The Role of Redox Dysregulation in the Effects of Prenatal Stress on Embryonic Interneuron Migration

Abstract

Maternal stress during pregnancy is associated with increased risk of psychiatric disorders in offspring, but embryonic brain mechanisms disrupted by prenatal stress are not fully understood. Our lab has shown that prenatal stress delays inhibitory neural progenitor migration. Here, we investigated redox dysregulation as a mechanism for embryonic cortical interneuron migration delay, utilizing direct manipulation of pro- and antioxidants and a mouse model of maternal repetitive restraint stress starting on embryonic day 12. Time-lapse, live-imaging of migrating GAD67GFP+ interneurons showed that normal tangential migration of inhibitory progenitor cells was disrupted by the pro-oxidant, hydrogen peroxide. Interneuron migration was also delayed by in utero intracerebroventricular rotenone. Prenatal stress altered glutathione levels and induced changes in activity of antioxidant enzymes and expression of redox-related genes in the embryonic forebrain. Assessment of dihydroethidium (DHE) fluorescence after prenatal stress in ganglionic eminence (GE), the source of migrating interneurons, showed increased levels of DHE oxidation. Maternal antioxidants (N-acetylcysteine and astaxanthin) normalized DHE oxidation levels in GE and ameliorated the migration delay caused by prenatal stress. Through convergent redox manipula-tions, delayed interneuron migration after prenatal stress was found to critically involve redox dysregulation. Redox biology during prenatal periods may be a target for protecting brain development.

Maternal stress experienced during pregnancy is associated with an increased risk for neuropsychiatric disorders in her offspring, such as schizophrenia (King and Laplante 2005; Bale 2009) and attention-deficit hyperactivity disorder (Huot et al. 2004). Stress alters multiple aspects of physiology (McEwen 2004; Sousa and Almeida 2012) that, when occurring in a pregnant mother, may pass to the fetus and alter the developing brain. These critical fetal changes may be mechanisms by which the risk for psychiatric disorders during adolescence and adulthood is increased many years later (Markham and Koenig 2011). The cellular and molecular alterations in the fetal brain that occur during prenatal stress are therefore a significant part of understanding psychopathology.

There are important relationships between the inhibitory neural circuitry of the forebrain and mental illness (Benes and Berretta 2001; Hashimoto et al. 2008; Yip et al. 2008). Disruption of GABAergic forebrain systems during embryonic development can have significant, enduring consequences for neural function in adolescence and adulthood (Powell et al. 2003; Levitt et al. 2004; Fu et al. 2012). Inhibitory cortical interneurons are of particular significance because of their role in cognition (Fishell and Rudy 2011) and links between atypical GABAergic system functioning and the psychopathology of schizophrenia, autism, and anxiety disorders (for a comprehensive review see (Fine et al. 2014)).

The effects of prenatal stress on the embryonic brain have revealed potential mechanisms through which the functioning of cortical GABAergic systems are affected by early environmental stressors. From their birthplace in the ganglionic eminence, GABAergic progenitor cells tangentially migrate into the developing neocortex, where they establish inhibitory neural connections to form functional groups such as cortical columns and limbic–cortical loops. Transcription factors that regulate GABAergic interneuron migration are significantly changed by prenatal stress (Stevens et al. 2013) and maternal immune activation (Oskvig et al. 2012). Prenatal stress and the maternal immune factor, IL-6, are also known to decrease overall migration of interneuron progenitors (Gumusoglu et al. 2017) and decrease the density of GABAergic progenitors in the embryonic cortical plate (Stevens et al. 2013). However, up-stream mechanisms by which prenatal stress leads to GABAergic progenitor changes have yet to be elucidated.

There are a number of mechanisms that direct interneuron tangential migration. Extracellular cues are crucial to migration, including motogens, chemoattractants, and neurotransmitters (Corbin and Butt 2011). Cell autonomous components of interneurons such as CXCR4, the chemokine receptor for the chemoattractant, SDF-1, also guide interneurons within the tangential migratory streams (Sanchez-Alcaniz et al. 2011). Genetic and syndromic risk for psychiatric illness influences these mediators and interneuron development (Meechan et al. 2012; Abbah and Juliano 2014). The role of maternal physiological disruptions such as prenatal stress in affecting these mechanisms has not been explored.

