The present study reveals that early life exposure to BPA elevates juvenile anxiety in rats of both sexes, but that these effects can be mitigated by concomitant intake of a soy-rich diet. Our data are novel in that they evaluated outcomes of exposure at adolescence, a life stage of great hormonal and neurobiological change, and identified specific gene targets in the amygdala. Furthermore, our data are consistent with recent studies reporting that developmental exposure to BPA is associated with hyperactivity and increased anxiety in young girls 
, but indicate that diet, or other factors, could alleviate these effects.
The behavioral outcomes described here are consistent with prior reports showing that developmental BPA exposure has anxiogenic effects in juvenile and adult C57BL/6J mice 
and deer mice (Peromyscus maniculatus
; and abrogates behavioral sex differences in adult CD-1 mice 
. A notable difference between the present and prior studies, however, is that here the overall anxiogenic effect of BPA did not strongly persist into adulthood. Several factors could account for this difference, the most obvious of which is a species effect. For example, mice exhibit marked strain differences in anxiety levels 
and there are numerous important neuroendocrine differences between mice and rats which manifest as species-specific behavioral attributes 
. Dose is also likely a factor as it varies considerably across studies. Use of more than one dose was beyond the scope of the present study, but future work will be necessary to identify the effective dose range for BPA-related effects on affective behavior. More likely, however is that the minimization of perinatal stress, a well-recognized confound of adult affective behavior, accounts for the less robust effects of BPA on adult anxiety 
. By using drinking water and diet as the route of administration for BPA and soy phytoestrogens, we did not have to disturb the dams or their pups to dose the animals. Thus, this study is unique because it is the first to orally administer BPA at a dose sufficient to produce human-relevant serum levels 
while simultaneously minimizing the pronounced confound of perinatal stress. With the adverse impacts of early life stress eliminated, the resulting behavioral changes reported here are more clearly attributable to BPA exposure.
Concomitant administration of a soy-based diet with BPA resulted in attenuation of the BPA effects. In both the Light/Dark box and the EPM, soy largely abrogated the anxiogenic impact of BPA on the percent of rats engaging in the “risky” behavior. Sex differences in adult behavior were maintained except in the case of latency to enter an open arm, where females displayed a longer latency. But again, co-administration of soy and BPA rescued the reversed sex difference induced by BPA exposure. These data are the first to suggest that adverse behavioral outcomes resulting from early life exposure to BPA may be mitigated, at least to some degree, by dietary or other factors. Presumably, this mitigating activity is attributable to GEN, or one (or more) of the numerous other isoflavone phytoestrogens aside from GEN that are commonly found in soy-based foods 
. Although identification of the specific dietary compounds capable of minimizing the adverse neurobehavioral impacts of BPA and other EDCs is potentially advantageous to pregnant women, it is important to recognize that GEN and other phytoestrogens can also act as EDCs during development. Thus, the critical windows in which they produce effects, and their mechanism(s) of action and interaction on neural systems that mediate behavior must be more clearly elucidated.
The mechanisms by which BPA alters affective behavior remain poorly understood, but the amygdalar gene expression data obtained here identified 8 genes associated with BPA-induced anxiety including Esr2
. These genes have previously been associated with sociosexual and affective behaviors. Interestingly, a recent paper showed transgenerational, epigenetic effects of vinclozolin (a fungicide) on the male F3 generation's responses to stress during adolescence 
. The male F3 vinclozolin-lineage descendants differed from vehicle-lineage counterparts in how adolescent stress affected subsequent performance on tests of anxiety/stress, and that these same animals differed in expression of genes in hippocampus and amygdala, including Mc3r
, similar to results in BPA-exposed F1 rats in the current study.
, changes to which were, in the present study, detected in the amygdala, play crucial roles in the regulation of oxytocin and vasopressin secretion 
. These two neuropeptides are well-established to be essential for mediating social interactions and affiliation in rodents and primates, including humans 
. For example, Esr2 agonism has been shown to be anxiolytic 
, and in the paraventricular nucleus (PVN), Esr2 is required to drive estrogen-dependent oxytocin and vasopressin production 
. α-Melanocyte stimulating hormone (α-MSH), acting through Mc4r, induces dendritic release of oxytocin from magnocellular PVN neurons 
. Central actions of oxytocin and vasopressin are implicated in a suite of sociosexual behaviors 
and confer anxiolytic effects 
. Collectively, these observations suggest that disrupted ontogeny of oxytocin and vasopressin signaling pathways may underlie the observed changes in juvenile affective behavior. In further support of this hypothesis, a recent study using mice found that prenatal BPA exposure had significantly lower whole brain levels of oxytocin and vasopressin just prior to birth compared to unexposed controls 
. Dysregulation of these neuropeptide signaling pathways has been implicated in a range of childhood affective disorders, including autism 
. Although oxytocin and vasopressin gene expression levels were not significantly altered by exposure or diet in the present study (), the amygdala may not be the most salient region to assess expression levels of these neuropeptides because they are generated primarily in the PVN and supraoptic nucleus (SON). Thus, subsequent investigation of oxytocin and vasopressin production in the PVN and SON will be necessary to confirm the hypothesis that these neuropeptides may be altered by early life BPA exposure.
