Stress and Androgen Activity During Fetal Development

Abstract

Prenatal stress is known to alter hypothalamic-pituitary-adrenal axis activity, and more recent evidence suggests that it may also affect androgen activity. In animal models, prenatal stress disrupts the normal surge of testosterone in the developing male, whereas in females, associations differ by species. In humans, studies show that (1) associations between prenatal stress and child outcomes are often sex-dependent, (2) prenatal stress predicts several disorders with notable sex differences in prevalence, and (3) prenatal exposure to stressful life events may be associated with masculinized reproductive tract development and play behavior in girls. In this minireview, we examine the existing literature on prenatal stress and androgenic activity and present new, preliminary data indicating that prenatal stress may also modify associations between prenatal exposure to diethylhexyl phthalate, (a synthetic, antiandrogenic chemical) and reproductive development in infant boys. Taken together, these data support the hypothesis that prenatal exposure to both chemical and nonchemical stressors may alter sex steroid pathways in the maternal-placental-fetal unit and ultimately alter hormone-dependent developmental endpoints.

Over the past several decades, fetal programming has emerged as an important model for understanding health development across the lifespan. According to this model, in utero exposures can play a significant role in shaping future physiology and there is particular interest in how prenatal stress¹ may influence fetal development, leading to altered childhood outcomes. Thousands of studies in humans and animal models have examined these issues, focusing on outcomes ranging from immune function (1, 2) to metabolism (3, 4). The influence of maternal stress can be seen even during fetal development (5, 6) and can be long-lasting, persisting into adulthood (7, 8). Much of the existing research on fetal programming related to maternal stress has centered on the hypothalamic-pituitary-adrenal (HPA) axis, particularly glucocorticoids. There is a compelling body of animal literature suggesting that when a dam is subjected to stressors (such as prenatal restraint), the resulting pups exhibit altered behavior and neurobiology, including dysregulation of the HPA axis (9, 10). In humans, however, associations between maternal stress (and related constructs) and cortisol regulation are often inconsistent or modest (11–13).

The emphasis on the HPA axis has overshadowed the possibility that other hormonal systems may be affected by prenatal stressors, either directly or through their interactions with the HPA axis. Sex steroid hormones are an obvious candidate, and, indeed, amniotic cortisol and testosterone concentrations are correlated (14). In both animal models and humans, the relationship between prenatal stress and childhood outcomes is frequently sex dependent. In rodents, males are typically affected more than females, resulting in stronger associations with postnatal outcomes including (but not limited to) body size (15), behavior (16), and brain development (17). In humans, there are notable sex differences in the prevalence of certain disorders linked to prenatal stress, including autism spectrum disorders, schizophrenia, and attention deficit hyperactivity disorder (18–20). Prenatal stress is also associated with asthma and depression, conditions for which the patterns of prevalence shift at puberty, concurrent with the dramatic rise in gonadal steroid production (21–24). Thus, there is suggestive evidence that (1) the sexes may differ in their vulnerability or sensitivity to the effects of prenatal stress and perhaps differently at different times in development and (2) there may be associations between prenatal stress and gonadal steroid activity, prenatally or postnatally. Unfortunately, in humans, directly measuring stress-related changes in in utero sex steroid activity is currently not feasible, and studying the downstream sequelae (which are arguably most relevant at reproductive maturity) is logistically difficult. Thus, creative approaches and new methods are needed to unravel the complex ways in which stress may act on the fetus.

