Corticosterone, Adrenal, and the Pituitary-Gonadal Axis in Neonatal Rats: Effect of Maternal Separation and Hypoxia

Abstract

Hypoxia, a common stressor in prematurity, leads to sexually dimorphic, short- and long-term effects on the adult hypothalamic-pituitary-adrenal (HPA) and hypothalamic-pituitary-gonadal (HPG) axes. We hypothesized that these effects are due to stress-induced increases in testosterone during early postnatal life. We evaluated this phenomenon by systematically assessing the short-term effects of normoxic or hypoxic separation on male and female pups at birth, postnatal hours (H) 2, 4, and 8, and postnatal days (PD) 2 to 7. Our findings were (a) hypoxic separation led to a large increase in plasma corticosterone from 4H-PD4, (b) neither normoxic nor hypoxic separation affected critical adrenal steroidogenic pathway genes; however, a significant decrease in baseline Cyp11a1, Mc2r, Mrap, and Star adrenal expression during the first week of neonatal life confirmed the start of the adrenal stress hyporesponsive period, (c) a luteinizing hormone/follicle-stimulating hormone–independent increase in plasma testosterone occurred in normoxic and hypoxic separated male pups at birth, (d) testicular Cyp11a1, Lhcgr, and Star expression was high at birth and decreased thereafter suggesting a hyporesponsive period in the testes, and (e) elevated estrogen in the early neonatal period occurred independently of gonadotropin stimulation. We conclude that a large corticosterone response to hypoxia during the first 5 days of life occurs as an adaptation to neonatal stress, that the testosterone surge during the first hours after birth occurs independently of gonadotropins but is associated with upregulation of the steroidogenic pathway genes in the testes, and that high postnatal estrogen production also occurs independently of gonadotropins.

Acute hypoxia in the premature neonate may have long-term consequences on the adult hypothalamic-pituitary-adrenal (HPA) and hypothalamic-pituitary-gonadal (HPG) axes in a sexually dimorphic manner (1-4). We have previously shown that hypoxia and maternal-neonatal separation exert long-term programming effects in the adult such as changes in insulin, glucose, and insulin sensitivity, an increase in metabolic markers such as leptin and adiponectin, and alterations in the HPA axis response to stress (1, 5). These long-term sexually dimorphic programming effects in the adult rat may be due to a stress-induced augmentation of the neonatal testosterone surge (6-9).

A surge in testosterone occurs in mammalian males during embryonic development, shortly after birth, and again at the onset of puberty (10). Although little is known about the mechanism that generates the neonatal testosterone surge, it occurs in many species (10-12) and is critical in rats for establishing adult reproductive behavior, sexually dimorphic gonadotropin secretion, and proper sexual differentiation of the developing brain (13, 14). Maternal exposure to environmental factors like stress, nicotine, and alcohol have been shown to decrease or shift the amplitude of the perinatal and postnatal testosterone surges (15-20), which correlate with feminization of the adult male rat (21-23). Gonadectomy-induced reduction of circulating testosterone in the neonatal male rat leads to increased basal corticosterone levels and an augmented adrenocorticotropin (ACTH) and corticosterone response to stress, both of which are reversed with the administration of testosterone (8, 24, 25). These studies collectively suggest that normal testosterone dynamics are required for proper development of the HPA and HPG axes in the neonate.

We hypothesized previously that hypoxia-induced increases in testosterone at and soon after birth (at the time of the normal neonatal testosterone surge) might program the subsequent behavior of the HPA axis (1, 26). The purpose of this study was to determine whether (a) hypoxia does indeed augment the neonatal male plasma testosterone surge measured by liquid chromatography–tandem mass spectrometry, (b) this augmentation correlates with changes in neonatal gonadotropin levels and/or testicular enzyme expression, and (c) there is a temporal relationship between the early neonatal gonadal steroid response to separation and hypoxia and the subsequent stress-induced augmentation of corticosterone during the first 7 days of life. We independently assessed the acute effects of normoxic separation (a potential stressor per se) and with the addition of hypoxia to separation stress in the neonatal male and female rat during the first week of life on plasma corticosterone, testosterone, estradiol, luteinizing hormone (LH) and follicle-stimulating hormone (FSH), and the associated changes in the messenger RNA (mRNA) expression of critical adrenal and testicular genes involved in steroidogenesis.

Materials and Methods

Animal treatment and experimental protocols

Animal protocols were approved by the Institutional Animal Care and Use Committee of Aurora Health Care and the federal guidelines for the use and care of laboratory animals were followed (https://grants.nih.gov/grants/olaw/references/phspol.htm). Timed-pregnant Sprague-Dawley (SD) female rats (N = 82) were obtained from Envigo at gestational ages E15 to E18. Dams and their litters were maintained in a controlled environment (lights on 6 am to 6 pm) and given a standard diet with water ad libitum (27, 28). Male and female pups were studied as described as follows.