GABAergic interneuron migration is also susceptible to changes to oxidative phosphorylation (Lin-Hendel et al. 2016). Both genetic disruption of mitochondrial oxidative phosphorylation in mice lacking adenine nucleotide transferase 1 (Ant1) and pharmacological inhibition of oxidative phosphorylation alter interneuron migratory morphology, direction, and migratory rates (Lin-Hendel et al. 2016). Mitochondrial disruption also significantly contributes to the generation of reactive oxygen species (ROS) (Burton and Jauniaux 2011), adding to the reactive forms of oxygen produced during normal metabolism (Guerin et al. 2003) from the one electron reductions of O₂ from mitochondrial electron transport chains. The developing embryonic brain is particularly vulnerable to ROS changes and redox dysregulation because of its low antioxidant capacity (Wells and Winn 1996; Wells et al. 1997, 2005), particularly migrating neuronal progenitors (Narasimhaiah et al. 2005). Therefore, ROS and redox balance may play a role in disruptions of embryonic brain development.

There are multiple links between the development of inhibitory neural systems, in particular, and redox regulation (Marín 2012; Hsieh et al. 2017). Redox dysregulation results from disruptions to the balance of reductive and oxidative molecules of cellular metabolism. Sizable, sustained redox dysregulation may occur when ROS overwhelm endogenous antioxidants (or result from impaired antioxidants) and may contribute to multiple disease states (Peuchant et al. 2004; Cambonie et al. 2007; Derks et al. 2010; Ghulmiyyah et al. 2011; Ziech et al. 2011; Pinney and Simmons 2012). Neuropsychiatric disorders have been specifically linked to redox dysregulation, including schizophrenia (reviewed in (Do et al. 2009)), epilepsy (Chevallier et al. 2014), and autism spectrum disorder (ASD) (Rossignol and Frye 2012). Inhibitory circuits are implicated in these disorders (Marín 2012) and are particularly sensitive to redox dysregulation (Do et al. 2009).

Prenatal stress increases lipid peroxidation and decreases total antioxidant capacity superoxide dismutase (SOD) activity, reduced glutathione (GSH) levels, and glutathione reductase (GR) activity in postnatal, juvenile offspring brain (Bernhardt et al. 2017). Postnatal treatment with N-acetylcysteine (NAC), a glutathione precursor and dietary antioxidant, reverses brain redox changes after prenatal stress in adult mice (Bernhardt et al. 2017) and reverses effects of prenatal stress on embryonic microglia (Bittle and Stevens 2018). Maternal immune activation also induces oxidative stress in adolescent and adult brain of mouse offspring (Al-Amin, Alam, et al. 2016; Al-Amin, Reza, et al. 2016). However, shifts in redox balance in the embryonic brain during maternal stress have not been identified.

In the present study, we examined the role of redox dysregulation in interneuron migration. To elucidate mechanisms by which prenatal stress alters GABAergic progenitor migration patterns, we also set out to evaluate the redox state of the embryonic brain in a mouse model of prenatal stress. To probe the role of ROS and redox balance in effects of prenatal stress, we manipulated redox states with antioxidants and pro-oxidants and evaluated embryonic brain outcomes.

Materials and Methods

Mice

GAD67-GFP+/− knock-in mice were bred on a CD1 background and housed in accordance with the University of Iowa Institutional Animal Care and Use Committee (IACUC) policies. All mice were housed in cages on a 12 h light/dark cycle with free access to food and water. For timed pregnancies, breeding-naïve GAD67-GFP−/− female mice were bred with GAD67-GFP+/− males. The detection of a vaginal plug established embryonic day 0 (E0). Prenatally stressed dams were singly housed from E12 onward and nonstressed dams were cohoused.

Treatments and Prenatal Stress

Half of pregnant females underwent prenatal stress in a clear, plastic restraint under bright lights for 45 min, 3 times a day, at 3–4 h intervals, starting on E12 and continuing daily until tissue collection (Stevens et al. 2013; Lussier and Stevens 2016). On E13 or E14, prenatally stressed dams underwent one additional restraint session prior to embryo collection. In another cohort of dams, beginning on E12, prenatally and nonstressed dams were injected intraperitoneally (ip) with either: 1) N-acetylcysteine (NAC) (200 mg/kg of 40 mg/mL saline solution with 30% sodium hydroxide (1 μM) to bring pH to 7.4; Sigma A7250) (Ho et al. 1999; Bielefeld et al. 2007; Davis et al. 2010; Flurkey et al. 2010; Azuelos et al. 2015), 2) astaxanthin (AST; 30 mg/kg [first daily injection] or 10 mg/kg [second and third daily injections] in a 10 mg/mL saline solution; with 3% dimethyl sulfoxide [DMSO]; Sigma SML0982) (Suzuki et al. 2006; Lee and Lee 2011), or 3) saline (200 μL, 0.9% saline), 20 min prior to restraint stress sessions or at equivalent times in nonstressed dams. All pregnant female mice were euthanatized by ketamine/xylazine anesthesia followed by rapid decapitation on E13 or E14.

