Developmental estrogen exposures and disruptions to maternal behavior and brain: effects of ethinyl estradiol, a common positive control

Abstract

Due of its structural similarity to the endogenous estrogen 17β-estradiol (E2), the synthetic estrogen 17α-ethinyl estradiol (EE2) is widely used to study the effects of estrogenic substances on sensitive organs at multiple stages of development. Here, we investigated the effects of EE2 on maternal behavior and the maternal brain in females exposed during gestation and the perinatal period. We assessed several components of maternal behavior including nesting behavior and pup retrieval; characterized the expression of estrogen receptor (ER)α in the medial preoptic area (MPOA), a brain region critical for the display of maternal behavior; and measured expression of tyrosine hydroxylase, a marker for dopaminergic cells, in the ventral tegmental area (VTA), a brain region important in maternal motivation. We found that developmental exposure to EE2 induces subtle effects on several aspects of maternal behavior including time building the nest and time spent engaged in self-care. Developmental exposure to EE2 also altered ERα expression in the central MPOA during both early and late lactation and led to significantly reduced tyrosine hydroxylase immunoreactivity in the VTA. Our results demonstrate both dose- and postpartum stage-related effects of developmental exposure to EE2 on behavior and brain that manifest later in adulthood, during the maternal period. These findings provide further evidence for effects of exposure to exogenous estrogenic compounds during the critical periods of fetal and perinatal development.

Introduction

The display of maternal behavior and interactions between the mother and her offspring are critical for development and health in rodents and humans alike (Batten et al., 2004; Francis et al., 1999b; Gilbert et al., 2009). Maternal behavior integrates neuroendocrine and physiological systems (Barrett and Fleming, 2011; Bowlby, 1951; Pereira and Ferreira, 2015; Rosenblatt, 1994), involves the interaction of numerous endogenous and environmental factors (Bale et al., 2010; Barrett and Fleming, 2011; Bridges, 2015) and is influenced by estrogen signaling (Pfaff et al., 2011). Furthermore, displaced or dysfunctional maternal behavior can have severe and long lasting neuropsychiatric and medical consequences for children who experience abuse and neglect (Felitti et al., 1998; Gilbert et al., 2009; Gunnar and Fisher, 2006).

The role of endogenous estrogens in maternal behavior, as well as the effects of low doses of exogenous estrogens in the disruption of these behaviors, remains in question. In the laboratory rat, classical studies have shown that at parturition, a decrease in progesterone followed by an increase in 17β-estradiol (E2) is necessary for the onset of maternal behavior, with prolactin, oxytocin and maternal-offspring interactions sustaining its display thereafter (Bridges et al., 1985; Lonstein, 2007; Morishige et al., 1973; Numan, 2003; Rosenblatt et al., 1988; Shaikh, 1971; Siegel and Rosenblatt, 1978). In laboratory mice, nulliparous females are frequently considered spontaneously maternal, which is taken to suggest a limited role for estrogen in maternal care in this species; pup exposure induces pup retrieval and other maternally relevant behaviors in ovariectomized and aromatase knockout mice (Stolzenberg and Rissman, 2011). However, virgin Swiss mice have also been found to commit infanticide in an intruder test (Parmigiani et al., 1999). Furthermore, a role for hormonal influence in mouse maternal behavior cannot be ruled out because in wild caught mice, ~60% of nulliparous females and ~90% of females during late pregnancy commit infanticide, a response which was found to be mediated by oxytocin. Interestingly, infanticide does not continue after parturition (McCarthy et al., 1986).

Additionally, there is evidence for a role of E2 in the establishment and display of maternal behavior in laboratory mice (Gandelman, 1973; Hauser and Gandelman, 1985; Stolzenberg and Rissman, 2011); ovarian hormones have been shown to mediate maternal motivation (Hauser and Gandelman, 1985). Estrogen receptor (ER)α knockout females display poor maternal behavior (Couse et al., 2000; Ogawa et al., 1998) and conditional silencing of ERα in the medial preoptic area (MPOA), a region of the forebrain critical for maternal care, abolishes maternal behavior (Ribeiro et al., 2012). Additionally, pregnant and lactating females have been shown to have increased ER immunoreactivity in the MPOA compared to virgins, with increased receptor density in lactating females (Koch and Ehret, 1989), findings which suggest that ERα is important for the display of mouse maternal behavior.

In non-human primates, circulating concentrations of E2 are similar in abusive and non-abusive mothers both before parturition, when levels are high (500–900 pg/ml), and after parturition, when levels are lower (<200 pg/ml) (Maestripieri and Megna, 2000a). Yet, rhesus macaque mothers with greater frequencies of abusive behaviors have higher E2:progesterone ratios at the end of pregnancy (Maestripieri and Megna, 2000b), suggesting a role for more complex hormone profiles in the quality of maternal care. In humans, a small study recently revealed polymorphisms in Esr1, the gene encoding ERα, that were associated with negative parenting in mothers (Lahey et al., 2012). These findings are consistent with a role for estrogen in human maternal behavior although the functional consequences of these polymorphisms have not yet been explored.

