Bisphenol S (BPS) Alters Maternal Behavior and Brain in Mice Exposed During Pregnancy/Lactation and Their Daughters

Abstract

Estrogenic endocrine disrupting chemicals have been shown to disrupt maternal behavior in rodents. We investigated the effects of an emerging xenoestrogen, bisphenol S (BPS), on maternal behavior and brain in CD-1 mice exposed during pregnancy and lactation (F0 generation) and in female offspring exposed during gestation and perinatal development (F1 generation). We observed different effects in F0 and F1 dams for a number of components of maternal behavior, including time on the nest, time spent on nest building, latency to retrieve pups, and latency to retrieve the entire litter. We also characterized expression of estrogen receptor α in the medial preoptic area (MPOA) and quantified tyrosine hydroxylase immunoreactive cells in the ventral tegmental area, 2 brain regions critical for maternal care. BPS-treated females in the F0 generation had a statistically significant increase in estrogen receptor α expression in the caudal subregion of the central MPOA in a dose-dependent manner. In contrast, there were no statistically significant effects of BPS on the MPOA in F1 dams or the ventral tegmental area in either generation. This work demonstrates that BPS affects maternal behavior and brain with outcomes depending on generation, dose, and postpartum period. Many studies examining effects of endocrine disrupting chemicals view the mother as a means by which offspring can be exposed during critical periods of development. Here, we demonstrate that pregnancy and lactation are vulnerable periods for the mother. We also show that developmental BPS exposure alters maternal behavior later in adulthood. Both findings have potential public health implications.

In the rat, estrogens play an important role in the regulation of maternal behaviors via estrogen receptor α (ERα) signaling in the medial preoptic area (MPOA) of the forebrain (1–6). The MPOA is critical for the onset of maternal behavior, and lesions to this area abolish maternal care (2, 7–9). In the mouse, the onset of maternal behavior is not strictly dependent on the high levels of estrogen produced during late pregnancy; nulliparous females demonstrate spontaneous maternal behavior (10–14), as do ovariectomized and aromatase knockout mice (14, 15). However, there is evidence indicating a role for estrogen (14–16) and demonstrating the importance of ERα in the display of mouse maternal behavior (17, 18).

Endocrine disrupting chemicals (EDCs) with ER agonist activities such as the xenoestrogen bisphenol A (BPA) (19–22) can interfere with a broad range of physiological processes, including neural development and reproduction (23). BPA induces deleterious physiological effects in a number of tissues (19, 24, 25), including structural changes in brain regions that are associated with changes in behavior (26). A small number of studies suggest that BPA can alter maternal behaviors in rodents (27–32). In 1 study, female mice exposed to 10 µg BPA/kg/d during pregnancy days 14 to 18 spent less time nursing and less time on the nest compared with unexposed females (28). Interestingly, the same effects were observed in the F1 offspring (exposed on gestational days 14 to 18). Yet, when females exposed prenatally were exposed again during pregnancy, no alterations in maternal behaviors were observed (28). This evidence suggests that xenoestrogen exposures can alter maternal behavior and that the timing of exposure might be critical; pregnancy as well as the developmental period may be sensitive to endocrine disruption. To our knowledge, studies examining the effects of BPA and other EDCs on maternal behavior have not yet investigated the potential underlying neural mechanisms.

Due to public health concerns with BPA, alternative chemicals have been developed for use in consumer products. One replacement, bisphenol S (BPS), is currently used in baby bottles, thermal receipts, consumer paper products, and personal care products and has been detected in foodstuffs and canned foods (33–36). Biomonitoring studies reveal that human exposures to BPS are likely to be low but widespread (37) and have increased over the last decade (38). Although BPS has not been examined extensively, several studies indicate that this compound displays estrogenic properties in both genomic and membrane-associated estrogen signaling similar to BPA (39–41). To date, studies of BPS in mammals are limited, and there are very few studies investigating the effects of exposure on behavior or brain (42, 43). Furthermore, to our knowledge, there are no studies examining whether this compound disrupts maternal behavior or the maternal brain.

We postulated that low-dose exposures to BPS would affect maternal behavior and related neural structures in both dams (the F0 generation) and their offspring (F1 females) in adulthood. Given that one mechanism of endocrine disruption is altered hormone receptor expression, we examined the effects of BPS on ERα expression in the MPOA. In addition, we investigated tyrosine hydroxylase immunoreactivity, a marker for dopaminergic cells, in the ventral tegmental area (VTA), a brain region receiving functional input from the MPOA. The MPOA output to the VTA is also important for maternal motivation (44, 45). Dopaminergic neural circuits in the MPOA are involved in the regulation of the onset and maintenance of maternal behavior in rats; Numan and Stolzenberg (44) proposed that estradiol and dopamine may act through the same signaling cascades during the onset of maternal behavior. Further evidence indicates that estrogen-sensitive oxytocin neurons project from the MPOA to the VTA (46–48). Thus, we hypothesized that both the MPOA and VTA would be sensitive to BPS. EDCs have been shown to cause long-term effects, including some following developmental exposures that may not manifest until adulthood and others that may only be apparent under specific conditions (i.e., after parturition). Here, we investigated potential effects of BPS exposures on maternal behavior and in neural regions important for maternal behavior. We tested the hypothesis that exposures to BPS affect the F0 generation exposed during pregnancy and lactation and the F1 generation exposed inutero and during the perinatal period. Our results suggest that BPS induces different effects on the F0 and F1 females.