Neonatal experiments

Rat pups of both sexes were studied starting immediately at birth, 2 hours postbirth (2H), 4 hours postbirth (4H), 8 hours postbirth (8H), and at postnatal days (PD) 2, 3, 4, 5, 6, and 7. The day of birth is denoted PD1, so PD2 is 24 hours postbirth. Dams were monitored during lights on (6 am to 6 pm) for the onset of parturition and allowed to deliver spontaneously. Dams that delivered during light hours were reserved for early time-point collections so as to not disrupt the dark phase. Birth samples: Pups were immediately and quickly separated from the dam when delivery was complete and only pups (not dams) were exposed to treatments described in the following sections. 2H, 4H, and 8H samples: Dams were observed for parturition during light hours, and time monitoring began when approximately half of the anticipated litter was born (6 or 7 pups).

Male and female rat pups within each age group were separated from their lactating dams and randomized into the following acute experimental groups: Baseline: Rat pups were removed from their home cage with lactating dams and immediately decapitated for trunk blood collection. Separation (normoxic time control): Rat pups were separated from their dams and placed into an environmental chamber vented with room air (21% O₂) for 60 minutes with low-heated bedding to prevent hypothermia associated with maternal separation (27). Hypoxia with separation: Rat pups of both sexes were separated from their dams and placed into an environmental chamber for 60 minutes with low-heated bedding as described previously. Hypoxia was achieved by reducing inspired oxygen in the chamber to 8% O₂ as described previously (29, 30). Trunk blood was collected at 60 minutes of separation or hypoxia.

Trunk blood was collected by decapitation into tubes containing K₂ EDTA. At each age group, blood from 2 to 4 pups was pooled to achieve an adequate amount of plasma for all hormone measurements as described previously (31). Each pooled plasma sample is n = 1 for statistical analysis.

Plasma hormone assays

Plasma corticosterone was measured by radioimmunoassay (MP Biomedical) as described previously (32, 33). Male plasma testosterone was measured in our laboratory-developed method by liquid chromatography–tandem mass spectrometry as described previously (5). Female plasma estradiol was measured by enzyme-linked immunosorbent assay (Calbiotech) as described previously (26, 34, 35). Plasma FSH and LH were analyzed at the Ligand Assay and Analysis Core within the Center for Research in Reproduction at the University of Virginia School of Medicine by multiplex testing (36) (EMD Millipore). Values less than 0.24 ng/mL (FSH/LH assay minimum detectible limit) were assigned a value of 0.24 ng/mL. Male LH samples were not subjected to statistical analysis and not reported because most of the results were less than 0.24 ng/mL. All samples were analyzed using a blinded protocol with unique, experimentally deidentified sample numbers.

Adrenal and testis messenger RNA expression

Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analysis followed MIQE (Minimum Information for Publication of Quantitative Real-Time PCR Experiments) guidelines (37). Immediately after decapitation, adrenals from both sexes and testes from males were flash-frozen into liquid nitrogen and stored at –80 °C until RNA extraction. Owing to their very small size before PD3 and the resultant difficulty in locating them quickly (38, 39), as well as the time limit of freezing tissue quickly to reduce RNA degradation, ovaries were not obtained at necropsy. Total RNA from whole adrenal glands and testes were lysed in QIAzol with a TissueLyser II (Qiagen) and extracted using the RNeasy Lipid Tissue Mini Kit (Qiagen) with an on-column DNase treatment following the instructions from the manufacturer. RNA was quantified using a Nanodrop 2000 ultraviolet-Vis spectrophotometer (Thermo Fisher Scientific) and quality was assessed via gel electrophoresis. Total adrenal RNA (300 ng) and testes RNA (500 ng) were reverse-transcribed to complementary DNA (cDNA) using Superscript IV according to the manufacturer’s instructions (Thermo Fisher Scientific). All cDNA was diluted 1:10 in molecular biology grade water. RT-qPCR was performed using Roche Light Cycler Prober Mater Mix (Roche) and TaqMan hydrolysis probes with FAM fluorophore (Thermo Fisher Scientific). The final reaction volume of 20 µL consisted of 1X Prober Master Mix, 1X TaqMan primer/probe mix, and 5 µL of previously diluted cDNA. The TaqMan hydrolysis probes used are listed in Table 1. Amplification and detection were performed on a Roche LightCycler 480 II real-time PCR machine using the following thermal cycler conditions: 1 cycle of 95 °C for 10 minutes, and 45 cycles of 95 °C for 10 seconds, 60 °C for 30 seconds, and 72 °C for 1 second. Samples were run one gene per 384-well plate in triplicate. C_q (quantification cycle) values of adrenal and testis mRNA expression were normalized to Actb gene and Rpl19 gene, respectively. Relative mRNA expression was calculated using the 2^–ΔΔCt relative quantification method (40), with the nadir time point within baseline of each gene of interest as the calibrator. The calibrator for each gene was identified as being the age group within baseline that had the highest mean C_q, and RT-qPCR data are presented as fold change compared to the calibrator (nadir time point). Mean ± SE C_q data used to calculate fold changes are provided in Supplemental Tables S1 and S2 (41). We want to emphasize that the RT-qPCR data are plotted as fold change using the validated 2^–ΔΔCt relative quantification method (40) for the purposes of graphical clarity and communication only, but that the statistical analyses annotated on these fold-change graphs were performed on the individual C_q data.

Table 1.

