectrophotometer (Thermo Scientific).

For real-time quantitative reverse transcription-PCR, 1 µg total RNA was reverse transcribed using Superscript II reverse transcriptase (Life Technologies), and PCR was performed at 94 °C for 5 min, followed by 40 cycles of 94 °C for 0.5 min, 55 °C for 0.5 min, and 72 °C for 1 min using primers and probes for the cholesterol side-chain cleavage enzyme (P450scc, encoded by CYP11A1), P450c17, 3βHSD2, StAR, glyceraldehyde-3-phosphate dehydrogenase, and actin ( Table 6 ). Reactions were performed in a total volume 25 µl containing 2 µl complementary DNA and 12.5 µl FastStart SYBR Green Master (Roche, Mannheim, Germany) on an iCycler iQ Real Time Detection System (Bio-Rad, Hercules, CA).

Table 6 Sequence of primers used for qRT-PCR
Full size table

For microarray experiments, 300 ng total RNA was used to produce biotin-labeled complementary RNA using Illumina TotalPrep RNA amplification kit (Life Technologies) according to the manufacturer’s recommended protocol. The biotinylated complementary RNA was eluted in nuclease-free water and was quantitated by NanoDrop spectrophotometer. Hybridization to the HumanHT-12 v4 BeadChip array (Illumina, San Diego, CA) was done at a complementary RNA concentration of 150 ng/µl in the UCSF Genomic Core Facility. The chips were scanned, and data were analyzed using Genome Studio Gene Expression Module (Illumina). Data were normalized using the “quantile” method of normalization in the software. Normalized data containing the signal levels and detection P values were exported into Microsoft Excel for gene expression analysis. The fold changes in gene expression levels were calculated as the gene signal levels under normoxic conditions divided by the signal levels under hypoxic conditions for control cells. The data are expressed as mean fold change in normoxia over control in hypoxia ± SD. Genes with signal detection of P > 0.05 in both normoxia and control groups were excluded from further analysis. An arbitrary fold change cutoff of >1.5-fold or <0.67-fold over control was chosen.

Statement of Financial Support

This work was supported by the University of California–San Francisco Molecular Endocrinology Fund; J.Q. was supported by the School of Medicine, Shanghai Jiao Tong University, Shanghai, China.

Disclosure

The authors have nothing to disclose. The authors report no conflict of interest.