Endocrine disrupting chemicals (EDCs) are compounds that interfere with hormone signaling (Zoeller et al., 2012) by affecting the synthesis, secretion, transport, binding, action, or elimination of natural hormones (Kavlock et al., 1996). Many of these chemicals have been shown to mimic the actions of estrogen via interactions with ERs (FDA, 2010). Although maternal behavior is not well studied in traditional toxicological evaluations of EDCs, several studies have shown disruptions to maternal care after adult or developmental exposures to a range of EDCs (reviewed in (Catanese et al., 2015; Palanza et al., 2002b; Palanza et al., 2016; Walker and Gore, 2011)).

17α-ethinyl estradiol (EE2) is the active estrogenic component found in oral contraceptives, used by 100 million women worldwide (Petitti, 2003; Pletzer and Kerschbaum, 2014). Due to its affinity for ERα (Anstead et al., 1997; Blair et al., 2000), its structural similarity to the endogenous estrogen E2, and its oral bioavailablity, EE2 has been used to study the effects of estrogens on sensitive organs at multiple stages of development. In fact, EE2 is a common positive control used in studies of other putative EDCs with estrogenic properties (vom Saal et al., 2005).

Two studies of female rats produced conflicting results on the effects of EE2 on maternal behavior after exposures during pregnancy (Arabo et al., 2005; Dugard et al., 2001). In the first study, pup retrieval was assessed; in the latter, pup retrieval and observations on the nest evaluating direct and indirect interactions with offspring (e.g. carrying, licking, moving on the nest) and self-directed behaviors (e.g., eating, drinking) were examined. In both studies, females were injected daily on days 9–14 of pregnancy with 15 μg EE2/kg/day. In addition to the high reproductive toxicity induced by this dose, these two studies produced conflicting effects on maternal behavior even though they were conducted by the same research group and used a similar experimental design. The contradictory results reported made it difficult to draw conclusions on the effects of exogenous estrogens on maternal care and led us to evaluate the effects of EE2 on maternal behaviors. We first assessed maternal behavior and brain in CD-1 mice exposed to low doses during pregnancy and lactation (Catanese and Vandenberg, 2017b). We found that females exposed to 0.01 or 1 μg EE2/kg/day from pregnancy day 9 through lactational day 21 had no significant disruptions to maternal behaviors, although EE2 treatment induced a significant reduction in tyrosine hydroxylase positive cells in the ventral tegmental area (VTA), a brain region important for maternal motivation; these females also spent more time displaying stereotypy behaviors (e.g., repetitive tail retrievals to the nest).

Although these studies suggest that the effects of EE2 on maternal behavior in exposed adult female rodents may be subtle or absent, prior evaluations examining neurobehaviors in offspring developmentally exposed to EE2 indicate that the pups experience detrimental effects, including those that are detected later in adulthood. In one of the rat studies examining maternal behavior in EE2 exposed mothers, the F1 offspring were found to exhibit behavioral changes including increased spontaneous motor activity, decreased exploration, increased anxiety-like behavior, as well as changes in cognitive processing in adulthood (Dugard et al., 2001). The second study examining maternal behavior in exposed rat dams revealed that EE2 increased anxiety-like and depressive-like behaviors in F1 offspring (Arabo et al., 2005). In a more recent study, female California mice exposed to EE2 during the gestational and perinatal period spent less time nursing and grooming pups, and more time away from the nest compared to controls (Johnson et al., 2015). In another recent study, female Swiss mice were exposed to 0.1 and 1 μg EE2/kg body weight/day from gestational day 10 through postnatal day 40. In addition to effects on reproductive and anxiety-like behaviors, maternal behavior was examined in nulliparous females who spontaneously display maternal behaviors (Derouiche et al., 2015). Interestingly, mice exposed to the lower dose of EE2 had longer latencies to retrieve pups and females from both EE2 groups spent more time in non-pup directed activities. However, these mice were nulliparous, making it difficult to adequately assess effects of developmental EE2 exposures on reproductively associated maternal behaviors.

Here, we examined effects of low doses of EE2 on maternal behaviors in female CD-1 mice exposed in utero and during the perinatal period [the F1 generation, raised from the mothers examined in our previous study (Catanese and Vandenberg, 2017b)]. Based on studies demonstrating that developmental exposures to estrogenic chemicals produce effects with long latency (Heindel and Vandenberg, 2015; Zoeller et al., 2012), we hypothesized that exposure to EE2 during early life would induce deleterious effects that manifest in the context of parenting behavior. In addition to effects on maternal behavior, we examined the effects of developmental EE2 exposures on expression of ERα in the MPOA, a brain region critical for the display of maternal behavior, and dopaminergic neurons in the VTA, a brain region receiving functional input from the MPOA. The VTA is implicated in maternal motivation (Numan, 2007; Numan and Stolzenberg, 2009) and there is evidence that projections from the MPOA to the VTA are responsive to estrogen (Fahrbach et al., 1986; Morrell et al., 1984), indicating that the VTA may be sensitive to endocrine disruption. To our knowledge, this is the first study to examine the effects of developmental low dose EE2 exposures on both the maternal brain and maternal behavior. Because we used the same methods to evaluate the F0 and F1 generation, and also used these methods in studies of another estrogenic EDC (Catanese and Vandenberg, 2017a), the completion of this study allows us to compare two critical periods and the effects of two xenoestrogens.