Live Imaging

At E14, GAD67GFP+ cells were visualized in live brain slices as they migrated tangentially out of the ventral forebrain in the developing neocortex. Embryonic mouse brains from nonstressed dams were dissected and sectioned with a vibrating microtome (350–450 μm thick) in cold artificial cerebral spinal fluid (aCSF) to create at least 2 coronal sections of forebrain. Embryos from 5 independent litters were used to measure migration of GABAergic interneuron progenitors, using multiple embryos of each litter, with 2 sections from each embryo matched—one in control and one in experimental media: 1) standard aCSF as the control media (5% fetal bovine serum in Hanks’ solution (Life Technologies, USA), supplemented with d-glucose (6 mg/mL) and streptomycin (Sigma, 20 mg/L) and 2) aCSF containing the pro-oxidant, hydrogen peroxide (1 and 5 μM concentrations) (Feeney et al. 2008). Tissues were mounted between a glass coverslip and flexible mesh within a plastic chamber and left at room temperature for 1 h. Time-lapse imaging was then performed of migrating interneurons using a Leica SP5 scanning laser confocal microscope equipped with automated image capture (Dailey et al. 2013) and a stage with a hot air supply to maintain a temperature of 36-38 °C. In each section, 15–18 images in a Z-stack (3 μm z-step interval) were taken at ×20 magnification and across 45–54 μm every 5–10 min over 1–2 h focused on the lateral, antihem region of the cortical plate. Images from matched sections were made with the same parameters used, and all analyses were done with paired statistical approaches. Images were analyzed using ImageJ to assess the direction and rate of movement of individual GABAergic progenitor cells in the superficial and deep migratory streams. Results were based on the average measurements of 10 randomly chosen migrating cells in each of 5 coronal tissue slices in each control and experimental group. Random selection of 10 points was made blindly on an outline of the section which was then overlaid on the time-lapse video image and the GFP+ progenitor cell closest to each of the 10 points were chosen for analysis. Rate of movement was calculated by dividing the path length in the x and y plane by the time period of recording. Angle of movement was calculated by calculating the angle between the vector from start to end of measured movement to the vector of slice tangent.

Embryonic Intracerebroventricular Injection

Pregnant dams with E13 embryos were anesthetized with 3% isoflurane vaporizer and a midline incision, approximately 3 cm in length, was made along the abdomen. The uterus was extracted from the abdominal cavity with DPBS. Either rotenone (0.031 μg/μL in 0.9% NaCl, Enzo Life Sciences), an inhibitor of mitochondrial respiratory chain complex I (Palmer et al. 1968), or 0.9% NaCl was injected into one of the lateral ventricles of the embryonic forebrain using a glass micropipette. Evan’s Blue dye (1% MP Biomedicals) was included in the injection to ensure successful delivery into the embryonic lateral ventricle. A total of 1 μL of rotenone solution or NaCl vehicle was injected into each embryo. After ICV injection of all embryos, the uterus was returned to the abdominal cavity and the abdominal wall and skin were closed with sutures. On E14, 24 h later, pregnant dams were sacrificed and whole embryonic heads were collected for measurement of GABAergic progenitor migration as described below.

