Maternal Stress and Effects of Prenatal Air Pollution on Offspring Mental Health Outcomes in Mice

Background: Low socioeconomic status is consistently associated with reduced physical and mental health, but the mechanisms remain unclear. Increased levels of urban air pollutants interacting with parental stress have been proposed to explain health disparities in respiratory disease, but the impact of such interactions on mental health is unknown. Objectives: We aimed to determine whether prenatal air pollution exposure and stress during pregnancy act synergistically on offspring to induce a neuroinflammatory response and subsequent neurocognitive disorders in adulthood. Methods: Mouse dams were intermittently exposed via oropharyngeal aspiration to diesel exhaust particles (DEP; 50 μg × 6 doses) or vehicle throughout gestation. This exposure was combined with standard housing or nest material restriction (NR; a novel model of maternal stress) during the last third of gestation. Results: Adult (postnatal day 60) offspring of dams that experienced both stressors (DEP and NR) displayed increased anxiety, but only male offspring of this group had impaired cognition. Furthermore, maternal DEP exposure increased proinflammatory interleukin (IL)-1β levels within the brains of adult males but not females, and maternal DEP and NR both decreased anti-inflammatory IL-10 in male, but not female, brains. Similarly, only DEP/NR males showed increased expression of the innate immune recognition gene toll-like receptor 4 (Tlr4) and its downstream effector, caspase-1. Conclusions: These results show that maternal stress during late gestation increases the susceptibility of offspring—particularly males—to the deleterious effects of prenatal air pollutant exposure, which may be due to a synergism of these factors acting on innate immune recognition genes and downstream neuroinflammatory cascades within the developing brain. Citation: Bolton JL, Huff NC, Smith SH, Mason SN, Foster WM, Auten RL, Bilbo SD. 2013. Maternal stress and effects of prenatal air pollution on offspring mental health outcomes in mice. Environ Health Perspect 121:1075–1082; http://dx.doi.org/10.1289/ehp.1306560


Introduction
Although low socioeconomic status (SES) has been repeatedly associated with a higher rate of chronic health problems and mental disorders (Adler and Rehkopf 2008;Reijneveld and Schene 1998), explicit characterization of the factors that underlie this phenomenon remains elusive (Evans and Kantrowitz 2002). A growing body of research suggests that maternal well-being during pregnancy is a crucial determinant of lifelong physical and mental health of the offspring (Case et al. 2005;Hackman et al. 2010;Susser et al. 1999). Notably, expectant mothers living in low-SES conditions experience the greatest burden of toxicants (Evans and Kantrowitz 2002), along with fewer resources and high psychological stress (Seguin et al. 1995). Exposures to toxicants such as lead are well known to adversely affect brain develop ment (Needleman et al. 1990; Weiss and Landrigan 2000). However, "social toxins," such as violence, poverty, and other factors that generate psychological stress in low-SES parents and children, have only recently begun to gain recognition as risk factors that can also alter the trajectory of brain development (Wright 2009). For example, Clougherty et al. (2007) noted an association between air pollution and asthma only in children who were also living with a chronic stressor (e.g., domestic violence). Similarly, parental stress can increase the effect of in utero exposures to toxicants (i.e., tobacco smoke) on childhood asthma risk (Shankardass et al. 2009). In such cases, stress may increase vulnerability, permitting a toxicant to initiate significant injury to physio logical systems greater than any injury resulting from exposure to the toxicant alone. Importantly, these synergistic effects of stress and pollutants are possible because they likely act on common biological systems, such as innate immune pathways (Frank et al. 2007;Levesque et al. 2011) within the developing nervous system.