Prenatal Stress and Sex-Dependent Development in Animal Models

Animal models have been widely used to understand how prenatal exposures affect sex steroid activity and reproductive development. For instance, exposure to the environmental chemical diethylhexyl phthalate (DEHP) disrupts prenatal androgen activity. As a result, at birth, exposed males often have reproductive anomalies and deficits including cryptorchidism, hypospadias, lower sperm counts, and lower testosterone levels (25–27). They may also exhibit feminization or demasculinization of sexual (eg, lordosis) and sex-dependent (eg, play behavior and aggression) behaviors that are concomitant with changes in sexually dimorphic brain nuclei (28, 29). Interestingly, many of the changes observed in DEHP-exposed male rodents are mirrored by those observed in response to prenatal stressors. Compared with controls, prenatally stressed males have lower testes weight, feminized play behavior, delayed puberty, altered mating behavior, lower adult testosterone levels, and decreased fertility (30–33). Some of the behavioral changes, particularly those related to attention and anxiety-related behaviors, may be ameliorated by postnatal testosterone administration, further suggesting disruption of androgen pathways (34). Notably, both exposures (prenatal DEHP and stress) result in shortened anogenital distance in male rodents. AGD, the distance from the anus to the genitals, has emerged as a particularly useful tool for examining the impact of prenatal exposures that affect androgen activity in that it (1) is a sensitive, continuous marker of prenatal androgen activity early in gestation and (2) can be measured in both sexes at birth or any time postnatally (25, 35). For these reasons, it is widely used as a key endpoint in reproductive toxicology (36). Shortened AGD in males is indicative of prenatal androgen disruption, due to, for instance, suppression of the testosterone surge that typically occurs during the male programming window early in gestation (37, 38).

AGD may also be informative about the prenatal hormonal milieu in females, who typically have very low prenatal androgen exposure and AGD that is 50% to 100% shorter than that in males (39, 40). Exposure to a hyperandrogenic prenatal environment results in masculinization of AGD, increased male-typical behavior, and decreased female-typical behavior (41–44). In contrast to the relatively consistent literature on the antiandrogenic effects of prenatal stress in male rodents, research in females is marked by widely variable results depending on the species studied, the paradigm used, and other factors. For example, in rats (but not mice) prenatally stressed females have decreased fecundity and fertility (45, 46). In guinea pigs, prenatal stress results in accelerated growth and reproductive maturity as well as more regular cycling (47), whereas in mice, vaginal opening is delayed and cycles become longer (48, 49). Even strain may matter; prenatally stressed females of some mouse strains exhibit increased aggressive behavior, whereas other strains do not (50, 51). Therefore, it is not surprising that various studies have reported positive (52), negative (51), and null (30) relationships between prenatal stress and AGD in female offspring.

Prenatal Stress and Sex-Dependent Development in Humans

That the sexes may differ in their response to prenatal stressors or insults is predicted by evolutionary theory and, as we have seen, supported by the animal literature demonstrating a consistent vulnerability among males in contrast to the highly variable effects in females (53, 54). Male vulnerability is also clinically evident: males are more likely to be born preterm and have poorer neonatal outcomes, particularly after complicated pregnancies (55–57). After catastrophic events (ie, natural and environmental disasters or terrorist attacks), the secondary sex ratio drops (58–60), most likely due to a disproportionate loss of vulnerable male fetuses (61). Whether androgens contribute to the male fetus' increased vulnerability to stressors is an unanswered question; however, their potential role is supported by experimental evidence that prenatal androgenization of female fetuses results in (1) adverse neurodevelopmental effects in response to prenatal stressors that are typically only seen in males (62), (2) a male-typical, blunted HPA axis response to stressors (63), and (3) intrauterine growth restriction (64). Interestingly, in humans, an elevated maternal androgen concentration during pregnancy has been associated with adverse outcomes including lower birth weight (adjusting for infant sex and other covariates) (65) and recurrent miscarriage (66), both of which are also linked to stress (67). Women with polycystic ovary syndrome, who have chronically high androgen levels, furthermore, may also be at elevated risk of miscarriage and perinatal complications compared with the risk of reproductively healthy women (68). Although conventionally it is believed that the human fetus is protected from excess maternal androgens, more recent evidence questions this assumption (69).