Gene Expression

To examine the source of migrating GABAergic progenitors, ventral forebrain tissue was dissected from E13 embryos and flash frozen on dry ice. Quantitative PCR was performed to assess changes in the expression of redox related genes, as well as genes involved in interneuron development: thioredoxin reductase 1 (Tr1; F: 5′-GACCAGGGAAACCAAGGGAG-3′ R: 5′-CACGCGTGTGCATCAACATC-3′), glutathione peroxidase 1 (Gpx1; F: 5′-AGTGCGAAGTGAATGGTGAGA-3′ R: 5′-GCACACCGGAGACCAAATGA-3′), sestrin 1 (Ses1; F: 5′-GGCCAGGACGAGGAACTTG-3′ R: 5′-AAGGAGTCTGCAAATAACGCAT-3′), sestrin 2 (Ses2; F: 5′-GGATTATACCTGGGAAGACC-3′ R: 5′-CGCAGTGGATGTAGTTCC-3′), sestrin 3 (Ses3; F: 5′-CCAGGACTACACCTGGGAAA-3′ R: 5′-AACCTTCAGGCTCCGTTCAA-3′), catalase (Cat; F: 5′-CGCAATTCACACCTACACGC-3′ R: 5′-TTTCCCTTCAGGAAACG GCA-3′), superoxide dismutase 1 (Sod1; F: 5′-CAGGACCTCAT TTTAATCCTCAC-3′ R: 5′-CCCAGGTCTCCAACATGC-3′), nuclear factor erythroid-derived 2 (Nrf2; F: 5′-CCACATTTCCTTCATGGTTTTG-3′ R: 5′-GACACTTCCAGGGGCACTATCT-3′), Apoptosis-inducing factor 1, mitochondrial (Aifm1; F: 5′-CGCTAAGCCATACTGGCATCA-3′ R: 5′-CAACTGTGGGCAAACTACTATCCA-3′), isocitrate dehydrogenase [NADP], mitochondrial (Idh2; F: 5′-TGGCTGGCTGTATCCATGG-3′ R: 5′-GGAAGTGCTCGTTCAGCTTC-3′), NADH dehydrogenase [ubiquinone] iron-sulfur protein 4, mitochondrial (Ndufs4; F: 5′-GAGCACATCCACTTGGAAGC-3′ R: 5′-GATGTGCTCTTCTGGAA CACC-3′), and C-X-C chemokine receptor type 4 (Cxcr4; F: 5′-ATGGAACCGATCAGTGTGAGT-3′ R: 5′-TGAAGTAGATGGTGGGCAGG-3′). RNeasy Mini Kit (Qiagen, Valencia, CA) was used per the manufacturer’s protocol to extract mRNA from the tissue. Total RNA concentrations were quantified on a spectrophotometer (Nanodrop, Thermo Scientific, USA) and reverse transcribed using Transcriptor First-Stand cDNA Synthesis Kit (Roche, USA). qPCR with Power SYBR Green Master Mix (Thermo Fisher Scientific, Warrington, UK) was performed using Applied Biosystems Model 7900HT instrument. The endogenous gene, glyceraldehyde-3-phosphate dehydrogenase (Gapdh; F: 5′-GGTGAAGGTCGGTGTGAACG-3′ R: 5′-CTCGCTCCTGGAAGATG GTG-3′), was used as a control gene and relative expression was calculated as: 2^{−[Ct[gene of interest]–Ct(GAPDH)]}, where Ct is threshold cycle. We found no significant differences due to prenatal stress or NAC exposure in Gapdh expression.

Glutathione Assay

The ratio of reduced glutathione (GSH) to glutathione disulfide (GSSG) is a common indicator of redox status and a gauge of antioxidant efficiency. Total GSH and GSGG content was determined as previously described (Tietze 1969; Griffith 1980), respectively, using spectrophotometry. Briefly, the assay was performed in 2 steps: first, the total amount of glutathione (GSH plus GSSG) was calculated, and second, the amount of GSSG alone was determined. GSH was calculated from the difference between these 2 direct measures. GSSG was determined by adding 2-vinylpryidine, which binds to GSH and prevents GSH from reacting in the recycling assay. Final concentrations for each reagent in the assay were as follows: 114 mM sodium phosphate, 5.04 mM EDTA, 0.209 mM NADPH, 0.600 mM DTNB, and 0.50 units/mL GSSG reductase. Total forebrain tissue from E13 embryos was pooled (2–3 forebrains per sample) and homogenized in 5-sulfosalicylic acid. The BCA Protein Assay Kit (Thermo Scientific, Rockford, IL) was used to determine protein content and for normalization of samples.

Antioxidant Activity

The following antioxidant enzyme activity assays were performed on E13 total forebrain. Upon collection, forebrains were homogenized in 50 mM potassium phosphate buffer, pH 7.8, with 1.34 mM diethylenetriaminepentaacetic acid (DETAPAC). Glutathione peroxidase 1 (GPx1), an antioxidant enzyme that enzymatically reduces hydrogen peroxide and organic hydroperoxides to water to limit its harmful effects (Carter et al. 2004), was measured spectrophotometrically. Enzymatic activity of GPx1 was determined spectrophotometrically using the method outlined in (Lawrence and Burk 1976) using the final concentrations: 50 mM potassium phosphate, 1 mM EDTA, 1 mM NaN₃, 0.2 mM NADPH, 1 E.U./mL GSSG-reductase, 1 mM GSH, 0.25 mM H₂O₂, and 1.5 mM cumene hydroperoxide. Changes in absorbance were observed with a Beckman DU-650 spectrophotometer at 340 nm continuously for 2.5 min. One unit of activity was defined as the amount of protein that oxidizes 1 μM of NADPH per minute. Protein concentrations were determined by the Lowry assay.