Methods

Animals

Timed pregnant female CD-1 mice (Charles River Laboratories, Stoneridge, NY), were acclimated for at least two days and individually housed in polysulfone cages (until parturition) with food (ProLab IsoDiet) and tap water (in glass bottles) provided ad libitum. The animals were maintained in temperature (23 ± 2°C), humidity (40 ± 10%) and light controlled (12h light, 12h dark, lights on at 0800 h) conditions at the University of Massachusetts Amherst Central Animal Facility. All experimental procedures were approved by the University of Massachusetts Institutional Animal Care and Use Committee.

From pregnancy day 9 – lactational day 20, F0 dams were provided a small wafer (Nabisco, East Hanover, NJ) treated with EE2 (Sigma Aldrich, St. Louis, MO; >98% purity) or vehicle alone (70% ethanol, allowed to dry prior to feeding) (Catanese and Vandenberg, 2017b; Gauger et al., 2007; Zoeller et al., 2005). Wafers were dosed with solutions designed to deliver 0.01 or 1 μg EE2/kg/day (n=12–16 for each dose). Dams were allowed to deliver naturally (birth designated lactational day [LD] 0) and litters were culled to 10 pups on LD1.

Pups exposed to EE2 during gestation and perinatal development (the F1 generation) were weaned on postnatal day (PND)21. F1 female pups from all treatment groups were then housed with same-sex littermates until 9 weeks of age. At that time, two females per litter were mated with proven fertile males (untreated, purchased from Charles River Laboratories). Pregnancy was verified by the presence of a vaginal plug. One F1 female was euthanized on LD2. The second F1 female was assessed for maternal behavior on LD2, 7, and 14 and euthanized on LD21. All animals were euthanized by carbon dioxide asphyxiation followed by decapitation.

Maternal behavior assays

Maternal behavior was assessed on LD2, 7, and 14 using quantitative measures described previously (Catanese et al., 2015). First, maternal behaviors were observed for a period of 90-minutes without disturbing the dams. Observations began at the start of the light phase and the frequency of behaviors were recorded every three minutes for the following measures: dam position on/off nest, pup licking and grooming, nest repair, and non-pup directed behaviors designated “self-care” (self grooming, eating, and drinking). After the observational period, the dam and pups were carefully removed from the cage and nest size and quality were then measured using the 5-point Hess scale (Catanese et al., 2015). Following nest measures, pup retrieval assays were conducted. Briefly, pups were scattered in the cage at the end opposite from the location of nest. Dams were returned to the cage opposite the pups and observed for 10 minutes. First active touching of the pups and retrieval of each pup was recorded. Pup-initiated nursing, characterized by active pursuit of the dam by the pup in order to nurse, was also recorded.

Open Field Behavioral Assay

On pregnancy day 16 and LD 10 or 11, dams were tested using a standard open field apparatus 40 cm × 40 cm × 40 cm. Behaviors in the open field were scored by independent observers blind to treatment group. Measures included rearing against the walls, rearing in the center of the open field, freeze/stops, and grooming events.

Immunohistochemistry

On LD2 and LD21, brains were collected from dams and fixed in neutral buffered formalin (10%) (Fisher Scientific, Pittsburgh, PA) using methods optimized in our laboratory (Catanese et al., 2015; Catanese and Vandenberg, 2017a, b). Sections spanning the MPOA and VTA were identified using a mouse brain atlas (Franklin, 1997, 2012) and incubated overnight at 4°C with rabbit anti-ERα antibody directed against the C-terminus of the rat ERα (Anti-ERalpha C1355, Fisher Scientific) or a polyclonal antibody for tyrosine hydroxylase (Abcam, ab112). The sections were then incubated with biotin labeled secondary antibody (goat anti-rabbit Ab 64256, Abcam, Cambridge, MA) followed by streptavidin peroxidase complex (Ab64269, Abcam). Diaminobenzidine (ab64238, Abcam) was used to detect the reaction. Sections were mounted on slides, dehydrated and coverslipped. Brain sections were imaged and analyzed by an observer blind to treatment using neuroanatomical landmarks as described previously (Catanese et al., 2015; Catanese and Vandenberg, 2017a, b).

Statistical Analysis

Both behavioral and immunohistochemical analyses were conducted by experimenters blind to treatment groups. Data were analyzed using SPSS Version 22. For assessments of maternal behavior, continuous variable data were analyzed using 2-way ANOVA General Linear Model analyses with lactational day and treatment groups as independent variables, followed by Fisher’s posthoc tests. Open field data and cell counts from immunohistochemical assays were analyzed using 1-way ANOVA followed by Fisher’s posthoc tests with treatment group as the independent variable. Categorical data were analyzed using Chi Square. Data were considered statistically significant at p<0.05. Graphs illustrate mean ± standard error unless otherwise stated.

Results

Developmental EE2 exposures do not affect litter size, litter weight, or sex ratios in F2 litters

To determine whether developmental EE2 exposure alters reproductive performance, F1 females exposed to vehicle or EE2 during gestation and perinatal development were mated with untreated males. There were no significant differences in the number of pups born to females based on treatment group (Table 1). There were also no differences in the total litter weight or the average pup weight on PND1, prior to litter culling (Table 1). The percentage of male pups was decreased in the litters born to EE2-treated dams, although these differences were not statistically significant (Figure 1).

Table 1.