List of rat genes and corresponding TaqMan hydrolysis probes

Gene	RefSeq	Name	TaqMan assay
Actb	NM_031144	Actin, beta	Rn00667869_m1
Rpl19	NM_031103	Ribosomal protein L19	Rn00821265_g1
Ar	NM_012502	Androgen receptor	Rn00560747_m1
Cyp11a1	NM_017286	Cytochrome P450, family 11, subfamily A, polypeptide 1	Rn00568733_m1
Cyp17a1	NM_012753	Cytochrome P450, family 17, subfamily A, polypeptide 1	Rn00562601_m1
Fshr	NM_199237	Follicle-stimulating hormone receptor	Rn01648507_m1
Hsd3b	NM_001042619	3 beta-hydroxysteroid dehydrogenase	Rn01789220_m1
Hsd17b1	NM_012851	Hydroxysteroid (17-beta) dehydrogenase 1	Rn00563388_g1
Ldlr	NM_175762	Low density lipoprotein receptor	Rn00598442_m1
Lhcgr	NM_012978	Luteinizing hormone/choriogonadotropin receptor	Rn00564309_m1
Mc2r	NM_001100491	Melanocortin 2 receptor	Rn02082290_s1
Mrap	NM_001135834	Melanocortin 2 receptor accessory protein	Rn01477212_m1
Mrap2	NM_001108774	Melanocortin 2 receptor accessory protein 2	Rn01445736_m1
Star	NM_031558	Steroidogenic acute regulatory protein	Rn00580695_m1

Statistical analyses

Hormone data were analyzed by 2-way analysis of variance (ANOVA) with and without log₁₀ transformation and Duncan’s multiple range test (SigmaPlot 14.0; Systat Software Inc). We chose sample size based on the 2 main outcomes: For corticosterone, at a power of 0.80 and an α of .05, the recommended samples size was 10 for the effects we predicted a priori based on our previous studies (29, 31). For testosterone, at a power of 0.80 and an α of .05, the recommended samples size was 6 for the effects we predicted a priori based on our previous study (26). In most cases, the sample sizes exceeded these minima and, for the most part, the power of the ANOVA residuals with P less than .05 was 0.9 or greater. One-way ANOVA (followed by Duncan’s multiple range testing) and t tests were occasionally used to explore data sets with high variance and are identified where appropriate within the text. Raw C_q values from RT-qPCR results were analyzed by 2-way ANOVA and Duncan’s multiple range test. P less than .05 was considered statistically significant. Hormone values are reported as mean ± SE. RT-qPCR data are presented as fold change calculated as described previously. Effect size was calculated and presented for all comparisons that we thought were most physiologically important and relevant to the main hypotheses (42, 43). Pertinent effect sizes with CI are presented within results, and the magnitude of effect size (Cohen’s d: small = 0.2, medium = 0.5, large = 0.8) are described in “Results” and figure legends where appropriate (42).

Results

Corticosterone and adrenal messenger RNA expression

There were no significant differences between sexes in plasma corticosterone concentrations, so male and female data were combined to analyze between treatment and age effects (Fig. 1). Within baseline, plasma corticosterone was highest at birth and significantly decreased with age. The pattern of plasma corticosterone during normoxic separation (time control for hypoxia) was not different from baseline from birth through PD4. However, at PD5, PD6, and PD7, plasma corticosterone at 60 minutes of normoxic separation was significantly increased compared to baseline (large effect sizes [2.20; CI, 1.20-3.07], [1.80; CI, 0.87-2.62], and [1.32; CI, 0.48-2.09], respectively). Sixty minutes of hypoxic separation significantly decreased plasma corticosterone at birth compared to baseline and normoxic separation time control (medium effect sizes [0.71; CI, 0.06-1.33] and [0.74; CI, 0.08-1.37], respectively), but significantly increased plasma corticosterone compared to baseline and normoxic separation at all ages starting at 4 hours postbirth (large effect sizes [1.03; CI, 0.41-1.62] and [0.93; CI, 0.26-1.56], respectively) through PD7 (large effect sizes [6.75; CI, 4.75-8.37] and [1.8; CI 0.86–2.64], respectively). Notice that baseline corticosterone and the magnitude of the corticosterone response to hypoxia gradually decreased from a peak at 8H to PD7 consistent with the adrenal stress hyporesponsive period (44-46).

Figure 1.

Plasma corticosterone at baseline and in response to 60 minutes of normoxic or hypoxic separation in pups of both sexes from birth to postnatal day (PD)7. Values are mean ± SE. N values are baseline (16-21), normoxic separation (12-18), and hypoxic separation (14-24). ^aDifferent from baseline within age. ^bDifferent from hypoxic separation within age. ^cDifferent from birth within baseline via one-way analysis of variance. All significant differences noted on the figure by multiple pairwise analyses were subjected to effect size analysis by Cohen’s d. All significant differences presented had a large effect (Cohen’s d > 0.8), except for the following with medium effect sizes: within birth, hypoxic separation vs baseline; effect size = 0.71 and within birth, hypoxic separation vs normoxic separation; effect size = 0.74.