Method

Animals

Timed pregnant female CD-1 mice (Charles River Laboratories, Raleigh, NC) were individually housed (until parturition) in polysulfone cages with food (ProLab IsoDiet, Brentwood, MO), which has been reported to have minimal estrogenic activity (49)] and tap water (in glass bottles) provided ad libitum. The animals were maintained in temperature- and light-controlled (12 hours light, 12 hours 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.

Beginning on pregnancy day 8, pregnant females were weighed daily. On day 8, dams were randomly allocated to treatment groups using statistical software that allowed normal distribution in each treatment group based on body weight. From pregnancy day 9 to lactational day 20, dams ingested a small wafer (Nabisco, East Hanover, NJ) treated with BPS or vehicle alone (70% ethanol). All ethanol was dried to completion prior to feeding (50). Wafers were dosed with solutions designed to deliver 2 or 200 µg BPS/kg/d (Santa Cruz Biotechnology, Santa Cruz, CA, catalog no. sc-238983, 99% purity) (n = 15 to 17 for each dose). BPS dosage was adjusted for body weight daily. The 200-µg/kg/d dose was selected because it appeared to alter some aspects of maternal behavior in a pilot study (data not shown). This dose is higher than suspected human exposure levels (37); thus, we also selected a lower dose of 2 µg/kg/d, which approaches human exposures. Both doses of BPS are far below the toxicological no observed adverse effect level of 10 mg/kg/d based on parental toxicity in a guideline developmental toxicity assay (51), and thus both doses would be characterized as “low” (52). During wafer administration, dams were briefly transferred to clean cages. During the lactational period, care was taken to avoid potential effects of separation (53). In addition, pups were not handled during this period to avoid potential effects of handling.

Dams delivered naturally [birth designated lactational day (LD) 0]. Litters were culled to 10 pups on LD1; male and female pups were kept in all litters (54). A small number of dams had fewer than 10 pups, and thus no pups were culled in these litters. Litters were weaned on LD21 and separated by sex. Two F1 female offspring from each F0 dam were raised to adulthood, mated with unexposed CD-1 males (Charles River Laboratories), and tested for maternal behavior using the same assays at the same time points across lactation as used to test their mothers. A schematic illustrating the exposure paradigm and timing of behavioral evaluations is shown in Figure 1.

Figure 1.

Schematic illustrating the experimental design. Red bar indicates the period of exposure (pregnancy day 9 through lactational day 21). The F0 dams were directly exposed; F1 pups were exposed via the mother (placental transfer in utero and lactational transfer during the perinatal period). Blue arrows indicate the timing of open-field behavioral assessments. Green arrows indicate the timing of maternal behavior assays.

Maternal behavior assays

For both the F0 and F1 generations, maternal behavior was assessed on LD2, 7, and 14. First, dams were observed without intervention for a period of 90 minutes at the beginning of the light phase. Observations were recorded every 3 minutes for the following measures: dam position on/off nest, self-care (self-grooming, eating, drinking), sleeping/resting, nest repair, and pup licking/grooming. At the end of the observational period, the dam and pups were gently removed from the cage. The dimensions of the nest were measured (55). Two independent observers scored the nest quality using a modified 5-point Hess scale (56). All litters were evaluated for these measures, regardless of size.

Finally, dams were assessed for pup retrieval. Following nest measurement, the pups were scattered in the cage at the end opposite the nest, and the dams were returned to the nest area. The latency to first touch 1 or more pups and the latency to retrieve each pup to the nest were recorded for a period of 10 minutes. To account for differences in litter size, only those litters with 9 to 10 pups were analyzed in pup retrieval assays. Furthermore, because a percentage of dams did not retrieve pups, regardless of treatment, some evaluations were conducted only in “active” retrieving dams (17).

Open-field behavioral assay

Dams were tested on pregnancy day 16 and LD10/11 using an open-field apparatus 40 cm × 40 cm × 40 cm (42). Measures were scored by 2 independent observers blind to treatment group, including rearing against the walls, rearing (without contact with walls), freeze/stops, and grooming events.