Effects of developmental EE2 treatment on F2 litter outcomes

	Control	0.01 μg EE2/kg/day	1 μg EE2/kg/day
Litter size (live pups on PND1 prior to culling)a	12.41 ± 0.61	12.77 ± 0.49	11.95 ± 0.77
Total litter weight (g) on PND1a	23.81 ± 1.13	23.73 ± 0.97	22.83 ± 1.39
Average pup weight (g) on PND1a	1.96 ± 0.050	1.87 ± 0.038	1.96 ± 0.063
Mortality rateb	3.5%	3.2%	4.8%

There were no effects of EE2 exposure on litter size, litter weight, or sex ratios in F2 litters

^aNo significant differences were noted using 1-way ANOVA.

^bNo significant differences were noted using Chi Square.

Figure 1. Sex ratios in F2-generation litters suggest possible effects of developmental EE2 exposure.

The percentage of male pups was decreased in the litters born to dams developmentally exposed to EE2, but these differences were not statistically significant.

Developmental EE2 exposure increases time spent on maternal self-care and time spent on nest building

To assess the effects of developmental EE2 exposure on maternal behavior, F1 females were observed undisturbed in their home cage. As expected, the time dams spent on the nest and the time spent nursing decreased across the postpartum period (e.g. as their pups became more mobile), but no effects of developmental EE2 exposure were observed (data not shown). On LD2, dams in the 1 μg EE2/kg/day exposure group spent significantly more time on self–care, measured by the time the dam spent eating, drinking, and self-grooming, when compared to dams exposed to 0.01 μg EE2/kg/day and controls (Figure 2A). Effects of EE2 were also observed for the percentage of time spent building the nest; on LD2, dams exposed to 1 μg EE2/kg/day spent more time nest building compared to controls and the 0.01 μg EE2/kg/day treatment group. On LD14, dams exposed to 0.01 μg EE2/kg/day spent more time nest building compared to both controls and dams exposed to 1 μg EE2/kg/day (Figure 2B). Across the lactational period, time spent grooming pups decreased, but there were no differences observed between treatment groups (Figure 2C).

Figure 2. Developmental EE2 exposure alters time spent on maternal self-care and time spent building the nest in the home cage.

A) On LD2, dams in the 1 μg EE2/kg/day group spent significantly more time on self–care (e.g. eating, drinking, and self-grooming) compared to controls. B) These same females spent more time nest building on LD2, whereas on LD14, dams exposed to the lower dose of EE2 spent more time nest building compared to controls. C) Developmental EE2 treatment did not alter time spent grooming pups at any postpartum timepoint evaluated. However these activities decreased over time. δ indicates significant effects of treatment compared to untreated controls of the same age, 2-way ANOVA with Fisher’s posthoc tests.

EE2 exposure did not alter nest size or quality in F1 females

Following the evaluation of behaviors in the home cage, we next asked whether developmental EE2 treatment would affect nest size or quality in F1 females. At all stages examined, we observed no effects of EE2 on internal nest volume, external nest volume, or nest quality (data not shown).

EE2 exposure alters time to touch pups in a retrieval assay

We next conducted a pup retrieval assay, which includes measures for the latency to first touch one or more pups, the latency to retrieve the first pup, and the latency to retrieve the entire litter. To control for differences in litter size, we limited our analyses only to litters with 9–10 pups. After scattering the pups at the end of the cage opposite the dam, dams exposed to 0.01 μg EE/kg/day showed significantly longer latency to touch the first pup on LD7 (Figure 3A). On LD2, the latency to retrieve the first pup was increased in dams exposed to 1 μg EE/kg/day, whereas the latency to retrieve the first pup was decreased in this group at LD7, although these differences were not statistically significant (Figure 3B). There were no significant differences in the time to retrieve the full litter based on treatment (data not shown). The number of dams that successfully retrieved at least one pup to the nest is reported in Table 2.

Figure 3. Developmental EE2 exposure alters latency to touch, retrieve pups.

A) On LD7, dams developmentally exposed to 0.01 μg EE/kg/day demonstrated significantly longer latency to touch the first pup. B) There were no treatment related effects on the time to retrieve the first pup at any timepoint across the postpartum period). * indicates a significant effect of age, 2-way ANOVA. δ indicates significant effects of treatment compared to untreated controls of the same age, 2-way ANOVA with Fisher’s posthoc tests.

Table 2.

The number of dams^a that successfully retrieved at least one pup to the nest.

	LD2	LD7	LD14
Control	14/16 (88%)	7/16 (44%)	1/15 (7%)
0.01 μg EE2/kg/day	9/16 (56%)	10/16 (67%)	0/16 (0%)
1 μg EE2/kg/day	9/9 (100%)	5/9 (56%)	0/9 (0%)

^aNote that only dams with 9–10 pups were evaluated for retrieval, so the overall sample size is lower than for other behavioral endpoints evaluated

EE2 exposure does not induce anxiety-like behaviors in the open field

We next assessed whether EE2 treatment induced anxiety-like behaviors in the open field assay. F1 females were evaluated in late pregnancy (pregnancy day 16) or mid-lactation (LD10 or LD11). At both time periods, there were no effects of treatment on number of center rears (a behavior associated with low anxiety) or grooming events (a behavior associated with heightened anxiety) (Figure 4A, 4B), suggesting that EE2 exposure does not alter anxiety-like behaviors.

