Prolonged HPA axis dysregulation in postpartum depression associated with adverse early life experiences: A cross-species translational study

Abstract

Childhood and adolescent stress increase the risk of postpartum depression (PPD), often providing an increased probability of treatment refractoriness. Nevertheless, the mechanisms linking childhood/adolescent stress to PPD remain unclear. Our study investigated the longitudinal effects of adolescent stress on the hypothalamic-pituitary-adrenal (HPA) axis and postpartum behaviors in mice and humans. Adolescent social isolation prolonged glucocorticoid elevation, leading to long-lasting postpartum behavioral changes in female mice. These changes were unresponsive to current PPD treatments but improved with post-delivery glucocorticoid receptor antagonist treatment. Childhood/adolescent stress significantly impacted HPA axis dysregulation and PPD in human females. Repurposing glucocorticoid receptor antagonists for some cases of treatment-resistant PPD may be considered.

Extended Data Figure 1.. Experimental schedule of the preclinical study.

A, Virgin female mice were group-housed. B, Virgin female mice were isolated from 5 to 8 weeks of age. C, Virgin female mice were group-housed, mated with a male mouse, and gave birth to pups. D, Virgin female mice were isolated from 5 to 8 weeks of age, mated with a male mouse, and gave birth to pups. TST, tail suspension test; FST, forced swim test; SIT, three-chamber social interaction test. Different cohorts of mice subjected to behavioral tests at 0 week, 1 week, and 3 weeks postpartum were studied to avoid the repeated exposure to stressful behavioral procedures.

Extended Data Figure 2.. Long-lasting behavioral changes in the social interaction test in dams exposed to adolescent social isolation.

S Sniffing time and indexes of sociability and social novelty recognition during the three-chamber social interaction test were measured at postpartum days 0, 7, and 21. No changes in sociability or social novelty recognition among the four groups were observed at postpartum day 0. Behavioral changes in social novelty recognition, but not sociability, were observed in stressed dams at postpartum days 7 and 21. Postpartum day 0/Unstressed virgins, N=10; Postpartum day 0/Stressed virgins, N=9; Postpartum day 0/Unstressed dams, N=10; Postpartum day 0/Stressed dams, N=9; Postpartum day 7/Unstressed virgins, N=10; Postpartum day 7/Stressed virgins, N=10; Postpartum day 7/Unstressed dams, N=10; Postpartum day 7/Stressed dams, N=11; Postpartum day 21/Unstressed virgins, N=10; Postpartum day 21/Stressed virgins, N=10; Postpartum day 21/Unstressed dams, N=12; Postpartum day 21/Stressed dams, N=11. Values are represented as mean ± SEM; **P<0.01 and *P<0.05. See Supplementary Table 1 for details on the statistical analyses.

Extended Data Figure 3.. No change in the levels of plasma estradiol, progesterone, oxytocin, and prolactin in mice exposed to adolescent social isolation.

Levels of estradiol, progesterone, oxytocin, and prolactin in plasma were measured at four-time points (virgin, late pregnancy, 0 week postpartum, and 1 week postpartum). No differences in the levels of estradiol, progesterone, oxytocin, and prolactin were observed between unstressed and stressed mice at any time point. N=9–20 (precise values detailed in Supplementary Table 1). Values are represented as mean ± SEM. See Supplementary Table 1 for details on the statistical analyses.

Extended Data Figure 4.. Positive correlation between plasma corticosterone levels and immobility time in the tail suspension and forced swim tests at 1 week and 3 weeks after delivery.

The levels of plasma corticosterone (CORT) were positively correlated with immobility time in the tail suspension test (TST) and forced swim test (FST) at 1 week and 3 weeks after delivery. 1 week postpartum/Unstressed dams, N=9; 1 week postpartum/Stressed dams, N=10; 3 weeks postpartum/Unstressed dams, N=12; 3 weeks postpartum/Stressed dams, N=11. Spearman and Pearson rank correlation coefficients were examined for the data at 1 week and 3 weeks postpartum, respectively. See Supplementary Table 1 for details on the statistical analyses.

Extended Data Figure 5.. Dose-response effects of a SSRI, a GABA_A receptor modulator, or a GR antagonist on behavioral changes in dams exposed to adolescent social isolation.