Immunohistochemistry

Brains were collected from F0 dams on LD21 and from F1 dams on LD2 or LD21 and fixed in neutral buffered formalin (10%) (Fisher Scientific, Pittsburgh, PA). Brains were sectioned in the transverse plane at 40 μm, and the MPOA and VTA were identified using a mouse brain atlas (57, 58). Free-floating sections were processed for immunoreactivity to ERα or tyrosine hydroxylase (TH, the rate-limiting enzyme in catecholamine synthesis, used as a marker for dopaminergic cells). Antigen retrieval was performed using 0.01 citric acid buffer (pH 6.0), followed by quenching of endogenous peroxidases using 3% hydrogen peroxide in methanol. Sections were washed, blocked with normal goat serum in 1.5% milk, and incubated overnight at 4°C with rabbit anti-ERα antibody directed against the C-terminus of the rat ERα (1:20,000 anti-ERα C1355; Millipore, Temecula, CA; catalog number 06-935, RRID AB_ 310305, lots 2488982, 2517829, and 2697437) or a polyclonal antibody for TH (1:2000; Abcam, Cambridge, MA; ab112, RRID AB_297840, lots GR-193639-4 and GR-166961-3). Sections were then washed and incubated with biotin-labeled secondary antibody (goat anti–rabbit, Ab64256; Abcam) followed by streptavidin peroxidase complex (Ab64269; Abcam). Colorimetric detection using diaminobenzidine (DAB) chromogen and substrate (ab64238; Abcam) was followed by a wash in tap water. Sections were stored in phosphate-buffered saline with 0.1% Tween-20 until mounted on slides, dehydrated, and cover-slipped. Details about the antibodies are included in Table 1.

Table 1.

Required Table of Information About Antibodies

Peptide/Protein Target	Antigen Sequence (if Known)	Name of Antibody	Manufacturer, Catalog No., or Name of Source	Species Raised in Monoclonal or Polyclonal	Dilution Used	Research Resource Identifier
ERα		Anti-ERα (C1355)	Millipore, 06-935	Rabbit; polyclonal	1:20,000	AB_310305
TH		Antityrosine hydroxylase	Abcam, ab112	Rabbit; polyclonal	1:2000	AB_297840
Secondary		Biotinylated goat anti–rabbit immunoglobulin	Abcam, ab64256	Goat; polyclonal	Ready to use (5 μg/mL)

One image per section was collected using a Zeiss AxioImager dissection microscope (×120 magnification for MPOA; ×100 magnification for VTA) and Zeiss high-resolution color camera (Carl Zeiss Microscopy, Thornwood, NY). ImageJ software (National Institutes of Health, Bethesda, MD) was used to convert the image from red, green, and blue color to 8 bit, subtract background, and automatically threshold. Cells expressing ERα in the MPOA and TH in the VTA were counted on anatomically matched sections. Owing to limitations in the quantification of DAB coloration, we did not measure staining intensity (59, 60). For feasibility reasons, ERα expression was quantified in 2 MPOA sections per animal: 1 from the rostral central MPOA (cMPOA), ∼0.14 mm from the bregma, and 1 from the caudal cMPOA, ∼0.02 mm from the bregma, by an observer blind to treatment. The cMPOA, which was selected because it is essential for pup retrieval in the mouse, was identified using neuroanatomical landmarks as described in Tsuneoka et al. (61). The VTA was identified using neuroanatomical landmarks (57, 58) and TH immunoreactivity as demonstrated in Yamaguchi et al. (62). TH immunoreactivity was assessed in one section located ∼2.92 from the bregma. For the MPOA, ERα labeling in the bed nucleus of the stria terminalis and the ventromedial nucleus of the hypothalamus were used as positive controls. This antibody has been validated in brain tissue [see Kelly et al. (63)]. For the VTA, TH labeling in the arcuate nucleus was used as a positive control. The specificity of the TH antibody is provided by the vendor using Western blot analysis. For both the MPOA and the VTA, sections without primary antibodies were used as negative controls.

Statistical analysis

Behavioral and immunohistochemical analyses were conducted by observers blind to treatment groups. Data were analyzed using SPSS version 22 (SPSS, Inc., Chicago, IL). For assessments of maternal behavior, continuous variable data were analyzed using 2-way analysis of variance general linear model analyses with lactational day and treatment as independent variables, followed by Bonferroni post hoc tests. Open-field data were analyzed using 1-way analysis of variance with treatment as the independent variable. Categorical data were analyzed using χ². Data were considered statistically significant at P < 0.05. Graphs illustrate means ± SEs unless otherwise stated. Graphs indicate post hoc statistical analyses only when treatment was statistically significant in the 2-way analysis of variance. Sample sizes for individual end points are provided in Table 2.

Table 2.

Sample Sizes Used for Evaluation of Different End Points

Characteristic	Control, n	2 μg BPS/kg/d, n	200 μg BPS/kg/d, n
F0 maternal behavior	15	17	15
F0 MPOA evaluation (ERα expression)	10	13	11
F0 VTA evaluation (TH expression)	10	13	10
F1 maternal behavior	14	15	15
F1 MPOA evaluation on LD2 (ERα expression)	10	15	15
F1 VTA evaluation on LD2 (TH expression)	10	15	15
F1 MPOA evaluation on LD21 (ERα expression)	14	15	15
F1 VTA evaluation on LD21 (TH expression)	13	14	14

Results

Infanticide and maternal neglect were observed in F1 but not F0 dams exposed to BPS

BPS treatment did not affect litter size, litter weight on LD1, average pup weight on LD1, or the sex ratio (% male pups) of litters born to either the F0 or the F1 generation (Table 3). No examples of infanticide or severe neglect were observed in the F0 dams for any treatment. However, moderate neglect, including poor cleaning of the pups and pups retaining their umbilical cords, was observed in 18% of F0 dams exposed to 2 μg BPS/kg/d (χ², P = 0.08 compared with controls).