Figure 4. Results from open field assays are not consistent with EE2-induced anxiety-like behaviors.

F1 females were evaluated late in pregnancy (pregnancy day 16) or mid-lactation (LD10 or LD11). A) EE2 treatment did not affect the number of center rears or B) the number of self-grooming events at either timepoint.

EE2 exposure attenuates pup induced nursing in mid-lactation

During the pup retrieval assay, pup induced nursing (PIN), in which the pup initiates nursing by pursuing the dam, rooting and latching onto the dam’s ventrum and actively nursing, was examined. PIN was not observed in any litters on LD2 (data not shown). On LD7, 31% of control litters displayed PIN behavior, whereas this behavior was seen less frequently (6–8%) in litters born to either EE2 dose (Figure 5A). On LD14, PIN behavior was observed at a high rate in controls (81% of litters) but in significantly fewer dams from both EE2 treatment groups (31–33%, Figure 5B).

Figure 5. Pup induced nursing is altered by developmental EE2 exposure.

A) On LD7, approximately one-third of control litters displayed PIN behavior, and this behavior was observed less frequently in litters born to either EE2 dose. B) On LD14, PIN behavior was observed in the vast majority of control dams but was less frequently observed in females from the two EE2 treatment groups. * p<0.01, Chi Square relative to controls.

ERα expression in the MPOA and dopaminergic cell number in the VTA are altered in EE2-exposed dams

To evaluate the effect of developmental EE2 exposure on maternally relevant brain regions, cells expressing ERα were quantified in the central MPOA of dams on LD2 and on LD21; two matched sections were selected, one from the rostral sub-region of the MPOA and another from the caudal sub-region of the MPOA (Figure 6A,B). On LD2, a significant reduction in ERα expression was observed in the rostral sub-region of MPOA in the 0.01 μg EE2/kg/day exposure group compared to controls (Figure 6C). Surprisingly, at LD21, evaluation of the same rostral sub-region of MPOA revealed a significant increase in ERα expression in the 0.01 μg EE2/kg/day group (Figure 6D).

Figure 6. ERα expression in the MPOA is affected by EE2 exposures in dams exposed during development.

Using ERα immunoreactivity to characterize expression in the central MPOA, we assessed one section from the rostral subregion of the MPOA and a second section from the caudal subregion of the MPOA (A,B). C) On LD2, there was a significant reduction in ERα expression in the lower EE2 group in the rostral subregion of MPOA, but not the caudal subregion. D) On LD21, there was a significant increase in ERα expression in the low EE2 treatment group in the rostral, but not the caudal, subregion of MPOA. * indicates significant difference from control, p<0.05, Fisher’s posthoc after significant 1-way ANOVA.

Because the VTA receives functional input from the MPOA and plays a role in maternal motivation (Numan, 2007; Numan and Stolzenberg, 2009), we next evaluated whether developmental EE2 exposure affects the number of dopaminergic cells in the VTA. We analyzed potential effects using antibodies against tyrosine hydroxylase, the rate limiting step in dopamine synthesis, traditionally used as a dopaminergic marker, in brains collected from dams on LD2 and LD21 (Figure 7A). On LD2, EE2-treated females in both groups had significantly fewer tyrosine hydroxylase positive cells in the VTA (Figure 7B). However, on LD21, there were no treatment related differences in tyrosine hydroxylase immunoreactivity in this brain region (Figure 7C).

Figure 7. Tyrosine hydroxylase immunoreactivity is altered in EE2-exposed females.

A) Tyrosine hydroxylase positive cells in the VTA were quantified on transverse sections in dams on LD2 and LD21. Shown here are representative photomicrographs from dams on LD2. B) On LD2, females exposed to both the lower and higher dose of EE2 had significantly fewer TH-ir neurons in the VTA. C) On LD21, there were no treatment related differences in the number of TH-ir cells in this region. * p<0.01, Fisher’s posthoc after significant 1-way ANOVA.

Discussion

Maternal behavior is a sensitive endpoint involving the interaction of physiological and environmental factors (Bale et al., 2010; Barrett and Fleming, 2011; Bridges, 2015) including estrogen signaling (Pfaff et al., 2011). Here, we investigated the effects of gestational and perinatal exposure to EE2 on maternal behavior and brain. We found that there were alterations in maternal behavior after developmental exposure to EE2 including time spent nest building, time on maternal self-care, and latency to touch the first pup. Increased time spent nest building and self-grooming may indicate alterations to maternal motivation, or repetitive behaviors (Albelda and Joel, 2012; Greene-Schloesser et al., 2011; Korff and Harvey, 2006) that may interfere with maternal behavior (Stern and Protomastro, 2000). Future studies are needed to assess if these observed behaviors are related to maternal motivation or OCD-like behavior using insistence on sameness modeling or a signal attenuation test (Crawley, 2007; Garner and Mason, 2002; Gross et al., 2012; Joel, 2006; Low, 2003). It is also possible that these behaviors may be related to the repetitive tail retrieving behavior we observed previously in F0 dams, the mothers of the current F1 generation, which were exposed to EE2 during pregnancy and lactation (Catanese and Vandenberg, 2017b).