If prenatal psychosocial stress alters the maternal and/or fetal androgen milieu in humans, we would predict subsequent changes in androgen-sensitive infant and child development as well as sex differences in the response to stressors. In a study of pregnant Israeli women who experienced frequent and unpredictable rocket attacks, the odds of preterm birth, low birthweight, small for gestational age, and small head circumference were increased in female infants, but not in male infants, compared with those of a control group (70). The secondary sex ratio was also lower in the exposed population, suggesting that the most affected male fetuses may have been culled. Similar sex-dependent results were found in a study of birth outcomes following a major Chilean earthquake (71). With the lens of sex differences in mind, Davis and Sandman recently reanalyzed their previous work on prenatal psychosocial and biological stress in relation to child neurodevelopment, finding that most associations were stronger in (or even limited to) girls (72). Finally, results from another natural disaster—a major ice storm in Quebec—demonstrated that some stress-related sex differences in offspring outcomes persist into later childhood. Higher levels of maternal subjective distress in response to the storm were associated with poorer motor function at age 5 (73) and wheezing, asthma, and corticosteroid use among girls at age 12, but not boys (21).

If prenatal stress influences the in utero androgenic environment, then reproductive function should be particularly affected. Unfortunately, the epidemiological literature is very sparse. In one study, Finnish girls who gestated during the nearby Chernobyl disaster had higher cortisol and testosterone levels in adolescence than girls born a year later, and it was estimated that the stressful exposure accounted for 18% of the variance in testosterone levels (74). On the other hand, few reproductive changes have been observed in women who gestated during the Dutch Hunger Winter famine, although outcome assessment was focused on fertility measures (eg, completed family size and interpregnancy interval) rather than biomarkers (75).

Stressful Life Events and Sex-Dependent Development in a Pregnancy Cohort

AGD, a biomarker that has been linked to both the prenatal hormone milieu and to clinical reproductive outcomes, may provide a unique “bridge” between prenatal stress and its long-term consequences on reproductive function and fertility (76–78). We examined associations between prenatal stress and AGD in the Study for Future Families. Pregnant women and their male partners were recruited and asked about stressful life events (SLEs) that occurred during those pregnancies, including job loss and unemployment (self or partner); death of close family members; divorce, separation, or serious relationship issues; serious legal or financial problems; or serious injury or illness (self or partner) (79–81).² AGD was measured in infants following protocols described elsewhere (82). In multivariable models, the infant daughters of couples experiencing more SLEs (4+) tended to have longer AGD than the daughters of couples reporting fewer SLEs (β = 2.61, P = .015) (Figure 1) (83). In contrast, greater exposure to SLEs was nonsignificantly associated with shorter AGD in infant males (β = −2.01, P = .23). Further follow-up with the cohort indicated that girls exposed to more SLEs in utero also exhibited more masculine play behavior in early childhood than girls exposed to fewer SLEs (84).

Figure 1.

AGD in relation to prenatal exposure to stressful life events.

Numerous epidemiological studies have shown that as in animal models, prenatal exposure to the phthalate DEHP is associated with shorter AGD in infant boys (85–88). Given that prenatal stress and prenatal phthalate exposure may both influence human AGD, we used Study for Future Families data to examine whether prenatal SLE exposure modifies associations between maternal DEHP metabolite concentrations and infant AGD. In adjusted models³ stratified by parental SLEs (<4 vs. 4+), associations between DEHP metabolite concentrations and two measures of AGD (anopenile distance and anoscrotal distance) were stronger in the sons of low stress couples than in those of higher stress couples (Table 1). These findings suggest a possible “protective effect” of stress on the impact of DEHP on AGD, which is consistent with the hypothesis that stress may have androgenic effects in humans. Similar models examined stress-phthalate interactions in relation to analogous AGD measures in female infants; however, there were no associations (not shown), which is not surprising given that most studies have not found associations between any phthalate metabolite and AGD in infant girls. Given the small sample size, we consider these results to be suggestive, but preliminary, and are currently replicating these analyses in a second, larger birth cohort.