An essential enzyme in the oxidoreductase system, thioredoxin reductase 1 (TR1), helps combat ROS and was detected spectrophotometrically with 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) as substrate using 5 mM DTNB and 0.2 mM NADPH in 0.1 M potassium phosphate, 1 mM EDTA, pH 7.0 plus 0.1 mg/mL bovine serum albumin. Because other antioxidant enzymes in biological samples, such as GR, can reduce DTNB, a thioredoxin reductase specific inhibitor was used in a separate reaction to determine TR1 activity. The difference between total DTNB reduction in the sample and DTNM reduction in the sample in the presence of the thioredoxin reductase specific inhibitor was determined. The calculated difference was the value of specific thioredoxin reductase activity in the sample. Absorbance was read at 412 nm for 6 s (as previously described by Holmgren and Bjornstedt 1995).

Dihydroethidium

The method of dihydroethidium (DHE) oxidation detection used here is a modified version of the method describe in (Mapuskar et al. 2017). Embryonic brain tissue was examined for DHE fluorescence which occurs when undergoing oxidation, an indicator of superoxide (Kimura et al. 2010) and other pro-oxidants in the cell. When DHE comes into contact with the superoxide anion and other pro-oxidants in tissue, several wavelengths of red fluorescence are detectable from the resultant ethidium and 2-hydroxyethidium that arise from oxidation. Antimycin-A, a known inhibitor of Complex III of the mitochondrial electron transport chain, was used as a positive control because of its ability to increase one electron reduction of O₂ to form one substantial contributor to DHE-oxidation, superoxide. E13 embryo head tissue was embedded fresh in Optimal Cutting Temperature (OCT) compound and coronally cryosectioned (Leica, 1510-3, Bannockburn, Illinois) at 10μm. Sectioned tissue was immediately frozen on slides. As a positive control, 10 μM antimycin-A and 10 μM DHE (Invitrogen) in 0.1% DMSO was applied to one tissue sample and incubated at 37 °C, 5% CO₂ for 10 min (Coleman et al. 2014). After 10 min, slides were gently rinsed with PBS (containing 5 mM pyruvate), coverslipped, and immediately imaged with an epifluorescent Zeiss Axiolmager M2 microscope with a filter cube for red detection (587/25 excitation/649/70 emission). Note: these filter parameters did not ideally overlap the spectra of ethidium and 2-hydroxyethidium (Zielonka et al. 2008) but had enough overlap to capture meaningful signal (Thermo Fisher Spectra Viewer) with a high-sensitivity monochrome digital camera, validated with negative (no DHE) and positive controls (antimycin A). Images were obtained with Orca-R² digital camera (Hamamatsu, Japan) and Stereoinvestigator software for digital imaging (Microbrightfield, Vermont). A 10 μM DHE was applied to a second sample, the tissue incubated for 10 min, and then rinsed with PBS before coverslipping and imaging. Gain intensity and exposure time settings for digital imaging were established using control tissue with antimycin-A as a maximum intensity measurement and maintained while imaging control embryonic brain (with and without antimycin-A) immediately followed by embryonic brain from a comparison experimental group. Images of DHE+ cells were obtained using the ×40 objective from 3 different places in the ganglionic eminence and were analyzed for fluorescence with ImageJ software. Within each brain, mean fluorescent intensity (MFI) levels from a total of 300 cells within the ganglionic eminence, the origin of GABAergic progenitors, were averaged. The average level from each experimental condition sample was normalized to average level in the matched nonstress control processed at the same time and compared across groups.

GABAergic Progenitor Migration

Pregnant dams were sacrificed and whole embryonic heads were collected and fixed in 4% paraformaldehyde (PFA) overnight and incubated in 20% sucrose for at least 20 h. Tissue was embedded in OCT, cryosectioned at 25 μm, and then incubated in 10% goat serum/PBS++ blocking solution (with 0.025% Triton X-100, 0.0125% Tween 20) at room temperature for at least 1 h. They were then incubated overnight with 5% goat serum/PBS++ and primary antibody, anti-GFP (1:1000, Abcam, AB13970). Sections were stained with the Alexa dye-conjugated secondary antibody (1:500–1000; Molecular Probes) in 5% goat serum/PBS++ incubation for 2 h. Tissue section slides were then coverslipped using DAPI mounting medium (4′,6-diamidino-2-phenylindole, Vector Laboratories, #H-1200). Migration of GAD67GFP+ cells in coronal tissue sections was measured using fluorescent microscopy with a Zeiss Axiolmager M2 microscope with Stereoinvestigator software. We measured the circumferential length of the superficial stream of GAD67GFP+ cells along the edge of the dorsal forebrain as a percentage of the entire circumferential length of the cortical plate and averaged across at least 4 anterior to posterior sections. compared between antioxidant, stress, pro-oxidant, and control groups.