In mice, the active display of maternal behavior declines across the postpartum period (Palanza et al., 2002a; Shoji and Kato, 2006). Our behavioral findings are consistent with these patterns (Figures 2, 3) and we also noted that the effects of EE2 observed in the current study were dependent on postpartum day and EE2 dose. For example, after developmental exposure to 1 μg EE2/kg/day, dams spent more time building the nest on LD2, whereas exposure to 0.01 μg EE2/kg/day led to increased time building the nest on LD14 (Figure 2). Another issue relevant to dose is the pattern in which effects were observed for the lower, but not the higher dose of EE2. Non-linear and non-monotonic dose responses are common for hormones and EDCs (Vandenberg et al., 2012); thus, effects observed at the lower EE2 dose may not manifest in dams exposed to the higher dose of EE2 (Vandenberg, 2014). Before we can conclude that EE2 induces non-monotonic effects on maternal behavior, additional dose groups should be evaluated, especially for those endpoints that displayed divergent effects at low and higher doses.

In prior rodent studies, gestational or neonatal exposures to EE2 were shown to induce alterations in reproductive tissues of males and females (Fisher et al., 1999; Naciff et al., 2002; Sawaki et al., 2003a, b). A multi-generational reproductive toxicity assay conducted by the US National Toxicology Program showed effects of EE2 on the timing of puberty in females, weights of male reproductive organs, incidence of mammary ductal hyperplasias in males, testicular sperm count, and disturbances to the estrus cycle (National Toxicology Program, 2010). Future studies are needed to determine whether the low doses we have evaluated in this study also affect these other relevant endpoints in CD-1 mice. Importantly, in contrast to several other studies examining the effects of EE2 (Arabo et al., 2005; Dugard et al., 2001; Howdeshell et al., 2008; Ryan et al., 2010), we examined low doses of EE2 which did not induce reproductive toxicity in F1 offspring at birth (Catanese and Vandenberg, 2017b). Although some adverse outcomes were observed in F1 dams/F2 litters (Table 1), these did not appear to be related to EE2 exposure.

Only a small number of studies have investigated the effects of exposure to EE2 on the brain and behavior of rodents. In one study, prenatal and early postnatal exposure to EE2 masculinized behaviors that are typically sexually dimorphic such as short-term spatial-memory and anxiety-like behaviors in females tested in adulthood (Ryan and Vandenbergh, 2006). In another study, pubertal rodents treated orally with EE2 showed alterations in the number of ERα-positive neurons in two regions of the brain, the MPOA and the ventromedial nucleus (VMN); ERα expression was increased in males on postnatal day (PND) 37 in the VMN and increased in females in the MPOA on PND 90 (Ceccarelli et al., 2007), providing evidence that ERα expression is sensitive to EE2 exposure. A few studies have examined the effects of EE2 and other EDCs on maternal behaviors in F1 females exposed during gestation (or gestation and the perinatal period) (Derouiche et al., 2015; Jefferson et al., 2007; Johnson et al., 2015; Palanza et al., 2002a; Venerosi et al., 2008). In one study in California mice, F1 dams exposed to EE2 during early development spent less time nursing, less time grooming pups and less time on the nest (Johnson et al., 2015). In contrast to these findings, we did not observe changes in time spent grooming pups or time on the nest; however, we did observe increased maternal self-care in the 1 μg EE2/kg/day group on LD2 and increased time spent building the nest in both groups on different days during the lactational periods (Figure 2). The outcomes in these studies may differ due to variations in the methods including timing of treatment; in the study in California mice, the parental generation was fed a diet supplemented with 0.1 part per billion EE2, which began two weeks prior to breeding and lasted through pregnancy and lactation, which is longer than in the current study. Additionally, the California mouse is a bi-parental mouse, thus the paternal contribution to pup care may have influenced the maternal results in that study.

Despite relatively modest differences in specific behavioral outcomes, the results of this study support other behavioral investigations which suggest that the gestational period is sensitive to disruption by xenoestrogens (Ball et al., 2010; Della Seta et al., 2005; Palanza et al., 2002a; Simmons et al., 2005). In fact, there is significant interest in identifying the most vulnerable periods for EDC exposures (Grandjean et al., 2015). In a study of CD-1 mice, exposures to BPA during either pregnancy (in F0 females) or gestation (in F1 females) altered maternal behaviors, suggesting that both of these periods are sensitive to xenoestrogens (Palanza et al., 2002a). However, females exposed during both periods did not display significant disruptions in measures of maternal behavior. The authors postulated that there may have been permanent changes to homeostatic mechanisms due to the early gestational exposure that may have altered the female’s response to adult exposure (Palanza et al., 2002a). Interestingly, the authors also hypothesized that decreases in maternal behavior that were found after in utero exposure only may have been due to organizational effects on the developing neuroendocrine systems important for later maternal behavior. Thus, it is plausible that in dams treated both in utero and in adulthood, an organizational change that may have occurred developmentally could have altered the activational status of the neuroendocrine underpinnings governing maternal behavior later in adulthood.