Table 1.

Boys' AGD in Relation to Maternal DEHP Metabolite Concentration by Level of Prenatal Stress^a

DEHP Metabolite	High Stress (n = 28)		Low Stress (n = 77)
	AGD-AS	AGD-AP	AGD-AS	AGD-AP
MEHP	−0.39 (.72)	−0.82 (.37)	−0.61 (.41)	−1.84 (.04)
MEHHP	−0.60 (.68)	−1.24 (.37)	−1.95 (.02)	−2.87 (.005)
MEOHP	−0.45 (.75)	−1.90 (.16)	−1.99 (.02)	−3.12 (.003)

Abbreviations: AGD-AP, anogenital distance from anus to penis; AGD-AS, anogenital distance from anus to scrotum; MEHP, mono-(2-ethylhexyl) phthalate; MEHHP, mono-(2-ethyl-5-hydroxyhexyl) phthalate; MEOHP, mono-(2-ethyl-5-oxohexyl) phthalate. Data are β (P values) in models controlling for center, creatinine, age at examination and weight for age.

^aHigh stress represents 4+ SLEs reported during pregnancy by woman and partner combined; low stress = <4 SLEs reported during pregnancy by woman and partner combined.

Possible Mechanisms

If stress influences the prenatal hormonal milieu, thereby altering sex-dependent development, additional research is needed to understand the underlying mechanisms. Unfortunately, the traditional rodent model may be inadequate, given that murine and human sex steroid production are quite different, particularly during pregnancy. Clinical conditions characterized by atypical steroid production may be informative. Girls with congenital adrenal hyperplasia, who experience supranormal adrenal androgen concentrations during gestation due to mutations in the cortisol synthesis pathway, have longer AGD than control girls in infancy as well as masculinized gender identity and behavior later in childhood (89–93). Thus, masculinization of reproductive and neurological development can occur through adrenal pathways. In normal pregnancies, adrenal androgen activity is stimulated by HPA axis hormones (CRH and ACTH) (94); when prenatal stress activates the fetal HPA axis, adrenal androgen production may also be up-regulated. Notably, adrenal androgen production starts in the late first trimester, coincident with the reproductive programming window (95). Although adrenal androgens are weaker than testicular androgens (96), stress-related up-regulation might nevertheless result in subtle masculinization in girls and counteract DEHP-induced testicular androgen production in boys. This would also help to explain why in murine species (which do not produce adrenal androgens), the relationship between prenatal stress and female reproductive development is unclear, whereas in guinea pigs (which produce adrenal androgens), prenatal stress is associated with earlier onset of estrus, more regular cycling, masculinized sexual and social behavior, increased testosterone levels, and male-typical patterns of sex steroid receptors in sexually dimorphic brain loci (47, 97, 98). Additional interspecies differences that may be relevant to consider in the context of sex steroid production include the lack of placental estrogen production in murine species as well the lack of placental chorionic gonadotropins (eg, human chorionic gonadotropin, which is needed to stimulate testicular androgen production) in all mammals except for primates and equids (99). Thus, finding a suitable animal model, such as a nonhuman primate, may be necessary to elucidate the mechanisms by which stress may affect sex steroid pathways.

Conclusion

Ultimately additional research is needed to better understand the ways in which prenatal stress may alter hormone systems. There is growing evidence that the effects are not limited to the HPA axis alone and may also include sex steroid pathways. Although we have mainly considered the potential effects of both phthalates and stress on androgenic pathways, other sex steroid mechanisms may be important as well. For instance, examining the androgen to estrogen ratios and aromatase activity may be useful. Moving forward, it will also be important to characterize stress-related changes in gene expression and epigenetic modification of steroidogenic pathways in the mother, fetus, and placenta. Ultimately, better understanding of the mechanisms by which prenatal stress may act on sex steroid pathways will help to further clarify the origins of sex differences in health and disease.