Fig. 2 shows the relative fold-change expression analysis of critical steroidogenic pathways gene mRNA in male and female rat pup adrenal glands. Fold change was calculated from C_q values using the 2^–ΔΔCt relative quantification method (40), with the nadir time point within baseline of each gene of interest as the calibrator. (C_q values are provided in Supplemental Table S1) (41). At birth, 2H, 4H, 8H, PD2, and PD6 were chosen for analysis because they reflect the major changes in baseline and stimulated plasma corticosterone levels from birth to PD7 (see Fig. 1). For most genes, 60 minutes of normoxic or hypoxic separation had little effect on adrenal steroidogenic mRNA expression; most changes occurred between ages. At birth, adrenal Ar, Ldlr, Mrap, and Star mRNA regardless of treatment were upregulated compared to subsequent age groups. Normoxic separation induced a significant increase in adrenal Ldlr and Star mRNA at birth despite no changes in plasma corticosterone. Although Cy17a1 showed a many-fold increase in PD2 and PD6 adrenals, there was a high variability of expression levels within each group of adrenal glands (PD2: median C_q: 27.6; range, 26.3-36.0; PD6: median C_q: 28.9; range, 26.7-34.5), so no significant difference between treatments were observed (Supplemental Table S1) (41).

Figure 2.

Adrenal messenger RNA expression of critical steroidogenic pathway genes at baseline and after 60 minutes of normoxic or hypoxic separation in pups of both sexes from birth to postnatal day (PD)7. Fold-change analysis was performed using the 2^–ΔΔCt relative quantification method with Actb (actin, beta) as the reference gene normalizing to the nadir time point within baseline of each gene of interest. Calibrator (denoted by arrow) was identified as the age group within baseline that had the lowest quantification cycle (C_q) mean value. N = 6 for each treatment and age. Ar, androgen receptor; Cyp11a1 (P450scc), cholesterol side-chain cleavage; Cyp17a1, 17-alpha-hydroxylase; Ldlr, low-density lipoprotein receptor; Mc2r, melanocortin 2 receptor (ACTH receptor); Mrap, melanocortin 2 receptor accessory protein; Mrap2, melanocortin 2 receptor accessory protein 2; Star, steroidogenic acute regulatory protein. Statistical analysis was performed on C_q values shown in Supplemental Tables S1 (41). ^aDifferent from baseline within age. ^bDifferent from normoxic separation within age. ^yAge group is different from all other age groups. ^zAge group is different from birth.

Male testosterone, follicle-stimulating hormone, and testis messenger RNA expression

Fig. 3 describes the effects of normoxic and hypoxic separation (compared to baseline) on male plasma FSH and testosterone concentrations from birth through PD7. Male LH data are not shown because more than 50% of the results were below the assay’s limit of detection. Within baseline, no differences were observed in plasma FSH concentrations. Baseline plasma testosterone was highest at birth; baseline testosterone decreased thereafter with no evidence of a testosterone surge at time points after birth.

Figure 3.

Plasma follicle-stimulating hormone (FSH) and testosterone at baseline and after 60 minutes of normoxic or hypoxic separation in male pups from birth to postnatal day (PD)7. Values are mean ± SE. N values are baseline (6-9), normoxic separation (6-8), and hypoxic separation (6-12). ^aDifferent from baseline within age. ^bDifferent from hypoxic separation within age. ^cDifferent from birth within baseline via one-way analysis of variance (ANOVA). *Different from hypoxic separation within age via one-way ANOVA. All significant differences noted on the figures by multiple pairwise analyses were subjected to effect size analysis by Cohen’s d. All significant differences presented had a large effect (Cohen’s d > 0.8).

Compared to baseline, normoxic separation for 60 minutes significantly decreased plasma FSH in male pups at 2 hours postbirth (2H; large effect size [0.8; CI, 0.2-1.72]) but significantly increased plasma FSH at PD3, PD5, and PD7 (large effect sizes [1.98; CI, 0.86-2.94], [3.25; CI, 1.41-4.63], and [0.93; CI, 0.19-1.93], respectively). Hypoxic separation resulted in a small but significant decrease in plasma FSH in male pups at PD2 compared to baseline and normoxic separation time control (large effect sizes [1.46; CI, 0.35-2.42] and [0.91; CI, 0.10-1.84], respectively).

Overall, the only significant effect of 60 minutes of hypoxia on plasma testosterone was an increase compared to baseline at birth (large effect size [1.5; CI, 0.46-2.44]). Interestingly, at 2 hours after birth, 60 minutes of normoxic separation also induced a significant increase in male plasma testosterone (large effect size [1.9; CI, 0.68-2.87]), but the significant increase did not occur during hypoxia with separation. There were no other significant changes in plasma testosterone due to normoxic or hypoxic separation.

Fig. 4 shows the relative fold-change expression of steroidogenic pathway gene mRNA in male rat pup testis calculated and presented as described for Fig. 2. As in the adrenal mRNA expression analysis, a subset of ages was chosen to reflect the major changes in plasma testosterone levels observed from birth through PD7 (see Fig. 3). Cyp11a1 (P450scc), Cyp17a1, Lhcgr (LH receptor), and Star mRNA in the testes were all highest at birth and gradually decreased with age, paralleling the decrease in plasma testosterone levels. Hypoxic separation for 60 minutes had minimal effects on testis mRNA expression, so most changes observed occurred between ages as in the adrenal mRNA analysis. At 4H and 8H after birth, hypoxic separation significantly decreased Ar, Fshr, Lhcgr, and Star mRNA. Additionally, hypoxia decreased Cyp11a1 (P450scc) and Hsd17b1 mRNA at 8H postbirth and Hsd3b mRNA at PD2. For the most part, normoxic separation had no significant effect on testis mRNA expression compared to baseline.