Stressed dams were treated with a SSRI fluoxetine (18, 36, 52 mg/kg, p.o.) (A), a GABA_A receptor modulator ganaxolone (10, 20, 30 mg/kg, i.p.) (B), or a GR antagonist CORT113176 (40, 80, 160 mg/kg, p.o.) (C) once daily from postpartum day 0 to 24 h prior to sampling at postpartum day 9. Post-delivery treatment with only the GR antagonist at 40 and 80 mg/kg ameliorated the behavioral changes in the tail suspension test on postpartum day 7 in stressed dams. Post-delivery treatment with only the GR antagonist at 80 mg/kg ameliorated the behavioral changes in forced swim test on postpartum day 8 in stressed dams N=12. Values are represented as mean ± SEM; **P<0.01 and *P<0.05. See Supplementary Table 1 for details on the statistical analyses.

Extended Data Figure 6.. No effect of a GR antagonist, a SSRI, and an allopregnanolone analog on body weight.

A, CORT113176 (80 mg/kg, p.o., once daily from gestation day 14 to 24 h prior to behavioral testing), a selective GR antagonist, did not affect body weight after the forced swim test at postpartum day 8 in either group. B, Post-delivery treatment with a SSRI fluoxetine (18 mg/kg, p.o.), a GABA_A receptor modulator ganaxolone (10 mg/kg, i.p.), or CORT113176 (80 mg/kg, p.o.) once daily from postpartum day 0 to 24 h prior to behavioral testing did not affect body weight after the forced swim test at postpartum day 8 in any of the groups. N=12. Values are represented as mean ± SEM. See Supplementary Table 1 for details on the statistical analyses.

Extended Data Figure 7.. No effect of a GR antagonist on blood glucose levels and weight of the thymus, spleen, and visceral fat.

A, After fasting for 6 hours, glucose (2 g/kg) was administered intraperitoneally and blood glucose levels at 0, 30, 60, 90, and 120 minutes were examined on postnatal day 8. CORT113176 (80 mg/kg, p.o., once daily from postpartum day 0 to 24 h prior to the blood glucose testing at postpartum day 8), a selective GR antagonist, did not affect blood glucose levels in either group. B-D, Thymus, spleen, and visceral fat weights were measured after blood glucose testing at postpartum day 8. CORT113176 (80 mg/kg, p.o., once daily from postpartum day 0 to 24 h prior to the blood glucose testing at postpartum day 8) had no effect on weight of the thymus (B), spleen (C), and visceral fat (D) in any groups. Unstressed/Vehicle, N=6; Unstressed/GR antagonist, N=6; Stressed/Vehicle, N=6; Stressed/GR antagonist, N=7. Values are represented as mean ± SEM. See Supplementary Table 1 for details on the statistical analyses.

Figure 1.. Long-lasting behavioral changes in the tail suspension and forced swim tests in dams exposed to adolescent social isolation.

A, B, Immobility time (seconds) during the tail suspension (A) and forced swim tests (B) was assessed at postpartum days 0 and 1, postpartum days 7 and 8, and postpartum days 21 and 22, respectively. Behavioral changes emerged at 1 week postpartum and remained until at least 3 weeks postpartum. No changes across groups were observed immediately after delivery. Unstressed virgins, N = 10; Stressed virgins, N = 10; Unstressed dams, N = 12, Stressed dams, N = 10. Values are represented as mean ± s.e.m.; **P<0.01 and *P<0.05. See Supplementary Table 1 for details on the statistical analyses.

Figure 2.. Long-lasting behavioral changes in the sucrose preference test in dams exposed to adolescent social isolation.

Sucrose preference was measured at 1 week top, (postpartum days 7–10) and 3 week postpartum (bottom, postpartum days 21–24). Stressed dams showed a reduction in sucrose preference when given a choice between 1.5 % sucrose and water. N = 11. W/W, water/water. S/S, 1.5 % sucrose/1.5 % sucrose. S/W, 1.5 % sucrose/water. Values are represented as mean ± s.e.m.; **P<0.01 and *P<0.05. See Supplementary Table 1 for details on the statistical analyses.

Figure 3.. Prolonged elevation of corticosterone levels and altered glucocorticoid signaling in dams exposed to adolescent social isolation: its pivotal role in long-lasting changes of postpartum behaviors.