Notably, BPA significantly impacted the expression of Kiss1
, a gene only recently discovered to be expressed in the amygdala 
. This gene codes for the peptide kisspeptin, expression of which in the hypothalamus is now recognized to play a critical role in the timing of sexual maturation, female ovulation, and feedback mechanisms on the hypothalamic-pituitary-gonadal (HPG) axis 
. Kisspeptin neurons are sparse in number and primarily confined to the medial amygdala, with males having more than females. In the present study, this sex difference in amygdalar Kiss1
expression was also observed, but, unexpectedly, EE did not masculinize expression in females. Instead, it feminized expression in males (), an effect which is unusual in the rodent brain 
. BPA also reduced Kiss1
expression in males, an effect which was enhanced in animals maintained on the soy diet, suggesting that this specific effect may be estrogenic. Although adult expression is readily identifiable in the medial amygdala 
, ongoing, concurrent studies in our laboratory indicate that expression is not detectable in pre-weanlings suggesting that a mature HPG axis may be required for maximal expression of Kiss1
. The functional role of these neurons remains to be delineated. Prior studies have shown that central, but not peripheral, kisspeptin administration elicits oxytocin secretion in females and vasopressin secretion in males, suggesting a role for amygdalar kisspeptin neurons in the modulation of affective behavior 
. Further anatomical studies are needed to more comprehensively characterize the ontogeny of this neuronal population, their sensitivity to endocrine disruption, and their functional role in sociosexual behaviors.
The hypothesis that early life exposure to EDCs such as BPA and soy phytoestrogens can alter behavior is further supported by parallel work in rodents and humans showing that early life epigenetic changes within the hypothalamic-pituitary-adrenal (HPA) axis, arising from environmental influences, can permanently alter stress sensitivity and coping strategies 
. It also implicates an epigenetic mechanism of action. Here, EE exposure did not completely recapitulate the behavioral or transcriptional effects of BPA, soy, or the combination of the two, demonstrating that BPA and soy are not simply acting as estrogens. Prior work has established that BPA can induce DNA methylation changes that are blocked by concurrent administration of the soy isoflavone GEN 
, raising the possibility that a similar interaction may have occurred here, resulting in the dietary-dependent behavioral outcomes. One possibility is that differential methylation of the ERβ promoter resulted in decreased expression, and subsequently the decreased expression of Mc4r
and other genes regulated by Esr2
activity, but more extensive work is needed to test this hypothesis. Understanding the specific cellular and molecular mechanisms by which early life BPA exposure alters behavior is critical for determining if effects observed in animal models have implications for human health.
Finally, it is important to highlight that the dosing method used here produced serum BPA levels at all phases of the project that were equivalent to, or below, those reported in humans 
. Because BPA was administered in drinking water, exposure was likely low but continuous throughout the day, a pattern that more closely models that of humans than gavage or other methods of bolus administration. Although trace levels of free BPA were observed in some unexposed controls, suggesting an alternative and uncontrolled source of exposure, it may be an artifact of the analysis because the levels were so close to the limit of detection. The most likely source is diet, as we routinely monitor our caging leachate and water to ensure that they are BPA-free. Soy phytoestrogen exposure was monitored by assessing serum GEN and its metabolites. Serum levels were well below those seen in infants exclusively fed soy-based infant formula 
. It has long been hypothesized that GEN, BPA and other EDCs are not readily metabolized in neonates, and the absence of the glucoronidated form in PND 12 serum is consistent with this view. Serum levels reported here are high enough to induce physiological effects in rat models 
. Exposure to BPA and GEN was likely lowest during lactation because, although placental transfer of both compounds have been established, neither is known to lactationally transfer readily 
Affective disorders in adults and children have well recognized sex differences in etiology. Boys are at higher risk of autism spectrum disorders, ADHD, and early onset schizophrenia 
while women disproportionately suffer from anxiety, major depression, panic and eating disorders 
. Notably, male-biased disorders appear to have their origins in development while female-biased disorders are generally post-pubertal in onset implying that the windows of sensitivity to environmental exposure may be sexually dimorphic with males being more sensitive during development and females later in life 
. The data shown here reveal the potential for BPA and other EDCs to alter behaviors associated with anxiety, but also the potential for those alterations to be modified/mitigated by a soy-rich diet. These observations are a critical reminder that not all EDCs are synthetic and that plant compounds also have the potential to interact with hormone-sensitive mammalian systems, including the brain. Gene expression data from the amygdala imply a role for Esr2, melanocortins, and oxytocin/vasopressin signaling pathways in the manifestation of these behavioral changes, but future work will be needed to confirm this assertion. Collectively, these data highlight the plasticity of complex behaviors, their sex differences, and their potential for alteration by chemicals in the environment.
Finally, while humans and rodents perceive and express stress and anxiety differently, important core elements of the genetic and neurobiological basis of anxiety phenotypes are evolutionarily conserved across species, particularly in the amygdala and hypothalamus 
. Thus the rat is an appropriate animal model for understanding how the interaction of BPA and diet influence sociosexual behaviors, and identifying the neural mechanisms by which these changes are induced.