Figure 4.

Testes messenger RNA expression of critical steroidogenic genes at baseline and after 60 minutes of normoxic or hypoxic separation in male pups from birth to postnatal day (PD)7. Fold-change analysis was performed using the 2^–ΔΔCt relative quantification method with Rpl19 (ribosomal protein L19) as the reference gene normalizing to the nadir time point within baseline of each gene of interest. Calibrator (denoted by arrow) was identified as the age group within baseline that had the lowest quantification cycle (C_q) mean value. N = 5 for each treatment and age. Ar, androgen receptor; Cyp11a1 (P450scc), cholesterol side-chain cleavage; Cyp17a1, 17-alpha-hydoxylase; Fshr, follicle-stimulating hormone receptor; Hsd3b, 3-beta-hydroxysteroid dehydrogenase; Hsd17b1, hydroxysteroid 17-beta dehydrogenase; Lhcgr, luteinizing hormone/choriogonadotropin receptor; Star, steroidogenic acute regulatory protein. Note that statistical analysis was performed on C_q values shown in Supplemental Table S2. ^aDifferent from baseline within age. ^bDifferent from normoxic separation within age. ^yAge group is different from all other age groups. ^zAge group is different from birth.

Female follicle-stimulating hormone, luteinizing hormone, and estradiol

Fig. 5 shows the effects of normoxic and hypoxic separation (compared to baseline) on female plasma FSH, LH, and estradiol concentrations from birth to PD7. For most ages, there were minimal effects of normoxic and hypoxic separation on plasma FSH and LH. Baseline levels of plasma FSH increased with age compared to birth. From PD2 to PD7, baseline plasma FSH concentrations were all significantly increased compared to at birth (large effect sizes ranged from 1.18; CI, 0.21-2.05 [PD3] to 3.42; CI, 1.89-4.64 [PD4]). Note that the general pattern of FSH in females was similar to males over the same ages. Hypoxic separation significantly decreased plasma FSH at PD2 and PD4 (large effect sizes [1.53; CI, 0.55-2.39] and [2.55; CI, 1.28-3.60], respectively). At PD7, 60 minutes of normoxic and hypoxic separation both significantly increased plasma FSH concentrations compared to baseline (large effect sizes [1.41; CI, 0.15-2.48] and [0.76; CI, 0.38-1.80], respectively). Additionally, 60 minutes of normoxic separation significantly increased plasma FSH compared to hypoxic separation at PD7 (large effect size [0.96; CI, 0.30-2.07]). The only significant differences observed in plasma LH were a decrease in LH concentration from normoxic separation at PD5 (large effect size [1.59; CI, 0.35-2.65]) and a large increase in plasma LH concentration during 60 minutes of hypoxic separation at PD7 (large effect size [1.14; CI, 0.07-2.19]).

Figure 5.

Plasma follicle-stimulating hormone (FSH), luteinizing hormone (LH), and estradiol (E2) at baseline and after 60 minutes of normoxic or hypoxic separation in female pups from birth to postnatal day (PD)7. Values are mean ± SE. N values are baseline (6-9), normoxic separation (6-9), and hypoxic separation (6-12). ^aDifferent from baseline within age. ^bDifferent from hypoxic separation within age. ^cDifferent from birth within baseline via one-way analysis of variance. All significant differences noted on the figures by multiple pairwise analyses were subjected to effect size analysis by Cohen’s d. All significant differences presented had a large effect (Cohen’s d > 0.8).

Baseline plasma estradiol was highest within the first 8 hours of birth, significantly peaking at 8H, and subsequently decreasing by PD2-PD7 (large effect sizes ranging from 3.06; CI, 1.54-4.27 [PD3] to 7.34; CI, 4.52-9.45 [PD4]). Interestingly, 60 minutes of normoxic and hypoxic separation both decreased plasma estradiol concentrations at 8H postbirth (large effect size [1.43; CI, 0.30-2.41] and [2.62; CI, 1.26-3.73], respectively), significantly increased at PD2 (large effect sizes [2.28; CI, 0.96-3.36] and [5.69; CI, 3.42-7.42], respectively), and again significantly decreased at PD3 (large effect size [0.90; CI, 0.17-1.88] and [1.15; CI, 0.00-2.17], respectively). No other significant differences in estradiol were observed between treatments.