Table 3.

Litter Outcomes From F0 and F1 Dams (in F1 and F2 Litters)

Dam	Litter Size, Mean ± SE	Litter Weight on LD1, Mean ± SE, g	Average Pup Weight, Mean ± SE, g	% Males in Litter, Mean ± SE	% Litters With Infanticide or Severe Neglect	% Litters With Moderate Neglect
F0
Control	9.9 ± 1.04	19.5 ± 1.77	2.04 ± 0.070	50.2 ± 5.0	0	0
2 μg BPS/kg/d	10.5 ± 0.85	19.7 ± 1.42	1.94 ± 0.056	46.0 ± 3.4	0	18a
200 μg BPS/kg/d	10.1 ± 0.97	20.0 ± 1.53	2.06 ± 0.073	46.9 ± 4.4	0	7
F1
Control	11.1 ± 0.74	21.7 ± 1.29	2.01 ± 0.058	51.6 ± 2.7	0	4
2 μg BPS/kg/d	12.2 ± 0.52	23.1 ± 1.12	1.94 ± 0.048	50.2 ± 2.4	13a	9
200 μg BPS/kg/d	11.2 ± 0.61	22.0 ± 1.04	2.05 ± 0.064	48.3 ± 2.4	0	17a

^a

P < 0.1, χ² comparison with controls in the same generation.

In the F1 generation, infanticide and severe neglect (requiring euthanasia of some pups or the entire litter) were observed in 13% of females exposed to 2 μg BPS/kg/d (Table 3). More moderate examples of poor care were also observed in the F1 generation, including pups that were stuck together, improperly cleaned, or severely bruised on LD1 (Table 3). Collectively, these results suggest that BPS may induce alterations to instrumental maternal care in females exposed during early development.

BPS exposure alters time spent on the nest and nest building in the home cage

To assess the effects of BPS exposure on maternal behavior, dams were observed in their home cage without any experimental intervention. In typical control dams, time spent on the nest decreases over the postpartum period as the pups move about the cage independently [Fig. 2(A)]. F0 dams exposed to 200 µg BPS/kg/d spent significantly more time on the nest on LD14, suggesting a failure to extinguish nesting behavior in these dams [Fig. 2(A)]. No significant differences between treatment groups were observed in F0 females at any other time evaluated. Strikingly, in the F1 generation, dams exposed to either 2 or 200 µg BPS/kg/d during development spent significantly less time on the nest on both LD2 and LD7 compared with controls [Fig. 2(A)]. BPS did not alter the time dams spent grooming pups in either the F0 or F1 generation [Fig. 2(B)]. The time that dams spent building and repairing the nest was also evaluated. In F0 dams, BPS treatment did not affect time spent nest building. However, in the F1 generation, dams that were developmentally exposed to 200 µg BPS/kg/d spent more time nest building on LD14 compared with controls [Fig. 2(C)]. Collectively, these results suggest that dams exposed to BPS as adults experience modest changes to traditional measures of maternal behavior (e.g., failure to extinguish high levels of care in the later postpartum period), whereas females exposed to BPS during development experience more detrimental changes to maternal behavior (e.g., failure to stay on the nest during early and midpostpartum periods).

Figure 2.

BPS exposure alters time spent on nest and time spent nest building in the home cage. (A) On LD14, F0 dams exposed to 200 µg BPS/kg/d spent significantly more of the observed time on the nest. F1 dams developmentally exposed to both 2 and 200 µg BPS/kg/d spent significantly less of the observed time on the nest on both LD2 and LD7 compared with controls. (B) Adult or developmental BPS exposure did not alter observed time spent grooming pups at any point in the postpartum period, but these activities decreased over time. (C) There were no significant changes in observed time spent nest building in F0 dams, whereas F1 dams exposed to 200 µg BPS/kg/d spent significantly more observed time nest building on LD14. Red graphs indicate F0 dams and blue graphs indicate F1 dams. *P < 0.05, Bonferroni post hoc after significant 2-way analysis of variance for treatment.

BPS affects nest size in the F0 but not F1 generation

The internal volume of the nest was calculated using the average internal diameter of the nest, measured using the internal walls constructed from a single cotton nestlet, and the average nest depth. In the F0 generation, dams treated with 2 µg BPS/kg/d were found to have smaller nests on LD2 compared with controls. There were no significant differences in nest size between treatments in F1 females at any point across the lactational period examined (Fig. 3). Nest quality was not affected in either the F0 or the F1 dams (data not shown).

Figure 3.

BPS exposure induces minor changes in nest size in the F0 generation at LD2. On LD2, F0 dams exposed to 2 µg BPS/kg/d built significantly smaller nests compared with controls. F0 dams exposed to 200 µg BPS/kg/d also built smaller nests, but this was not significant. There were no treatment-related effects on nest size in F1 dams at any time point across the postpartum period. *P < 0.05, Bonferroni post hoc after significant 2-way analysis of variance for treatment.