Data Analysis and Sampling

Unless otherwise noted, all experiments were conducted with equal sample sizes of male and female embryos and sex genotyping was performed using PCR. For each assay, male and female outcomes were first compared separately for sex differences. If no sex differences were found, samples were pooled and sex-balanced sample sizes were reported.

In evaluating the velocity and angle deviation of migrating GAD67+ cells in the in vitro, live imaging experiment, 2-tailed paired t-tests were used to detect differences between cells in the hydrogen peroxide aCSF and cells in the control aCSF. For the ICV rotenone experiments, Mann–Whitney U test was performed to compare MFI levels of rotenone-exposed embryonic brains to saline-exposed embryonic brains and paired t-tests were used to compare the migration rates of the 2 conditions. To examine the effects of prenatal stress on gene expression, unpaired t-tests correcting for multiple comparisons using the Bonferroni method were performed. For the DHE experiments, 2 separate Kruskal–Wallis tests were performed to detect significant differences in DHE MFI between the prenatally stressed and nonstressed embryos of the PBS and NAC groups and PBS and AST groups. Dunn’s tests were performed to evaluate group differences correcting for multiple comparisons.

First, a priori t-tests were used to determine if prenatal stress affected outcomes. Then, 2-way ANOVAs were used to evaluate the effects of prenatal stress and each antioxidant manipulation independently for each assay, looking for interactions that would indicate the ability of antioxidant treatment manipulation to significantly modify the main effect of prenatal stress. The GSH and GSSG tests, the GPx1 activity assay, the E13 and E14 progenitor migration experiments, and the Cxcr4 gene expression assay were all analyzed using a priori t-tests and then 2-way ANOVAs with follow-up post hoc tests when appropriate.

Results

GABAergic Progenitor Cell Migration is Sensitive to Redox Dysregulation

GAD67GFP+ cell movements were captured in vitro during migration into the developing neocortex of E14 embryo forebrain sections using live imaging (Fig. 1A–C). Matched forebrain sections showed that normal tangential interneuron migration (Fig. 1D–G) was disrupted by hydrogen peroxide (H₂O₂; Fig. 1H–K; equivalent results for 1 and 5 μM H₂O₂, so data were pooled). H₂O₂ caused GAD67GFP+ cells to deviate in their direction of migration compared with control. The angle of deviation of cell movement in control tissue measured from the direction of the tangent ranged from 0° to 170°, with the average cell movement in single sections ranging 20.5°–53.0° (Fig. 1D–G, M). The angle of deviation in H₂O₂-exposed interneurons ranged from 0° to 180°, with the average cell movement in single sections ranging 54.0°–96.5° (P < 0.01 from control; Fig. 1H–K, M). While the direction of cell movement was divergent with H₂O₂, the velocity of migrating GAD67GFP+ cells was also increased, from an average velocity across all samples of 41.7 to 72.2 μm/h—a 71.2% increase (P < 0.01; Fig. 1L). Cellular velocity ranged from 6.3 to 84.0 μm/h in control conditions and 29.4–122.4 μm/h with H₂O₂ exposure. Together, a model of these altered movements shows an overall delay in the population migration in the appropriate direction (Fig. 1N).

Figure 1.

Time-lapse imaging of migrating interneurons in acutely excised forebrain tissue slices. (A–C) Arrow indicates direction of migrating GABAergic progenitor cells and arrowheads indicate the starting position and end position of each interneuron after 30 min. (A) Tissue section of E14 embryo in artificial cerebral spinal fluid (aCSF) at beginning time point (t = 0 min). (B) Tissue section pictured in A at t = 30 min. (C) Magnification of A and B with mash-up of time points 0 and 30 min. Schematic of migration of 10 interneuron progenitors after approximately 1–1.5 h of E14 embryo in aCSF (D–G) and aCSF with either 1 μM H₂O₂ (H–J) or 5 μM H₂O₂ (K). Arrow indicates direction of migration stream and dot is end-point. (L) Bar graph shows increased average velocity of GAD67GFP+ cells and (M) depicts increased average degree of deviation from migration stream with H₂O₂ exposure. (N) Schematic of hypothesized migration delay showing average velocity and average angle deviation for each of the 5 live imaging sessions. For the purpose of calculating migration deviation, we designated interneurons confined to migration stream as deviating 0°. Dashed lines represent the boundaries of the tissue slices. (**P < 0.01 by 2-tailed paired t-test, n = 5 embryos, aCSF and H₂O₂ groups).