Discussion

The purpose of this study was to determine whether exposure to acute hypoxia augments the neonatal plasma testosterone surge in male rats, whether it is associated with changes in plasma gonadotropins and/or testicular enzyme mRNA expression, and then to describe its temporal relationship to the subsequent corticosterone response to stress using our neonatal (birth to PD7) rat model of human prematurity (26-28). The main findings in this study were as follows: 1) Only at birth did hypoxic separation decrease plasma corticosterone compared to baseline and normoxic separation. Thereafter, there was a large increase in plasma corticosterone with hypoxic separation starting at 4H postbirth through PD4 (the beginning of the adrenal stress-hyporesponsive period) that did not demonstrate sexual dimorphism. The small but significant increases in corticosterone in response to normoxic separation developed at the same time after birth as the start of the adrenal stress-hyporesponsive period. 2) Neither acute normoxic nor hypoxic separation had a significant effect on the expression of critical steroidogenic pathway genes in the adrenal; however, a decrease in baseline Cyp11a1 (P450scc), Mc2r, Mrap, and Star adrenal mRNA expression, in addition to a decrease in baseline plasma corticosterone from birth through PD7, confirms that the classic stress hyporesponsive period is due to a change in adrenal function rather than an effect at the hypothalamus or anterior pituitary. 3) At birth, hypoxic and normoxic separation significantly increased plasma testosterone in male pups that was independent of gonadotropins but associated with upregulated testicular steroidogenic mRNA. 4) A dramatic decrease in testicular Star, Lhcgr, and Cyp11a1 (P450scc) mRNA from a peak at birth through PD7 suggests a possible hyporesponsive period in the testes similarly to the adrenal cortex. 5) A peak in baseline plasma estradiol in females was observed at 8H postbirth that was attenuated by normoxic and hypoxic separation without significant changes in plasma gonadotropins. 6) Finally, plasma estradiol was very low by PD4 despite small increases in FSH and LH, also suggesting an ovarian steroidogenic hyporesponsive period.

Corticosterone and adrenal gene expression

The synthesis of the primary glucocorticoid in the rat, corticosterone, is stimulated by ACTH through the adrenal melanocortin 2 receptor. We have previously shown that hypoxic stress at PD2 induces a large increase in corticosterone that occurs independently of an increase in immunoassayable plasma ACTH but is mediated by an increase in bioactive ACTH (28). This immunoreactive ACTH-independent adrenal response to hypoxia at PD2 shifts to the classic immunoassayable ACTH-dependent response by PD8 (29, 47). Because of these prior studies in which ACTH bioactivity is not detected at PD2 by our plasma ACTH immunoassay, we chose instead to use the plasma obtained to measure gonadotropins and gonadal steroids rather than ACTH. In the present study, we observed a large hypoxia-induced increase in plasma corticosterone at 4H after birth, peaking at 8H through PD2 and decreasing over time through PD7. This response was not sexually dimorphic. It is interesting to note that the small but significant increase in corticosterone due to separation per se did not occur until the latter time period.

To investigate the mechanism of the corticosterone dynamics described previously, we used RT-qPCR to evaluate the expression of genes that encode critical proteins involved in adrenal secretagogue transduction and steroidogenesis (28, 31, 48, 49). We have previously reported that acute hypoxia induces a subtle increase in adrenal Ldlr, Mrap, and Star mRNA expression in the PD2 and PD7 neonate (28). These mRNA encode adrenocortical proteins that mediate ACTH receptor activation, cholesterol uptake from the extracellular space, and subsequent cholesterol transport into the mitochondria to be converted to pregnenolone by the first enzymatic step (Cyp11A1). We infer that these significant increases in mRNA expression are mediated by increased nonimmunoassayable, bioactive ACTH in the early postnatal period because these responses are blocked by our highly specific anti-ACTH neutralizing antibody in vivo (28). In the present study, profound differences in steroidogenic pathway mRNA expression was not induced by normoxic or hypoxic separation, except for a 3- to 10-fold increase in Ldlr and Star in adrenals from normoxic pups separated for 60 minutes starting immediately at birth. Our previous studies did not examine the effect of hypoxia earlier than PD2 (28, 29, 47). A closer inspection of the PD2 mRNA data in Fig. 2 shows small-fold increases in Cyp11a1, Ldlr, Mc2r, and Star due to hypoxic separation that are dwarfed statistically by larger stochastic changes in other age groups and are consistent in magnitude with our previous studies (29, 31).

Adrenal Ar, Cyp11a1, Ldlr, Mrap, and Star were all significantly increased at birth regardless of treatment compared to the subsequent age groups before PD2. Birth is a significant and stressful event in mammalian life, so it seems reasonable that a large increase in critical adrenal steroidogenic mRNA expression occurs during and immediately after birth to facilitate increases in corticosterone necessary to adapt to a stressful postnatal environment (50-52). It is of interest that Ldlr, Mrap, and Star mRNA expression significantly decreased at the time points after birth, paralleling the decrease in basal corticosterone in the baseline and normoxic separation groups. There is obviously still adequate steroidogenic gene expression demonstrated by the dramatic increases in corticosterone induced by hypoxia between 8H after birth and PD4. Furthermore, although PD5 to PD7 is during the adrenal stress hyporesponsive period, there was still a significant hypoxia-induced increase in corticosterone compared to baseline. These findings are striking and highlight the critically important role that endogenous glucocorticoids play in the metabolic adaptation to hypoxic stress in the neonate (53).