Air pollution, one of the most relevant and pervasive environmental toxicants in the modern world, is a particularly important threat to child health and is increasingly associated with neuro developmental disorders such as autism (Volk et al. 2013). Mechanistic studies have revealed that diesel exhaust, a major component of air pollution, markedly activates micro glia, the resident immune cells of the brain, in adult rats (Levesque et al. 2011). Furthermore, we observed in mice that maternal exposure to diesel exhaust causes long-term increases in micro glial antigen expression in the brains of adult offspring (Bolton et al. 2012). Notably, prior stress has been shown to enhance pro inflammatory cytokine expression and associated neural damage following an immune challenge in adult rats (De Pablos et al. 2006), most likely by sensitizing micro glia (Frank et al. 2011). Because cytokines are important for normal brain development and adult function (Boulanger 2009;Yirmiya and Goshen 2011), signifi cant perturbations in their expression may have enduring consequences for lifelong mental health (Dantzer et al. 2008).
We hypothesized that the addition of maternal stress to the impact of pre natal air pollution exposure would act synergistically in offspring to impair mental health outcomes, compared with the effects of either exposure alone. To test this hypothe sis, we combined our animal model of pre natal diesel exhaust exposure  with an adaptation of a novel model of maternal resource deprivation [nest restriction (NR)] (Rice et al. 2008). Pregnant mice exposed by inter mittent aspiration to vehicle or diesel exhaust particles (DEP) throughout gestation either lived in normal housing or had reduced nesting materials during the last one-third of pregnancy.
Background: Low socioeconomic status is consistently associated with reduced physical and mental health, but the mechanisms remain unclear. Increased levels of urban air pollutants inter acting with parental stress have been proposed to explain health disparities in respiratory disease, but the impact of such interactions on mental health is unknown. oBjectives: We aimed to determine whether prenatal air pollution exposure and stress during pregnancy act synergistically on offspring to induce a neuroinflammatory response and subsequent neuro cognitive disorders in adulthood. Methods: Mouse dams were intermittently exposed via oropharyngeal aspiration to diesel exhaust particles (DEP; 50 μg × 6 doses) or vehicle throughout gestation. This exposure was combined with standard housing or nest material restriction (NR; a novel model of maternal stress) during the last third of gestation. results: Adult (postnatal day 60) offspring of dams that experienced both stressors (DEP and NR) displayed increased anxiety, but only male offspring of this group had impaired cognition. Furthermore, maternal DEP exposure increased pro inflammatory interleukin (IL)1β levels within the brains of adult males but not females, and maternal DEP and NR both decreased antiinflammatory IL10 in male, but not female, brains. Similarly, only DEP/NR males showed increased expression of the innate immune recognition gene tolllike receptor 4 (Tlr4) and its downstream effector, caspase1. conclusions: These results show that maternal stress during late gestation increases the suscepti bility of offspring-particularly males-to the deleterious effects of pre natal air pollutant exposure, which may be due to a synergism of these factors acting on innate immune recognition genes and downstream neuroinflammatory cascades within the developing brain.

Animals.
We obtained adult male and female C57BL/6 mice from Charles River Laboratories (Raleigh, NC) and timemated them in four separate cohorts (see Supplemental Material, p.2). After confirming successful mating of animals [vaginal plug, considered to be embryonic day (E) 0], we paired females in individually ventilated cages with specialized bedding (AlphaDri; Shepherd Specialty Papers, Milford, NJ) and ad libitum access to food (PicoLab Mouse Diet 5058; Lab-Diet, Philadelphia, PA) and filtered water. Each of the four cohorts was used for a separate analysis: a) cytokine protein in fetal brain, b) gene expression in postnatal day (PND) 30 brain, c) adult behavior and brain cytokine protein, and d) adult micro glial isolations and gene expression. In addition, one pup was randomly selected at PND1 from litters across the second and third cohorts for cortico sterone (CORT) analysis. All mice used in this study were treated humanely and with regard for alleviation of suffering, and experiments were conducted using protocols approved by the Duke University Animal Care and Use Committee.
Prenatal stressors. DEP exposures. Beginning on E2, time-mated females were lightly anesthetized with 2% isoflurane for approximately 1 min and treated with DEP via oro pharyngeal aspiration ) (see Supplemental Material, pp. 2-3). Females received 50 μg DEP suspended in 50 μL vehicle [phosphate buffered saline (PBS), pH 7.2, 0.05% Tween-20] or vehicle alone (VEH) every 3 days during E2-E17 for a total of six doses, as a model of intermittent exposure. This dose and route of delivery induces maternal lung inflammation (e.g., white blood cell infiltration) comparable to levels observed following intermittent maternal inhalation of diesel exhaust at environmentally relevant concentrations . Moreover, both routes of delivery result in similar levels of particle deposition within the lung (Foster et al. 2001).