BPS treatment alters pup retrieval

Pup retrieval assays were conducted on LD2, LD7, and LD14 and include measures for the latency to first touch 1 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 included only those litters with 9 to 10 pups in our analysis. F0 dams exposed to 200 µg BPS/kg/d showed significantly longer latency to touch the first pup on LD14. In contrast, there were no significant differences in time to first touch pups between treatment groups for F1 females at any time point [Fig. 4(A)]. When we evaluated retrieval times for all dams, including those that did not retrieve 1 or more pups (retrieval times of 600 seconds were assigned for these females), there were no significant differences based on treatment of the time to retrieve the first pup [Fig. 4(B)] or the time to retrieve the entire litter [Fig. 4(C)] in either generation.

Figure 4.

BPS alters pup retrieval parameters in F0 dams. (A) On LD14, F0 dams exposed to 200 µg BPS/kg/d displayed a significantly longer latency to touch the first pup. There were no treatment-related changes to this parameter in F1 dams. (B) Time to retrieve the first pup and (C) time to retrieve the entire litter were not affected by BPS treatment in either F0 or F1 dams. Dams that did not retrieve were assigned a retrieval time of 600 seconds, the full length of the retrieval assay. Red graphs indicate F0 dams and blue graphs indicate F1 dams. *P < 0.05, Bonferroni post hoc after significant 2-way analysis of variance for treatment.

Because some females failed to retrieve any pups to the nest and others failed to complete retrieval of the entire litter [Supplemental Fig. 1(A)], we next evaluated only those dams that retrieved at least 1 pup. Effects of BPS were observed for the time to retrieve the first pup in dams from both the F0 and F1 generations. On LD2, F0 dams exposed to 2 µg BPS/kg/d displayed a significantly longer latency to retrieve the first pup [Supplemental Fig. 1(B)]. In contrast, F1 dams developmentally exposed to 2 µg BPS/kg/d demonstrated significantly shorter latency to retrieve their first pup on LD7 [Supplemental Fig. 1(B)]. Retrieval of the entire litter was common during early parturition in females that retrieved at least 1 pup but was never observed in late parturition [Supplemental Fig. 1(C)]. When we evaluated the time to retrieve the entire litter in females that successfully retrieved at least 1 pup, there were no effects of BPS in the F0 generation. However, F1 females developmentally exposed to 2 µg BPS/kg/d had significantly shorter latency to retrieve their entire litter on LD7 [Supplemental Fig. 1(C)]. Taken together, these results suggest different effects of BPS on pup retrieval depending on the period of exposure: F0 mothers that attempt retrievals have a longer latency to interact with and retrieve pups, whereas F1 females that attempt retrievals have a shorter latency to interact with and retrieve pups.

BPS does not affect behavior of F0 or F1 dams in open-field assays

To determine whether some of the differences we observed in maternal behavior assays could be attributed to BPS-induced changes in anxiety-like behavior or locomotor activity levels, we used the open-field test to evaluate F0 and F1 females before and after parturition. Increased time spent in the center of the open field is associated with lower anxiety-like behavior (64–69). In evaluations conducted on pregnancy day 16 and LD10/11, BPS exposure was not associated with any changes in behavior in the open field, including the number of rears in the center of the field or in the number of grooming events (Fig. 5). No significant effects of BPS were observed for the number of freeze/stops, number of fecal pellets, or the number of rears on the walls of the open-field apparatus (data not shown). Collectively, these data suggest that BPS does not induce anxiety-like behaviors in females exposed during pregnancy and lactation (F0 females) or in females exposed during development upon reaching adulthood (and pregnancy/lactation in particular).

Figure 5.

BPS exposure does not affect behaviors in the open-field test in either the F0 or F1 generations. (A) BPS exposure did not affect the number of center rears on pregnancy day 16 in F0 or F1 dams. (B) BPS exposure did not affect the number of grooming events in F0 or F1 dams on pregnancy day 16. (C) BPS exposure did not affect the number of center rears on LD10/11 in either F0 or F1 dams. (D) BPS exposure did not affect the number of grooming events on LD10/11. Red graphs indicate F0 dams and blue graphs indicate F1 dams.

ERα expression in the cMPOA is affected by BPS exposures in the F0 but not the F1 generation

To determine the effect of BPS on brain regions that are considered relevant for maternal behavior, neural cells expressing ERα were quantified in 2 matched sections of the cMPOA in F0 dams (on LD21) and F1 dams (on LD2 and LD21). The first subregion of interest was selected from the rostral cMPOA [Fig. 6(A)], whereas the second was selected from the caudal cMPOA [Fig. 6(B)]. BPS treatment appeared to increase ERα expression in the rostral cMPOA in F0 dams, but these differences were not statistically significant [Fig. 6(C)]. We did observe a significant increase in ERα immunoreactivity in the caudal cMPOA in F0 dams exposed to 200 μg BPS/kg/d relative to the control dams [Fig. 6(D)].

Figure 6.