A, Levels of corticosterone were measured at five time points: virgin, late pregnancy, 0 week postpartum, 1 week postpartum, and 3 weeks postpartum. Corticosterone levels in stressed dams were consistently higher than those in unstressed controls at 1 and 3 weeks postpartum. Unstressed/Virgin, N = 30; Stressed/Virgin, N = 29; Unstressed/Late pregnancy, N = 10; Stressed/Late pregnancy, N = 14; Unstressed/0 week postpartum, N = 11; Stressed/0 week postpartum, N = 10; Unstressed/1 week postpartum, N = 9; Stressed/1 week postpartum, N = 10; Unstressed/3 week postpartum, N = 12; Stressed/3 week postpartum, N = 11. B, Causal effects of the GR on postpartum behavioral changes in the TST and FST. CORT113176 (80 mg/kg, p.o., once daily from gestation day 14 to 24 h prior to sampling at postpartum day 9) ameliorated the behavioral changes in the tail suspension and forced swim tests in stressed dams at postpartum days 7 and 8, respectively. The antagonist did not affect behavior in the TST and FST in unstressed dams. Unstressed/Vehicle, N = 17; Unstressed/GR antagonist, N = 16; Stressed/Vehicle, N = 14; Stressed/GR antagonist, N = 14. C, Differences in CRHR and GR mRNA levels in the pituitary, but not PVN, between unstressed and stressed dams. Differences in the mRNA levels of CRHR and GR in the pituitary were observed between unstressed and stressed dams at postpartum day 9. No difference in the mRNA levels of CRH and GR in the PVN was observed between unstressed and stressed dams at postpartum day 9. There was no difference in the mRNA levels examined in this study between unstressed and stressed virgin mice. N = 7. D, Changes in HPA axis negative feedback in stressed postpartum dams. In the dexamethasone (DEX) suppression test, DEX (0.1 mg/kg, i.p.) was administered at postpartum day 8. Treatment with low dose DEX reduced the levels of ACTH and corticosterone in unstressed dams, whereas it failed to suppress these hormones in stressed dams. N = 11. Values are represented as mean ± s.e.m.; **P<0.01 and *P<0.05; ^##P<0.01 versus unstressed virgins; ⁺⁺P<0.01 versus stressed virgins. See Supplementary Table 1 for details on the statistical analyses.

Figure 4.. Normalization of postpartum behavior changes in dams exposed to adolescent social isolation following a post-delivery treatment with a GR antagonist.

Mice were treated with an SSRI fluoxetine (18 mg/kg, p.o.), a GABA_A receptor modulator ganaxolone (10 mg/kg, i.p.), or a GR antagonist CORT113176 (80 mg/kg, p.o.) once daily from postpartum day 0 to 24 h prior to sampling at postpartum day 9. Post-delivery treatment with only the GR antagonist ameliorated the behavioral changes in the tail suspension and forced swim tests in stressed dams. N = 12. Values are represented as mean ± s.e.m.; **P<0.01; ^##P<0.01 versus unstressed dams treated with the SSRI; ⁺⁺P<0.01 versus unstressed dams treated with the GABA_A receptor modulator; ^$ $P<0.01 versus unstressed dams treated with the GR antagonist. See Supplementary Table 1 for details on the statistical analyses.

Figure 5.. A link between early life stress, a sustained increase in the glucocorticoid signaling, and PPD in humans.

A, Significance of adverse early life events on PPD in human subjects. Participants with a history of mental illness were more likely to be diagnosed with PPD than those with other risk factors. The white and black slices of the pie charts show the percentages of participants without and with PPD, respectively. N = 116. B, Risk factors for PPD in human subjects. A history of mental illness was the most important risk factor with the highest effect size positively correlated and linked to the development of PPD compared to childhood behavior, home environment, and traumatic events. N = 116. C, A prolonged elevation in plasma cortisol levels in PPD patients with a history of mental illness. Participants with a history of mental illness and no diagnosis of PPD showed a significant decline in cortisol levels after delivery. Participants with a history of mental illness and PPD showed a sustained elevation of cortisol levels until at least 6 weeks postpartum. No PPD, second trimester, N = 38; no PPD, third trimester, N = 56; no PPD, 2 weeks postpartum, N = 60; no PPD, 6 weeks postpartum, N = 60; PPD, second trimester, N = 10; PPD, third trimester, N = 14; PPD, 2 weeks postpartum, N = 15; PPD, 6 weeks postpartum, N = 15. Values are represented as mean ± s.e.m.; **P<0.01 and *P<0.05. See Supplementary Table 1 for details on the statistical analyses.