Maternal resource deprivation. We adapted a model of postnatal NR by applying the method of Rice et al. (2008) to the prenatal period, a degree of restriction that produced minimal behavioral changes with NR alone. On E14, after DEP exposures, half of the VEH-and DEP-treated dams were singly housed in clean cages with a thin layer of bedding under an elevated fine-gauge aluminum mesh platform (mesh dimensions, 0.4 cm × 0.9 cm; McNichols Co., Tampa, FL) and provided with two-thirds of one square of felt-like nesting material (~ 1.9 g; NR group). The remaining dams were singly housed in clean cages with bedding and one full square of nesting material (~ 2.8 g; control group). On E19, we placed NR dams in clean cages with normal bedding and one full square of nesting material; from that point on, NR animals were treated identically to control dams. This design resulted in four groups of dams: VEH/control (n = 8), DEP/control (n = 10), VEH/NR (n = 8), and DEP/NR (n = 10).
Neonatal outcomes and maternal behavior. Birth weights. All animals were allowed to deliver spontaneously on E19-E20 (defined as PND0), and offspring were not cross-fostered. To acquire a litter average for birth weight (n = 8-10 litters/group from three cohorts), we weighed pups (sex not determined) on PND1. Four weeks later, offspring were weaned into clean cages of two to five same-sex siblings and provided with ad libitum access to standard chow and filtered water.
Neonatal CORT measurement. On PND1, we randomly selected one pup from each litter (males, n = 5-8/group; females, n = 3-7/group from two cohorts), performed rapid decapitation, and collected trunk blood to obtain a measure of basal circulating CORT levels soon after birth. We assessed total serum CORT concentrations using an enzyme-linked immuno sorbent assay (ELISA; Enzo Life Sciences Inc., Ann Arbor, MI) (Bilbo et al. 2007).
Maternal behavior assessment. To characterize the effect of pre natal stressors on maternal care, we observed dams (n = 3-7/group from two cohorts) twice daily with their litters on PND2-PND9 to determine time spent on the nest, nursing, and licking and grooming their pups (Myers et al. 1989) (see Supplemental Material, p. 4).
Fetal brain cytokine analysis. At E18, VEH-or DEP-treated dams with or without NR (n = 2-3/group from one cohort) were euthanized by sodium pento barbital anesthesia [250 mg/kg by intra peritoneal injection (i.p.)]. Fetuses were removed by hysterotomy, placed on ice, and decapitated. Whole fetal brains were snap-frozen and stored at -80°C until processing. To determine the sex of each fetus, we extracted genomic DNA (Kouduka et al. 2006) from tail snips for later geno typing.

PND30 neuroimmune gene expression.
To determine the long-term effect of pre natal stressors on genes critical for innate immune recognition and the subsequent cytokine response in the brain, we assessed male (n = 8-13 pups from 2-4 litters per treat ment group from one cohort) and female (n = 5 pups from 2-4 litters per treatment group from one cohort) offspring at PND30. Mice were deeply anesthetized with ketamine (430 mg/kg) and xylazine (65 mg/kg; i.p.) and trans cardially perfused with ice-cold saline for 2 min to clear brains of blood. Afterward, we extracted the brains, removed the cerebellum and hindbrain, and cut the remaining forebrain sagittally in half. Half brains were snap frozen and stored at -80°C until analysis by quantitative realtime polymerase chain reaction (qRT-PCR) as described by Williamson et al. (2011) (see Supplemental Material, p. 4 and Table S1).
Behavioral procedures. We assessed behavioral outcomes as a result of pre natal stressors in young adult (PND60-PND90) male and female offspring (n = 7-9 animals/sex from 2-4 litters per treatment group from one cohort) using a sequence of behavioral tests, with 1 week between each test. We tested males and females separately and performed all testing during the dark cycle (between 1000 hours and 1600 hours). Throughout testing, we also monitored females' estrous cycles.