Neonatal testosterone surge

The early-life mammalian testosterone surge is well documented in numerous species including rats, mice, and humans (10-12). This testosterone surge in males is critical for normal central nervous system (CNS) development, sexual differentiation of the brain, and establishing reproductive behavior and sexually dimorphic gonadotropin secretion (13, 14). In most species, the surge is typically transient, declining rapidly within the first few hours of birth, and remaining low during the first few weeks to months of life (10). In rats, there are 2 early-life surges in testosterone: prenatally (gestation days 18 and 19) and then immediately after birth (1-4 hours) (10). To assess the immediate effect of normoxic and hypoxic separation on this postnatal surge, we evaluated numerous time points from birth through PD7 to encompass the first week of life.

Environmental stressors such as prenatal exposure to maternal stress, drugs, alcohol, and nicotine have been shown to demasculinize the behavior of male offspring mediated by an alteration in the neonatal testosterone surge (23, 54). Although the rat fetus is relatively resistant, postnatal hypoxic exposure and/or maternal-neonatal separation augments the neonatal testosterone surge (17, 55, 56). We observed a significant increase in circulating plasma testosterone after 60 minutes of normoxic or hypoxic separation immediately at birth. Additionally, although plasma LH was not detectible in the majority of samples from male pups, testicular LH receptor (Lhcgr) mRNA expression was significantly increased at birth and 2H postbirth. This is consistent with previous studies suggesting that the fetal and neonatal testosterone surge can occur independently of increases in plasma LH (57). Again, this perinatal testosterone surge is a critical factor in important developmental effects in the CNS (13, 14).

What might be the mechanism of this surge in circulating testosterone if it is not driven by pituitary gonadotropins? There is a paucity of previous studies examining the mechanism that generates the neonatal testosterone surge in rats. Ironically, chronic stress leading to elevated levels of ACTH may inhibit LH and testosterone secretion in the adult male rat (58). A hypoxia-induced, ACTH-mediated effect on circulating plasma testosterone that is independent of N-methyl-d-aspartate–mediated LH secretion has been proposed (17). This is consistent with the hypothesis that a neonatal testosterone surge mediated by a hypoxic-induced increase in bioactive ACTH is an additional layer of neuroprotection during critical perinatal brain development as the period of sexual differentiation of the CNS in rats (17). As described earlier, stress-induced increases in bioactive ACTH up to PD2 in the rat are not detectible by immunoassay, so we were not able to correlate hypoxia-induced increases in immunoreactive ACTH with the testosterone surge. However, we plan future studies of this concept using our novel and highly specific ACTH-neutralizing antibody in vivo (28).

In another study, the early postnatal surge in testosterone was partially attributed to a decrease in the clearance of testosterone at birth, and that environmental stress, such as hypoxia or maternal-neonatal separation, could have an effect on these processes (59). It is not possible in our study to differentiate whether normoxic or hypoxic separation in the first few hours of postnatal life increased testosterone secretion or decreased its clearance rate. The increase in testicular steroidogenic pathway mRNA observed at birth favors increased testosterone synthesis, which occurs independently of pituitary gonadotropins. Like neonatal bioactive ACTH (28), there may be a form of LH and FSH not detected by immunoassay but still bioactive (60). One example of such a modification could be a change in glycosylation of the pituitary gonadotropins (61).

Plasma estradiol

It has been purported that the rodent ovary does not secrete significant amounts of estradiol before PD7 (62) despite our findings that it is increased at birth through PD1. At least a portion of the plasma estradiol we measured in female rat pups from birth through PD2 could be from residual maternal-placental-fetal estrogen. It is also possible that residual placental chorionic gonadotropin stimulated neonatal ovarian function in the absence of an increase in pituitary gonadotropins (63, 64). That said, the decreased in plasma estradiol during normoxic and hypoxic separation at 8H after birth may indicate that separation could increase the clearance of estradiol and/or chorionic gonadotropin. The testosterone surge was increased at birth by separation and acute hypoxia in males, whereas estrogen was not increased by separation and acute hypoxia until PD2 in females. This suggests that the subsequent, nonsexually dimorphic corticosterone responses to stress at 4H to PD7 were probably not mediated by these prior changes in gonadal steroids. However, it is still possible that the stress-induced augmentation of the testosterone surge at birth is a mechanism for subsequent sexual dimorphism in the adult (1-5).

LH secretion in females gradually increases from PD12 to PD20 (65, 66). In female rats, LH release starts to occur in pulses by the third and fourth week of life (67). Although this is after the time points within the present study, it is possible LH may begin to have a pulsatile pattern earlier in postnatal life accounting for some of the changes in plasma estradiol we observed. Of most interest is the hypoxic-induced LH secretion in PD7 females (again without an increase in plasma estradiol), further supporting our conjecture of a gonadal hyporesponsive period.