In our previous study of maternal behavior in the F0 generation, we found no changes in traditional measures of maternal behavior induced by EE2 (Catanese and Vandenberg, 2017b). However, we did observe an increase in an infrequently quantified stereotypy or OCD-like behavior, tail retrieval, in the parental dams exposed to EE2. It is important to note that these EE2-induced effects on the F0 dams may have affected the behavior of the offspring; the quality of maternal behavior in the parental generation can influence the quality of maternal behavior in offspring (Champagne and Curley, 2008; Francis et al., 1999a; Meaney, 2001). While there were no significant changes in traditional measures of maternal behavior in the F0 dams, it is possible that there were behavioral impacts of the stereotypy we observed in the F0 generation on offspring. It is also plausible that there were changes to maternal behavior that were not captured due to our study design; future studies should include more frequent observations each day, including during the dark cycle as well as across the post-partum period. Further, many factors including EDC exposures can potentially alter offspring in some manner, including changes in behavior or sensory systems, which could in turn alter the dam’s responsiveness to her pups; for example, the decreased initiation of PIN behavior in EE2-treated F2 generation pups could be an effect of developmental exposure on the pups themselves, an effect on the F1 mothers, or an interaction between both factors. Cross fostering studies would help to address these issues, although such studies face limitations because fostering of pups can also modify maternal behaviors (Francis et al., 1999a; Maccari et al., 1995).

Our F0 dams (the mothers of the females evaluated in this study) were shipped during early pregnancy, i.e., pregnancy days 3–6. Although shipping occurred prior to implantation and the dams were allowed to acclimate before treatment commenced, we cannot rule out potential effects of shipping stress on dams or offspring during early stages of embryonic development. Importantly, the mothers of the females evaluated in this study displayed maternal behaviors that were indistinguishable from that of controls (Catanese and Vandenberg, 2017b), making it unlikely that severe stress reactions occurred.

To our knowledge, this study is the first to examine the effects of EE2 on both maternal behavior and the maternal brain in the F1 generation. We found that ERα expression in the MPOA was significantly decreased on LD2 and significantly increased on LD21 in dams developmentally exposed to 0.01 μg EE2/kg/day (Figure 6). The sub-region of the MPOA in which ERα was quantified has been shown to be dedicated to pup retrieval (Tsuneoka et al., 2013). Although quantitative deficits in pup retrieval were not observed at LD2, we did find that on LD7, the dams in the same group demonstrated a longer latency to make initial contact with pups (Figure 3); additional experiments are needed to determine whether these behaviors are mediated via ERα. A number of studies have demonstrated the importance of the MPOA in the onset of maternal behavior and as a site of hormone response to pup stimuli (Bridges et al., 1990; Fahrbach and Pfaff, 1986; Giordano et al., 1990; Giordano et al., 1989; Moltz et al., 1970; Numan, 2006; Numan et al., 1977; Rosenblatt et al., 1988; Rosenblatt et al., 1994). Studies have also demonstrated that the MPOA is involved in motivational aspects of maternal behavior using approaches such as operant responding and modified conditioned place preference paradigms (Lee et al., 2000; Numan, 2007; Numan and Stolzenberg, 2009; Olazabal et al., 2013; Pereira and Morrell, 2011). Developmental exposures to xenoestrogens can affect ERα expression in the MPOA (Ceccarelli et al., 2007) and our results suggest that developmental alterations to ERα expression can manifest during the lactational period. It is possible that the low level of EE2 exposure experienced during development altered the response of the MPOA to estrogen-mediated expression of its receptor early in the postpartum period (Borras et al., 1994). Further, the alterations observed during early postpartum may have induced a compensatory increase in receptor expression later in postpartum, e.g. at weaning.

The reduction in the expression of tyrosine hydroxylase positive cells in the VTA as a result of developmental exposure to both doses of EE2 on LD2, but not LD21 (Figure 7), suggest that the effects of developmental exposure to EE2 may be specific to the early postpartum period for this endpoint. It is not possible to determine whether the effects found in the VTA are relevant to the modest alterations in maternal behavior that we observed. Stereotypies arise after loss of dopaminergic neurons in the VTA in Borna disease (Solbrig et al., 1995). However, further investigation is necessary to test whether conditional loss of dopaminergic cells in the VTA might also induce repetitive or OCD-like behaviors. Further, while dopaminergic neurons of the VTA play a role in reward and motivation, we did not conduct tests of maternal motivation, thus it will be important for future studies to use a bar press test to examine pup reinforcement (Hauser and Gandelman, 1985; Lee et al., 2000) or signal attenuation (Joel, 2006). Increased repetitive behavior during the postpartum period may be relevant to humans, as post-partum OCD has been found to develop in both men and women during early stages of parenthood (Abramowitz et al., 2003; Fairbrother and Abramowitz, 2007; Leckman et al., 1999; Maina et al., 1999). Further, approximately 11% of women manifest symptoms of OCD at two weeks postpartum, and as many as half of these women’s symptoms can persist for up to 6 months (Miller et al., 2015).