Contextual and auditory cue fear con ditioning. We assessed memory as described in detail for rats (Williamson et al. 2011), but with slight modifications for mice (see Supplemental Material, pp. 5-6). Briefly, mice were trained to associate a foot shock with a specific context and an auditory tone. We tested each mouse's memory of these associations 48 hr later by assessing freezing behavior (the proto typical rodent fear response) in the fear context, in a new context, and in response to the auditory cue [conditioned stimulus (CS)] in the new context. We used contextual fear conditioning to assess hippocampaldependent memory, which has been reported to be uniquely vulnerable to early-life insults (Williamson et al. 2011), compared with auditory cue fear conditioning, which does not require the hippocampus (Phillips and LeDoux 1992).
Elevated zeromaze. We assessed anxietylike behavior using an adaptation of a widely used method for rodents that measures time spent in the closed versus open arms of a circular maze (Shepherd et al. 1994) (see Supplemental Material, p. 6). Immediately after the test (< 5 min), we collected a blood sample (~ 100 μL) from the facial vein of each mouse (in a separate room). We determined CORT levels in these blood samples by ELISA, as described above.
Forced swim test. We assessed depressivelike behavior of the mice by measuring time spent immobile in a container of water (Castagné et al. 2010) (see Supplemental Material, p. 6).
Adult brain cytokine analysis. To determine the enduring effects of pre natal stressors on brain cytokines and their potential role in observed behavioral changes, we assessed cytokine levels in the brains of adult offspring 10 days after behavioral testing. Offspring (n = 7-9/group/sex) were anesthetized and perfused as described above. We then extracted brains and dissected them on ice into hypothalamus (HYP), prefrontal cortex (PFC), hippocampus (HIPP), and adjacent parietal cortex (PCX). We selected these regions for their known roles in the cognitive and affective behaviors that we assessed. To obtain enough total protein for analysis, we pooled the dissected regions from each animal, snap-froze them, and stored them at -80°C until processing. We performed ELISA protein analyses of IL-1β and IL-10 as described above.
Microglial isolation and gene expression analysis.We determined the cellular source of the measured cytokines in brain tissue from adult (~ PND60) behaviorally naïve offspring (n = 5/sex from 2-3 litters/group from one cohort). We again pooled dissected HYP, PFC, HIPP, and PCX tissue from each animal to obtain enough cells for later analysis. We isolated microglia by magnetic-activated cell sorting (Williamson et al. 2011), using the Neural Tissue Dissociation Kit (P), antimyelin microbeads, and anti-CD11b (an established marker for microglia) microbeads, all from Miltenyi Biotec Inc. (Auburn, CA). We then washed cells in sterile PBS and stored them at -80°C until analysis by qRT-PCR as described by Williamson et al. (2011) (see Supplemental Material, p. 6).

Data analysis.
We analyzed all data using SPSS statistical software (IBM, Armonk, NY). Because of hetero genous variance, neonatal CORT data were log-transformed. For ELISA analyses, samples that had undetectable levels of IL-1β or IL-10 were assigned a value of one-half the lowest detectable value in the assay (Thompson et al. 2012). We used three-way analysis of variance (ANOVA) (sex × DEP × NR) to analyze all data, except for the PCR data from CD11b-positive (CD11b+) and CD11b-negative (CD11b-) isolated cells, for which we used four-way ANOVA (sex × DEP × NR × cell population). We followed up inter actions with sex using separate 2-way ANOVA (DEP × NR) for males and females to identify sex-specific effects. We also followed up significant DEP × NR inter actions within each sex using post hoc comparisons [Tukey's HSD (honestly significant difference)] to identify group differences, assuming significance at p < 0.05. All reported p-values are two-tailed, except for the correlations between behavioral measures and cytokine measures from adult brains; for those we used one-tailed p-values because we had clear a priori hypotheses based on the apparent correspondence between observed group differences in the two measures. Finally, we controlled for litter effects by using multiple litters per treatment group. In addition, litter effects can be excluded because of the sex-specific effects observed in males and females from the same litter.