Rotenone Increases DHE Oxidation and Delays Migration of GABAergic Progenitor Cells

Intracerebroventricular (ICV) rotenone injections were utilized to create a pro-oxidant environment in the embryonic brain in utero, confirmed by assessing a proxy for DHE oxidation levels, DHE fluorescent intensity, in ganglionic eminence (Fig. 2A,B,G). Injections of rotenone into the ventricle of E13 embryonic brain, which generates ROS by inhibiting Complex I of the electron transport chain, resulted in increased DHE mean fluorescent intensity (MFI) in embryonic ganglionic eminence (Mdn = 1.3) compared with saline control (Mdn = 1.0), U = 0.0, P < 0.01. We further assessed the impact of this redox dysregulation on interneuron migration in utero. Rotenone ICV injections resulted in delays to the leading edge.

Figure 2.

Rotenone increased levels of DHE oxidation and delayed migration in the E14 brain. ICV-injected rotenone increased levels of DHE oxidation in E14 ganglionic eminence (GE) tissue compared with ICV-injected saline, as measured by dihydroethidium (DHE) mean fluorescent intensity (MFI) of cells and normalized to MFI = 1.00 (A, B, G). Prenatal stress (PS) delayed progenitor migration at E14 (D) compared with nonstressed (C) embryonic neocortical tissue (H). ICV-injected rotenone (F) also delayed migration of GAD67GFP+ cells at E14 compared with ICV-injected saline (E, H). For the DHE experiments, ICV Saline: n = 5, and ICV Rotenone: n = 5 and Mann–Whitney U test for hypothesis testing (*P < 0.05). For the migration experiments, NS: n = 10, PS: n = 4, ICV Saline: n = 8, and ICV Rotenone: n = 8. (*P < 0.05 compared with ICV saline by paired t-test, ^$P < 0.05 compared with NS by paired t-test).

of migrating GABAergic neurons 24 h later compared with saline injected controls (P < 0.05; Fig. 2E,F,H). The delay of GABAergic neurons recapitulates the migration delay at E14 after prenatal stress (P < 0.05; Fig. 2C,D,H) as shown previously (Stevens et al. 2013).

Prenatal Stress Alters the Expression of Several Key Genes Involved in Redox Regulation

We assessed the expression of genes involved in redox regulation in E13 brain after prenatal stress at E12, including those for antioxidant defense (for a review of the antioxidant enzyme defense system in the embryonic brain, please see (Shim and Kim 2013)), Gpx1, Cat, Tr1, Sod1, Ses1, Ses2, and Ses3 (Kopnin et al. 2007), mitochondrial proteins involved in generating ROS (Jo et al. 2001; Lemarié et al. 2004; Leman et al. 2015), Idh2, Aifm1, and Ndufs4, and the antioxidant response factor Nrf2 (Nguyen et al. 2009) (Fig. 3). After adjusting for multiple comparisons, prenatal stress increased gene expression of the antioxidant defense gene, Tr1 (P < 0.05; Fig. 3A) and increased expression of a gene that contributes to ROS production, Aifm1 (P < 0.01; Fig. 3I). The increase in Aifm1 may be a contributor to an increased pro-oxidant state in embryonic brain which would drive increased expression of Nrf2, a master regulator of the antioxidant response, after prenatal stress (P < 0.01; Fig. 3H). While Nrf2 expression can upregulate multiple antioxidants and other factors through binding in promoters with antioxidant response elements (i.e., TR1, glutathione peroxidases), most antioxidant genes we assessed were unchanged. This may suggest that increased Nrf2 gene expression did not lead to increased nuclear translocation of NRF2 and/or other factors led to an insufficient attempt of cells to upregulate antioxidant capacity, contributing to a dysregulated redox environment.

Figure 3.

Prenatal stress (PS) altered the expression of some genes involved in redox regulation. Only Tr1 (A) among antioxidant activity genes (A–G) was altered of other genes related to redox regulation (H–K), only Nrf2 and Aifm1 were altered. (NS = nonstressed; NS: n = 20, PS: n = 20; *P < 0.05, **P < 0.01 by unpaired t-tests correcting for multiple comparisons using Bonferroni method).

Prenatal Stress Dysregulates Glutathione Stores and Antioxidant Enzyme Activity

To further assess embryonic brain changes as a result of prenatal stress (Stevens et al. 2013), we assessed embryonic brain for thiol states, indicative of redox conditions. Prenatal stress trended to increase the level of reduced to oxidized glutathione (GSH:GSSG) in the embryonic forebrain, compared with the nonstressed condition (P = 0.05; Fig. 4A). There was also no significant difference in reduced GSH levels in control versus prenatally stressed embryonic forebrain (Fig. 4B). However, prenatal stress significantly decreased glutathione disulfide, GSSG (P < 0.01; Fig. 4C), suggesting a lack of antioxidant enzyme (glutathione peroxidase, GPx1) capacity in embryonic brain. Indeed, we found a trend towards a decrease in GPx1 activity after prenatal stress (P = 0.06; Fig. 4D). Prenatal stress also led to a trend 23.7% decrease in thioredoxin reductase 1 (TR1) activity (P = 0.06; Fig. 4E).