The adrenal stress hyporesponsive period

The adrenal stress hyporesponsive period occurs in rodents from PD4 to PD14 (44, 68, 69). This period in early postnatal development results in a reduced capacity of the adrenal to increase corticosterone in response to stress and has been hypothesized to be a protective measure to ensure proper brain development (45, 70, 71). Although maternal-neonatal separation for short periods of time during the first few weeks of life occurs naturally in rats (72), longer periods of maternal-neonatal separation during this period elicit long-term programming effects on the HPA axis (73-76). We have previously confirmed in our rat model of prematurity that the adrenal stress hyporesponsive period occurs in PD7 to PD15 pups likely due to a decrease in adrenal sensitivity to ACTH (26, 47, 77). We have also previously demonstrated that the adrenal hyporesponsive period occurs in the absence of large increases in circulating ACTH (26, 29, 48, 77). In the present study, we observed an unexpected adrenal hyperresponsive period in response to hypoxic separation starting at 4H after birth and continuing through PD4. The high concentrations of corticosterone across each treatment group between birth and 2H after birth may be from residual maternal-placental-fetal glucocorticoids (78-80). This is consistent with a previous study showing that rat pups evaluated immediately after cesarean section at 22 days of gestation had significantly decreased plasma corticosterone levels within 1 hour (53). In our study, only at birth did hypoxic separation significantly decrease plasma corticosterone. It is possible that hypoxia at birth increased the clearance rate of plasma corticosterone, working through a mechanism no longer present by 2H after birth.

Normoxic and hypoxic separation did not have a profound effect on critical adrenal steroidogenic pathway mRNA expression, except for an overall increase in Ldlr and Star activity at birth. Of particular interest is the dramatic 6-fold increase in Ldlr mRNA at 60 minutes of normoxic separation at birth compared to baseline with a small increase in Star mRNA and no appreciable change in other critical steroidogenic genes or plasma corticosterone level. Although the adult rat adrenal does not express appreciable Cyp17 and cannot synthesize adrenal androgen precursors (81), it seems possible that Cyp17 expression occurs in the neonatal adrenal providing androgen precursors for local tissue activation to testosterone and dihydrotestosterone not reflected in plasma levels (82, 83). Previous studies have reported low levels of steroid products arising from 17-hydroxylation in the adrenal in vitro (84, 85). A more recent study has measured appreciable levels of Cyp17 mRNA in male and female rat adrenals from PD10 to PD25 that could be responsible for increased adrenal androgen precursors (82). In our study, adrenal Cyp17a1 mRNA was very low until PD2, when a 15-fold increase occurred in response to normoxic and hypoxic separation. It is important to note that the C_q values from Cyp17a1 were highly variable between samples (see Supplemental Table S1) (41), so further studies are needed to confirm this novel finding of an increase in Cyp17a1 expression.

We previously suggested that either locally produced adrenal androgens or testosterone from the testes may directly influence neonatal adrenocortical function (26). The present study found a small decrease in adrenal Ar mRNA expression further supporting a potential role for effects of neonatal androgens on adrenocortical function (86-88). As discussed earlier, testis mRNA analysis again revealed a remarkable decrease in baseline Cyp11a1 (P450scc), Cyp17a1, Hsd3b, Lhcgr, and Star from birth to PD6, paralleling the decrease in testosterone measured suggesting a hyporesponsive period in the testes that may begin immediately after the clearance of the neonatal testosterone surge. This dramatic decrease in gene expression analyzed in whole testes after birth could theoretically be due to a decrease in the number of Leydig cells per se or because of a theoretical increase in the proliferation of Sertoli cells, thus decreasing the percentage of steroidogenic Leydig cells. However, Sertoli cell number does not increase in the testes from birth through PD7 in the neonatal rat nor does Leydig cell number decrease (89, 90). In fact, there is an increase in fetal Sertoli cell number up to 21 days of gestation, but then there is a dramatic decrease in postnatal pup Sertoli cell number from birth through PD21 (91).

Perspectives

To summarize our main findings, we found that separation and hypoxia increased plasma testosterone only during the immediate postnatal testosterone surge, that this occurred in the absence of an increase in gonadotropins, and that relevant, associated testicular mRNA were also highest in the early postnatal period. This very early increase in testosterone during hypoxia was followed temporally by an augmented corticosterone response to hypoxia that was first evident at 4H after birth. We suggest that a large corticosterone response to hypoxia during the first 5 days of life is critical for the metabolic adaptation to neonatal stress. This is preceded by the testosterone surge that occurs during the first hours after birth independently of gonadotropins but is associated with upregulated steroidogenic pathway genes in the testes. The decrease in testosterone thereafter is likely important to allow normal brain and behavioral development and may be related to corticosterone effects (13, 14, 92). There are many studies that suggest the neonatal male may respond differently from females to stressors, which may be responsible for sex differences in the adult (3, 93-96). Because environmental stressors may alter the normal dynamics of the neonatal testosterone surge leading to long-term programming effects on the male, they are likely to have important clinical implications in the management of the premature hypoxic neonate. Only by careful evaluation of potential interactions between adrenal and gonadal steroids soon after birth can these implications be revealed.

Glossary

Abbreviations

ACTH

adrenocorticotropin

ANOVA

analysis of variance

cDNA

complementary DNA

Cq

quantification cycle

FSH

follicle-stimulating hormone

HPA

hypothalamic-pituitary-adrenal axis

HPG

hypothalamic-pituitary-gonadal axis

LH

luteinizing hormone

mRNA

messenger RNA

PD

postnatal day

qPCR

quantitative polymerase chain reaction

RT

reverse transcription

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability: All data generated or analyzed during this study are included in this published article or in the data repositories listed in the References.