Neonatal outcomes and maternal behavior.
Previous research has shown that postnatal NR results in more frequent dam departures from the nest, decreased pup weights, and increased plasma CORT at PND9 (Rice et al. 2008). In our adaptation of the model, prenatal NR decreased birth weights [main effect of NR, F(1,32) = 12.09, p < 0.005] regardless of pre natal DEP exposure, but weights normalized by PND8 ( Figure 1A). Importantly, we observed no significant differences in litter size or composition due to either environmental stressor (see Supplemental Material, Table S2). In addition to the effect on birth weight, pre natal NR also increased PND1 serum CORT in male pups [main effect of NR, F(1,20) = 7.89, p < 0.05] but not in females [sex × NR inter action, F(1,35) = 2.78, p < 0.05] ( Figure 1B). However, unlike postnatal NR, pre natal NR did not affect maternal behavior during PND2-PND9, a critical period for changes in maternal care to affect offspring brain development (Avishai-Eliner et al. 2001). Specifically, we observed no significant group differences in the percentage of time dams spent on the nest ( Figure 1C) or in nursing ( Figure 1D) or licking and grooming their pups ( Figure 1E). Furthermore, pre natal stressors did not have any enduring effects on maternal anxiety-like behavior when dams were tested in the elevated zero-maze 60 days post partum ( Figure 1F).  Fetal brain cytokine analysis. The proinflammatory cytokine IL-1β was not detectable in most of the E18 brain samples, and we observed no significant group differences due to sex or pre natal stressors (Figure 2A). In contrast, the anti-inflammatory cytokine IL-10 was detectable in a greater proportion of samples, and there was a significant sex × DEP inter action [F(1,55) = 4.24, p < 0.05]. Follow-up tests revealed that males tended to down-regulate IL-10 in response to DEP exposure [trend for main effect of DEP, F(1,27) = 2.43, p = 0.1], whereas females tended to up-regulate IL-10 in response to DEP [trend for main effect of DEP, F(1,28) = 2.395, p = 0.1] ( Figure 2B).
PND30 neuroimmune gene expression. A preliminary mouse inflammatory response 84-gene PCR array performed on PND30 male brains, identified Tlr4 as the only gene that exhibited a synergistic effect of DEP and NR (data not shown). Notably, TLR4 is an innate immune receptor critical for the response to both environ mental toxicants and stress (Arbour et al. 2000;Caso et al. 2008). We replicated this result with single-analyte qRT-PCR, finding a significant DEP × NR inter action [F(1,34) = 16.79, p < 0.001] for PND30 male brains. Specifically, the DEP/NR group had significantly higher Tlr4 expression than did the DEP/control and VEH/NR groups (p < 0.05; Figure 3A). Interestingly, we observed no significant differences among PND30 females [sex × DEP × NR inter action, F(1,51) = 9.18, p < 0.005]. In addition, expression of caspase-1 (Casp1), a downstream effector molecule and key enzyme for IL-1β production (Black et al. 1988), exhibited a similar pattern to Tlr4 expression. Post hoc tests revealed that DEP/NR male brains had significantly higher expression of Casp1 than all other groups [DEP × NR interaction, F(1,37) = 14.66, p < 0.001; post hoc, p < 0.01], whereas there were no significant differences among females [sex × DEP × NR inter action, F(1,54) = 6.45, p < 0.05] ( Figure 3B).
Memory. When assessed for contextual fear recall, DEP/NR males froze significantly less than did all other groups in the fear context [DEP × NR inter action, F(1,29) = 5.15, p < 0.05; post hoc, p < 0.05]. However, DEP/ NR females exhibited no such hippocampal-dependent memory impairment [trend for sex × DEP × NR interaction, F(1,57) = 2.23, p = 0.1] ( Figure 4A). Importantly, this male-specific deficit was not due to generalized hyperactivity or an inability of the DEP/NR males to freeze, because there were no significant differences in freezing in a novel context ("new context") or freezing to the auditory cue (CS) ( Figure 4A). Furthermore, the females' stage of the estrous cycle did not significantly affect their behavior in any of the tests (data not shown).