BPS treatment increases ERα expression in the cMPOA of F0 but not F1 dams. (A) Photomicrograph illustrating representative examples of the rostral cMPOA from control and BPS-treated F0 females, ×120 magnification. (B) Photomicrograph illustrating representative examples of the caudal cMPOA from control and BPS-treated F0 females, ×120 magnification. (C) There were no treatment-related effects on ERα expression in the rostral cMPOA of F0 females on LD21 or F1 females on LD2 and LD21. (D) On LD21, F0 dams exposed to 200 µg BPS/kg/d had increased ERα expression in the caudal cMPOA. There were no exposure-related effects in F1 females on either LD2 or LD21 in the caudal cMPOA.*P < 0.01 Bonferroni post hoc statistic after significant analysis of variance.

In contrast to the effects observed in F0 dams, there were no significant differences in ERα expression in either subregion of the MPOA for F1 dams at LD2 or LD21 [Fig. 6(C) and 6(D)]. These results are consistent with an activational effect but not an organizational effect of BPS on ERα expression in the MPOA.

TH immunoreactivity is not affected by BPS exposure in either the F0 or the F1 generation

We next quantified TH-immunoreactive cells in the VTA in anatomically matched sections in F0 dams on LD21 and F1 dams on LD2 or LD21 [Fig. 7(A)]. The VTA receives functional input from the MPOA and also plays an important role in maternal behavior. There were no significant effects of BPS treatment on TH immunoreactivity in the VTA in the exposed groups in either the F0 or F1 generation [Fig. 7(B)].

Figure 7.

TH immunoreactivity in the VTA is not affected by BPS treatment. (A) Photomicrograph illustrating the region of the VTA that was evaluated in F0 females (at LD21) and in F1 females (at LD2 and LD21). Shown here are representative images of TH immunoreactivity in the VTA of F0 females, ×100 magnification. (B) There were no treatment-related effects on TH immunoreactivity in the VTA in either F0 or F1 dams at any time point examined.

Discussion

To our knowledge, we have evaluated for the first time the effects of exposure to BPS, a BPA replacement chemical, on maternal behavior and brain in a mammalian model using CD-1 mice. We specifically characterized the effects of BPS on F0 generation dams, exposed during pregnancy and lactation, and F1 generation females, exposed in utero and during the perinatal period. We found that maternal behavior is sensitive to BPS exposure and observed different outcomes depending on generation, postpartum period, and dose. We observed a surprising increased incidence of infanticide in F1 females exposed to the lower dose of BPS. Although these same effects were not seen at the higher dose, more than 10% of females exposed to 2 μg BPS/kg/d either killed their pups or provided such poor instrumental maternal care that 1 or more pups needed to be euthanized. Although not statistically significant, the neglect and poor maternal care we observed were striking. In contrast to our behavioral findings, ERα expression in the MPOA was significantly affected by BPS exposure in only the females in the F0 generation.

With regard to maternal behavior, F0 dams exposed to the higher dose of BPS (200 µg/kg/d) spent significantly more time on the nest on LD14, with no significant differences between treatment groups at other time points across the lactational period. As pups grow and develop and the postpartum period progresses, dams spend less time on the nest (70). Thus, these results suggest that BPS exposure in the F0 dams induces an extension of a behavior that is generally diminished by this stage of the postpartum period; this may indicate a lack of adjustment in the dam to the changing needs of her pups (71). In contrast, dams of the F1 generation developmentally exposed to either 2 or 200 µg BPS/kg/d spent significantly more time away from the nest on LD2 and LD7 compared with controls. During these early postpartum periods (LD2 and LD7), the pups require protection and constant care for their growth and survival, including attention to thermoregulation and feeding (71, 72); thus, the F1 BPS-exposed dams display quantitatively diminished care.

Observations suggesting an inability to attend to the changing development and needs of the pups may also be extended to measures of nest building. Nest building is dependent on hormone levels during pregnancy and is important for thermoregulation and protection of pups [see Catanese et al. (55)]. In F1 dams, BPS treatment increased the observed time spent nest building on LD14, which may indicate a repetitive or obsessive-compulsive disorder–like behavior (73–76). By this age, pups have developed fur and are better able to thermoregulate in comparison with earlier time points. The ability of most BPS-exposed females to build a maternal nest and attend to and groom their young indicates that the effects we observed in some components of maternal behavior are not due to wholesale changes in instrumental maternal care.