We have now evaluated many of the same endpoints, including maternal behaviors and evaluations of regions in the postpartum brain relevant to maternal care, in two generations of mice (exposed adult females and their developmentally exposed daughters) exposed to two xenoestrogens: EE2 (Catanese and Vandenberg, 2017b) and bisphenol S (BPS) (Catanese and Vandenberg, 2017a; LaPlante et al., 2017). These studies allow us to compare two potentially vulnerable periods and two compounds that bind to ER but may act distinctly. In fact, our studies indicate different effects as a result of exposure to these chemicals, as well as different effects when comparing the F0 and F1 generations (Figure 8). For example, reductions in tyrosine hydroxylase-immunoreactive cells in the VTA were observed in both the F0 and F1 females after EE2 exposure, yet this outcome was not observed in either the F0 or F1 dams exposed to BPS. It will be important to investigate regions of the brain involved in maternal care not investigated here, such as the prefrontal cortex, nucleus accumbens and amygdala, which may have been impacted by direct or indirect effects due to in utero exposure to EE2. Additoinally, it is possible that the differences we observed in EE2 and BPS treated females were due to the different doses selected for these two chemicals. Questions of dose may be resolved by a establishing a more complete dose response relationship for behavioral and neural endpoints for both compounds. It is likely however, that these two compounds act in distinct ways. For example, BPA has been shown to have different effects on the ER when compared to E2 (Gould et al., 1998), and does not completely reproduce the effects of EE2 on neural endpoints (Ceccarelli et al., 2007). Thus, it is important to consider that BPS, and by extension other estrogenic environmental chemicals, may act via the ER in a manner that is distinct from EE2.

Figure 8. Summary of effects of two xenoestrogens on two generations of mothers.

We have now evaluated the same endpoints from females exposed to EE2 or BPS either during pregnancy and lactation (the F0 generation) or during gestation and perinatal development (the F1 generation). Significant effects (p<0.05) are indicated by bold arrows; the direction of the arrows indicates whether the treatment increased or decreased the endpoint relative to untreated controls. Non-bold arrows indicate statistical trends. Gray blocks represent endpoints that have not yet been evaluated. Data from F0 dams treated with EE2 were published in (Catanese and Vandenberg, 2017b). Data from F0 and F1 dams treated with BPS were published in (Catanese and Vandenberg, 2017a) and (LaPlante et al., 2017).

The mechanisms underlying EDC effects on hormone receptors are complex, as are the actions of estrogen on its receptors (Heldring et al., 2007; Katzenellenbogen et al., 2000; Nilsson et al., 2001; Welshons et al., 2006; Zoeller et al., 2012) One plausible mechanism by which BPS may act differently compared to EE2 is the possibility that BPS can bind the receptor and induce atypical conformational changes, which further affect downstream events including gene expression. Although ER can adapt to the molecular structure of diverse ligands (Anstead et al., 1997; Heldring et al., 2007), conformational changes associated with divergent ligands can alter co-activator or co-repressor recruitment to the ligand receptor complex (Heldring et al., 2007; Katzenellenbogen et al., 2000; Klinge et al., 2004; Nilsson et al., 2001). Thus, differences in co-regulators or in the subsequent transcriptional activation after binding to the estrogen response element may be differentially affected by BPS and EE2 (Heldring et al., 2007; Klinge et al., 2004).

Given the differences in ERα expression in the MPOA after exposures to EE2 and BPS, additional studies are needed to further investigate whether the effects of BPS are due to action at ERα. While evidence demonstrates the critical importance for ERα in the MPOA for the display of maternal behavior in the mouse (Ogawa et al., 1998; Ribeiro et al., 2012), ERβ is also expressed in the MPOA and there is evidence for co-expression of these receptors in a number of neurons (Shughrue et al., 1998). Ultimately, our studies lead to further questions, and demonstrate that ER expression in the MPOA is dynamic and the regulation of ER appears to be plastic; receptor expression appears to be selective and sensitive to environmental as well as developmental impacts. Importantly, EE2 is often used as an estrogenic control in EDC studies and the differences in response to exposures between EE2 and BPS uncovered in our work indicate that estrogenic EDCs may disrupt maternal behaviors through divergent mechanisms compared to EE2. Thus, EE2 may not provide mechanistic relevance for use as an estrogenic positive control in EDC studies.

Maternal behavior is multifactorial, encompassing numerous behavioral components influenced by emotional, psychological, physiological, neuroendocrine and social factors (Numan, 2003; Pereira and Ferreira, 2015). Here we demonstrate alterations to components of maternal behavior and effects on the maternal brain in CD-1 mice after developmental exposure to low dose EE2. We have additionally demonstrated effects of EE2 on dams exposed during pregnancy (Catanese and Vandenberg, 2017b) and BPS exposures in the F0 and F1 generation (Catanese and Vandenberg, 2017a; LaPlante et al., 2017) (Figure 8). While a number of additional studies demonstrate the profound effects of maternal exposures on the health of offspring, our work and that of others provides strong evidence that there are notable effects of treatment on the maternal brain and behavior of the mother, which can in turn, impact offspring (Cummings et al., 2010; Cummings et al., 2005; Tomihara et al., 2015). From an evolutionary perspective, maternal behavior is often defined in light of its importance for the survival of offspring, however, the importance of maternal care in humans stems from its contribution to the intellectual, physical, emotional and psychological development of children. Therefore, uncovering mechanisms that might influence proper maternal care has broad social and public health implications.

Highlights.

• Developmental EE2 exposure increases time spent on maternal self-care

• Pups born to EE2-treated females were less likely to initiate nursing on LD14

• Exposure to EE2 altered ERα expression in the MPOA both early and late in lactation

• Developmental EE2 exposure decreased dopaminergic cells in the VTA on LD2

• Effects of EDCs on maternal behavior depend on chemical and period of exposure