Anxiety-and depressive-like behavior.
In the elevated zero-maze, both adult male and female DEP/NR offspring spent more time in the closed arms, indicative of increased anxiety [DEP × NR inter action with sexes combined, F(1,55) = 7.350, p < 0.01] ( Figure 4B). Post hoc tests revealed that DEP/ NR males were significantly more anxious than were VEH/control and DEP/ control males (p < 0.05). In addition, there was a trend for DEP/NR males to be more anxious than VEH/NR males (p = 0.07), which did not significantly differ from VEH/control males. DEP/ NR females were more anxious than DEP/ control females (p < 0.05), whereas VEH/control and VEH/NR females did not differ from each other. However, this increase in anxiety in the DEP/ NR animals was not associated with increased serum CORT because NR males exhibited a slight decrease in CORT immediately after the test, regardless of pre natal DEP exposure [main effect of NR, F(1,20) = 5.286, p < 0.05], and females displayed no significant differences Figure 2. Effects of prenatal DEP and NR on the cytokine response in fetal brain. (A) No significant differences in IL-1β protein levels were detected between treatment groups in the E18 brain. (B) A sexually dimorphic IL-10 response was observed in the brains of fetuses prenatally exposed to DEP: Male brains exhibited a down-regulation of IL-10, whereas female brains exhibited up-regulation. Data presented are mean ± SE of 7-8 animals/group.  Figure S1A). Finally, neither males nor females displayed any significant group differences in the forced swim test (see Supplemental Material, Figure S1B). Adult brain cytokine analysis. Ten days after behavioral testing, male brains from DEP groups exhibited increased levels of IL-1β protein [main effect of DEP, F(1,22) = 7.84, p < 0.01], whereas there were no group differences among female brains [trend for sex × NR inter action, F(1,45) = 2.49, p = 0.1] ( Figure 5A). In contrast, maternal DEP exposure and NR resulted in decreased levels of IL-10 protein, in an additive fashion, in male brains [main effect of DEP, F(1,22) = 12.13, p < 0.005; main effect of NR, F(1,22) = 6.89, p < 0.05], whereas female brains again did not exhibit any signifi cant group differences due to pre natal stressors [sex × DEP interaction, F(1,45) = 7.10, p < 0.05] ( Figure 5B). Overall, DEP/NR males exhibited a greater pro inflammatory bias (IL-1β/IL-10 ratio) (de Wit et al. 2010) than did DEP/NR females [sex × DEP inter action, F(1,39) = 4.26, p < 0.05; sex × NR inter action, F(1,39) = 4.53, p < 0.05; post hoc, p = 0.08] ( Figure 5C).
We found it striking that the group differences in brain cytokine measures seemed to parallel the behavioral differences we had observed in the same animals, and indeed, further analyses revealed that the brain levels of IL-1β and IL-10 correlated significantly with the behavioral measures of memory and anxiety, although in a divergent manner in males and females. Males exhibited a significant negative correlation between IL-1β and contextual fear memory [r(26) = -0.33, p < 0.05], such that higher levels of pro inflammatory IL-1β were associated with decreased freezing in the fear context, whereas females showed a trend for a positive correlation [r(27) = 0.27, p = 0.09] ( Figure 5D). IL-1β was also positively correlated with anxiety-like behavior in males [r(24) = 0.37, p < 0.05], such that higher levels of IL-1β were associated with increased time spent in the closed arms of the elevated zero-maze, whereas in females IL-1β was negatively correlated with anxiety-like behavior [r(26) = -0.46, p < 0.01] ( Figure 5E). On the other hand, anti-inflammatory IL-10 exhibited no significant correlations with memory in males or females ( Figure 5F), but it was negatively correlated with anxiety-like behavior in males [r(25) = -0.42, p < 0.05] and not females ( Figure 5G).

Gene expression of isolated CD11b+ and CD11b-cells.