Additional components of maternal behavior evaluated include time to first touch and time to retrieve pups to the nest. When all dams were evaluated, we observed no significant effects of BPS on time to retrieve 1 or all pups in either F0 or F1 dams and relatively modest effects on the time to first touch pups in the F0 generation. Some dams did not retrieve pups, even on LD2, but this failure to retrieve was not influenced by treatment in either F0 or F1 dams (Supplemental Fig. 1). To investigate effects of BPS treatment on retrieval strategies, one analysis we conducted evaluated only those dams that did retrieve pups. In the F0 maternally “active” dams, exposure to either dose increased the latency to interact with pups (Supplemental Fig. 1), suggesting alterations to approaching and retrieving pups to the safety of the nest, which could have serious repercussions for litter survival in the wild. In striking contrast, quantitatively opposite effects of BPS were observed in F1 dams (Supplemental Fig. 1). In this generation, of those dams that retrieved 1 or more pups, BPS-exposed females showed significantly shorter latency to retrieve their first pup and significantly shorter latency to retrieve their entire litter on LD7. It is tempting to interpret this rapid pup retrieval as “improved” maternal behavior, but it remains in question whether more rapid retrieval should be considered an improvement in care, as it may indicate hyperactivity, compulsivity-like behavior, heightened stress response to scattered pups, or a displaced form of retrieval. Results from open-field tests are not consistent with increased anxiety-like behaviors in BPS-treated females in either generation. However, preliminary data from our laboratory suggest that BPS may induce hyperactivity in F1 females prior to the onset of puberty [Catanese and Vandenberg, unpublished observations, and Kim et al. (42)], which is consistent with the effects observed in F1 dams’ pup retrievals.

We hypothesized that the maternal brain may be sensitive to exogenous estrogenic exposures and evaluated whether BPS treatment altered 2 brain regions considered important for maternal behavior, the MPOA and VTA. Whereas the VTA was not affected by BPS exposure in either generation, we observed a significant increase in ERα expression in the caudal cMPOA in F0 dams exposed to 200 µg BPS/kg/d on LD21 and a nonsignificant increase in the rostral subregion of the cMPOA in the same animals (Fig. 6). There were no significant differences observed in ERα expression in the cMPOA of F1 dams at either LD2 or LD21.

The effects of overexpression of ERα are considered to be tissue dependent (77). In the brain, for example, induced overexpression of ERα in the hippocampus of aging female rats has been shown to enhance memory (78, 79). ERα expression is upregulated in response to neuronal injury and is considered neuroprotective (79). Although this present study did not investigate mechanisms underlying the increase in receptor expression, ERα expression in the MPOA is critical for maternal care in the mouse, and its regulation remains under investigation. Importantly, there appear to be differences in regulation of ERα depending on stage of estrous as well as in relation to pregnancy and lactation, indicating that endocrine state is key in assessing ER regulation in this region. Estradiol downregulates ERα gene expression in the MPOA of cycling adult female rats (80), and protein levels fluctuate across the estrous cycle (81). Increased messenger RNA and protein expression occur later in pregnancy and may be regulated by transcriptional mechanisms such as transcription rate (3, 82, 83); receptor turnover and receptor phosphorylation may also account for increases in ER expression (82). In the mouse, ERα expression in the MPOA is thought to be positively correlated with estradiol levels, but the relationship between estradiol and ERα expression is not fully understood (84). A recent study found that letrozole, an aromatase inhibitor, interfered with pup retrieval and latency to build a nest in ovariectomized virgin females and induced increased messenger RNA levels of ERα, indicating a role for brain-derived estrogen in the regulation of ERα gene expression in the MPOA (85).

To date, there are relatively few data to predict the effects of xenoestrogens on ER expression in the MPOA. A study of nonpregnant or lactating Sprague-Dawley rats examined the effects of high doses of BPA (40 mg/kg/d) on ERα immunoreactive cells in the MPOA, ventromedial hypothalamus, and arcuate nucleus. Although lactating females had fewer ERα-expressing cells in the MPOA compared with nonlactating females, there were no significant effects of BPA in the lactating group (86). Here, the response to BPS we observed in the MPOA in F0 females supports the idea that pregnancy/lactation represents a vulnerable period for exposures and that the effects may be “activational” (e.g., only occurring during the period of exposure) (87–89). Determining the mechanism underlying the increase in ERα in the MPOA in the F0 generation, the persistence of this effect beyond the period of exposure, as well as potential downstream neural and/or genomic effects, will be important to investigate in the future.

We were surprised to find that ERα expression in the cMPOA of F1 dams was not altered by developmental BPS exposure as early life exposures to BPA have been shown to alter ERα expression. Neonatal exposure to 0.05 mg/kg and 20 mg/kg BPA demonstrated dose-dependent effects; there was increased ERα expression on postnatal day (PND) 8 and PND21 in response to the lower 0.05-mg/kg dose and decreased expression on PND8 and PND21 in response to the higher 20-mg/kg dose (90). In an additional study, maternal exposure to 2, 20, or 200 µg BPA/kg/d in BALB/c mice during pregnancy (gestational days 0 to 19) led to dose- and sex-specific changes in the expression of gene encoding of ERα, ERβ, and ERRγ in the hypothalamus of offspring (30). Effects of 20 µg BPA/kg/d on ERα expression in the hypothalamus of female offspring was correlated with reduced methylation of the gene coding for ERα. It is, of course, plausible that our F1 females may have been affected by BPS exposure, but ERα immunoreactivity in the MPOA is not the appropriate measure of this effect. Other relevant levels of analysis may include gene expression or epigenetic modifications, as ERα methylation status differs depending on the quality of maternal behavior in rodents (91, 92). We cannot rule out the possibility of effects on other neural end points not examined or that developmental exposures to BPS may have had early transient or stage-dependent effects; further experiments are necessary to investigate additional end points and to uncover any potential change to expression earlier in life. Alterations to maternal behavior in F1 females exposed to BPS were striking, suggesting that effects on other neural end points are likely; identifying the correct end points and establishing a causal link between the neural and behavior effects will be important to investigate in future work.