To assess the purity of the isolated cells, we confirmed that Cd11b mRNA was expressed only in CD11b+ cells, as expected [main effect of cell population, F(1,64) = 71.68, p < 0.001] (see Supplemental Material, Figure S2A). In contrast, CD11bcells expressed approximately 20-fold higher levels of Bdnf (brain-derived neurotrophic factor) mRNA and Gfap (glial fibrillary acidic protein) mRNA than did CD11b+ cells (all p-values < 0.001; see Supplemental Material, Figure S2B and C, respectively), indicating that this cell population (~ 80% of total cells) contains astrocytes and neurons. Isolated CD11b+ cells (microglia) had approximately 170-fold higher levels of Il1β mRNA and approxi mately 80-fold higher levels of Il10 mRNA than did CD11bcells (all p-values < 0.001; see Supplemental Material, Figure S2D,E). In addition, CD11b+ cells expressed approximately 40-fold more Casp1 and approximately 20-fold more Tlr4 than CD11b-cells (all p-values < 0.001; see Supplemental Material, Figure S2G and F, respectively). However, we observed no significant group differences in gene expression due to sex or pre natal stressors.

Discussion
We observed that maternal stress during late pregnancy exacerbated the impact of in utero exposure to DEP on mental health outcomes in adult offspring-outcomes that are associated with alterations in neuro inflammatory tone (i.e., increases in IL-1b and decreases in IL-10). The adult DEP/NR males (prenatally exposed to both stressors) exhibited significant memory deficits and anxiety, whereas DEP/ NR females displayed only slightly increased anxiety. Importantly, the impact of pre natal events on offspring behavior is always complex and may involve changes in several physiological pathways, as well as potential alterations in maternal-offspring inter actions after birth-both of which can profoundly modify neural develop ment (Caldji et al. 1998). In the present study, NR males (exposed to maternal stress) had reduced birth weight and increased CORT on PND1, independent of maternal DEP exposure. However, maternal care was not significantly altered by resource deprivation, and we observed no changes in maternal anxiety as a consequence of treatment. We observed no differences in corticosterone in adult offspring that could explain behavioral changes in response to the combined stressors. Instead, the altera tions in brain cytokines, which occurred in a sexually dimorphic manner, may underlie distinct behavioral phenotypes in adult male and female offspring. An optimal balance of central pro inflammatory cytokines, such as IL-1β, is critical for mental health, including mood regulation and hippocampal-dependent learning and memory (Yirmiya and Goshen 2011). In contrast, high levels of anti-inflammatory IL-10 are protective against behavioral changes volume 121 | number 9 | September 2013 • Environmental Health Perspectives due to microglial-driven neuro inflammation (Schwarz et al. 2011). Therefore, the diminished IL-10, in combination with increased IL-1β, that we observed in DEP/NR males could affect their relative vulnerability to cognitive impairments and mood dys regulation, compared with DEP/ NR females, which did not exhibit such a pro inflammatory bias. The underlying mechanism of this sex difference warrants further exploration. Importantly, there is a male bias in the prevalence of neurodevelopmental disorders, including learning disabilities (Flannery et al. 2000) and autism (Stone et al. 2004), in addition to differences between the sexes in childhood outcomes following maternal stress during pregnancy (Cao et al. 2012;Fang et al. 2011).
Our data from isolated CD11b+ and CD11b-cells demonstrate that microglianot neurons or astrocytes-are the primary source of the measured cytokines in the brain, suggesting that they are a target of "programming" by the pre natal stressors. Microglia begin to colonize the rodent brain around E9-E10 (Ginhoux et al. 2010) and critically shape several aspects of normal brain develop ment. Importantly, microglia largely remain in an activated, amoeboid state until the early post natal period (Ling and Wong 2004), which makes them especially sensitive to long-term functional changes by perinatal inflammatory events (Bilbo and Schwarz 2009;Williamson et al. 2011). However, we did not detect any signifi cant differences in cytokine gene expression in isolated adult micro glia due to pre natal stressors, despite the clear differences in cytokine protein. This discrepancy may stem from the fact that the animals we used for protein analysis underwent behavioral testing, whereas the animals used for CD11b isolation and gene expression analysis were behaviorally naïve. It is possible that behavioral testing may serve as a sufficient stressor to elicit relatively longterm increases in cytokine levels (i.e., enduring until tissue collection) in the brains of the DEP/NR animals, which would not be observed at baseline. Alternatively, there may be additional regulatory mechanisms at work. For instance, the selective increase in caspase-1 expression in DEP/NR males supports a role for the inflammasome, which is critical for the cleavage of proIL-1β into its mature form (Latz 2010), although we did not examine this directly. Importantly, TLR4 signaling is required for the activation of the inflammasome (Bauernfeind et al. 2009). Furthermore, DEP and maternal stress-induced signals may converge on microglia via TLR4, which was predominantly expressed by microglia, consistent with our previous findings in rats (Schwarz and Bilbo 2013), and is exaggerated in the brains of DEP/NR males. TLR4 is an important innate immune receptor that recognizes pathogen-associated molecular patterns (e.g., lipo poly saccharide), as well as endogenous danger-associated molecular patterns released in response to cellular distress (e.g., DEP-induced hyaluronan or high-mobility group box 1) (Bianchi 2007). Notably, gluco corticoids may up-regulate TLRs on microglia, augmenting subsequent Figure 5. Effects of prenatal DEP and NR on cytokine levels in the adult brain. (A) DEP-exposed males exhibited a significant increase in brain IL-1β, but DEPexposed females did not. (B) Males exhibited a significant decrease in brain IL-10 concentration in response to both DEP and NR, but no significant changes were observed in females. (C) Overall, DEP/NR males exhibited a greater pro inflammatory bias (IL-1β/IL-10 ratio) than DEP/NR females. Data for A-C are presented as the mean ± SE of 5-8 animals/group. (D-E) Brain IL-1β concentration was significantly correlated with memory (D) and anxiety (E) measures in both males and females, although in opposite directions. (F) Brain IL-10 concentration was not correlated with memory performance in either males or females. (G) Brain IL-10 concentration was negatively correlated with anxiety-like behavior in males but not in females. Data points in D-G represent correlated measures for individual animals from one entire cohort (n = 24-27 total animals per sex). # p < 0.05 DEP compared with VEH groups. **p < 0.05 NR compared with both control groups. ## p = 0.08, DEP/NR males compared with DEP/NR females; significant sex × DEP and sex × NR interactions (p < 0.05). neuro inflammatory responses (Frank et al. 2010;Garate et al. 2013). Thus, Tlr4 up-regulation may occur to a greater extent in males than in females as a result of the significant increase in serum cortico sterone in males born to stressed dams. Although the alterations in IL-1β we observed in adult offspring were not present in the fetal brain, we did observe a significant sex difference in IL-10 in the E18 brains of DEPexposed pups. Specifically, although DEP males exhibited a down-regulation of IL-10, DEP females displayed an up-regulation of this anti-inflammatory cytokine, suggesting that IL-10 may have been protective against the neurodevelopmental effects of pre natal stressors. Along with brain cyto kines, the placenta has been increasingly implicated as a key player in brain develop ment, as well as in the fetal response to pre natal insults (Hsiao and Patterson 2012), and warrants further exploration in our model. Thus, our working hypothesis is that maternal stress-induced changes in TLR4 signaling enhance the effects of a chemical exposure such as DEP, likely involving the maternal-placental-fetal interface (Auten et al. 2009, and ultimately converging onto micro glia within the fetal brain, resulting in the long-term alteration of brain function and behavior.

Conclusions
We observed that maternal psychological stress induced by resource deprivation during late pregnancy increased the vulnerability of murine offspring, particularly males, to in utero air pollutant exposure. Furthermore, developmental exposure to maternal stress and air pollution, similar to other environmental compounds such as pesticides and lipo poly saccharide , may have a long-lasting impact on micro glial function and neuro inflammation. Future studies aimed at elucidating the complex inter actions of psycho social and chemical stressors will be critical for informing environ mental and public health policy and identifying effective interventions.