Prior work has demonstrated an impact of BPA (28) and many other EDCs on maternal behavior (55, 93, 94). Due to differences in study design, timing of treatment, and dosage, it is difficult to compare our study with this seminal work (28). In many studies examining developmental effects of EDCs, the dam is considered a vessel through which offspring are exposed and thus is rarely examined herself (55, 93). Alonso-Magdalena et al. (95, 96) are challenging this paradigm, and strong evidence suggests that pregnancy may be considered a “vulnerable period of development” for the mother. Here we demonstrate that adult exposures can induce both neural and behavioral changes in exposed dams, including changes to maternal behavior, thus suggesting that pregnancy and the lactational period are sensitive to endocrine disruption. It is important to note that effects on the mother may affect her offsprings’ behavioral and neural development (97). One major challenge is to distinguish the direct effects of EDC exposures on the pups from indirect effects observed in pups due to altered maternal behaviors (28, 29, 98). In a small number of studies, cross-fostering has been used to isolate the “direct” effects of EDC exposures on the mother’s behavior from “indirect” effects on offspring (29, 98, 99). Here, our F1 females were directly exposed to BPS during gestation and perinatal development, and these females were also raised by dams that were exposed and affected by this compound. Thus, the effects observed in F1 females may be due to their own exposures or to alterations to the maternal behavior in their mothers, or a combination of these factors. Furthermore, changes in maternal behavior (displayed by either F0 or F1 dams) could be indirect effects (e.g., in response to changes in their nursing pups). Cross-fostering experiments will be necessary to delineate whether the effects of exposures on the F1 generation are direct or indirect. However, care must be taken in study design as maternal behavior can be affected in both a positive and negative direction by cross-fostered pups (29, 100–102).

To our knowledge, ours is the first study to examine whether exposure to one of the current BPA replacement chemicals, BPS, affects maternal behavior as well as maternally relevant neural correlates. A growing body of evidence indicates that exposure to environmental chemicals can alter maternal behavior [refer to Catanese et al. (55) and Cummings et al. (93)], and disruptions to maternal care can profoundly influence the development of subsequent generations, which are also sensitive to developmental and environmental impacts (97, 103–106). A combination of neural, endocrine, behavioral, and cognitive processes drives proper maternal care, allowing mothers to adjust to the needs of the young over the course of their early development (71, 107–109). Our results suggest that these processes become impaired after BPS exposures with differing effects based on dose, postpartum period, and generational timing of exposure. The results from our study highlight the need to understand the effects of xenoestrogens on maternal behavior and to better understand the neural mechanisms underlying these behavioral changes. From an evolutionary perspective, maternal behavior is often defined in light of its importance for the survival of offspring; the importance of proper maternal care in humans stems from its contribution to the physical, emotional, and psychological development of children, whereas the negative impact of poor maternal care can last into adulthood. Therefore, uncovering effects of environmental chemicals that might influence proper maternal care have broad social and public health implications.

Abbreviations:

 
• BPA

  bisphenol A

 
• BPS

  bisphenol S

 
• cMPOA

  central medial preoptic area

 
• EDC

  endocrine disrupting chemical

 
• ERα

  estrogen receptor alpha

 
• LD

  lactational day

 
• MPOA

  medial preoptic area

 
• PND

  postnatal day

 
• TH

  tyrosine hydroxylase

 
• VTA

  ventral tegmental area

Acknowledgments

The authors gratefully acknowledge input from Drs. R. Thomas Zoeller, Jerrold Meyer, and Mariana Pereira. We also thank members of the Vandenberg laboratory who helped with behavioral data collection, including Corinne Hill, Durga Kolla, Charlotte LaPlante, Sarah Sapouckey, and Alfred Kimani.

The authors acknowledge support from the University of Massachusetts (startup funds to L.N.V.) and Award K22ES025811 from the National Institute of Environmental Health Sciences of the National Institutes of Health (L.N.V.). M.C.C. was supported by the Center for Research on Families Graduate Student Research Grant, 2015–16, University of Massachusetts–Amherst. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the University of Massachusetts.

Author contributions: M.C.C. and L.N.V. designed and conducted experiments and collected and analyzed data. M.C.C. wrote the first draft of the manuscript. L.N.V. edited the manuscript.

Disclosure Summary: L.N.V. has received travel reimbursement from universities, governments, nongovernmental organization, and industry to speak about endocrine-disrupting chemicals. M.C.C. has nothing to disclose.

Author notes

Address all correspondence and requests for reprints to: Laura N. Vandenberg, PhD, School of Public Health & Health Sciences, Department of Environmental Health Sciencces, University of Massachusetts–Amherst, 149B Goessmann, 686 North Pleasant Street, Amherst, Massachusetts 01003. E-mail: .