Figure 4.

Prenatal stress (PS) disrupted glutathione stores and dysregulated antioxidant enzyme activity. (A) Ratio of reduced glutathione (GSH) to glutathione disulfide (GSSG) showed a trend increase as the result of prenatal stress (vs. nonstress [NS], P = 0.05 by a priori t-test), but showed a main effect of NAC treatment (^##P < 0.01 by 2-way ANOVA) and an interaction of stress and treatment (^ϕP < 0.05). (B) A main effect of NAC (^#P < 0.05) on GSH. (C) A significant baseline difference was observed between nonstress and prenatal stress control brains (**P < 0.01), and main effects of NAC (^##P < 0.01) and stress (^αP < 0.05), as well as an interaction of stress and NAC (^ϕP < 0.05) was found in GSSG brains. (D) A trend decrease in GPx1 activity resulting from prenatal stress (vs. NS, P = 0.06 by a priori t-test) and a main effect of NAC (^##P < 0.01) was detected. (E) A trend decrease in TR1 activity was found between NS and PS embryonic forebrains (P = 0.06 by a priori t-test). For glutathione experiments, NS: n = 6, PS: n = 6, NS NAC: n = 10, and PS NAC: n = 7. For GPx1 experiment, NS: n = 12, PS: n = 14, NS NAC: n = 3, and PS NAC: n = 3. For TR1 experiment, NS: n = 12, and PS: n = 14.

Because of these changes, we further assessed whether prenatal stress effects on embryonic brain redox dysregulation would be influenced by maternal N-acetylcysteine (NAC) administration, which restores intracellular glutathione and cysteine (Olsson et al. 1988). A 2-way ANOVA revealed significant main effects of NAC (F[1, 24] = 10.13, P < 0.01) and an interaction of NAC with prenatal stress effects (F[1, 24] = 4.66, P < 0.05) on the reduced to oxidized glutathione (GSH:GSSG) ratio in the embryonic forebrain (Fig. 4A). NAC treatment also increased levels of reduced glutathione (F[1, 25] = 4.31, P < 0.05; Fig. 4B) and decreased oxidized GSSG levels (main effect of NAC: F[1, 24] = 12.10, P < 0.01, NAC, stress interaction: F[1, 24] = 5.91, P < 0.05; Fig. 4C). In general, maternal NAC had dramatic impacts on the critical glutathione components of redox regulation and also prevented any additional dysregulation of glutathione stores in prenatally stressed embryonic brain compared with control.

Maternal NAC also significantly increased GPx1 enzyme activity level (main effect: F[1, 22] = 357.20, P < 0.01; Fig. 4D). Both nonstressed and prenatally stressed embryos had increased GPx1 activity (174.9% increase, P < 0.01 and 231.5% increase P < 0.01 respectively; Fig. 4D), with prenatal stress no longer showing any decrease (trend or otherwise) in activity compared with nonstressed embryonic brain in the presence of NAC.

Prenatal Stress Increased Levels of DHE Oxidation in Embryonic Brain

Prenatal stress influenced not only antioxidant enzymes but also shifted redox balance enough in the pro-oxidant direction to alter levels of DHE oxidation in embryonic GABAergic progenitors, as assessed by DHE MFI in ganglionic eminence. Antimycin-A increased MFI as predicted in nonstressed control brains as compared with control (a priori t-test, P < 0.01; Fig. 5G). The nonparametric test, Kruskal–Wallis, showed that there was a statistically significant difference in DHE MFI among all groups of the prenatally stressed and nonstressed NAC embryos and the prenatally stressed and nonstressed control embryos (H[3] = 18.56, P < 0.01, Fig. 5H), with a mean rank MFI value of 19.5 for nonstressed control (Fig. 5A), 33.8 for prenatally stressed control (Fig. 5B), 14.4 for nonstressed NAC (Fig. 5C), and 15.0 for prenatally stressed NAC (Fig. 5D). Post hoc tests of Dunn’s multiple comparisons test found differences between the MFI of nonstressed and prenatally stressed control embryos (P < 0.05), prenatally stressed control and nonstressed NAC embryos (P < 0.01), and prenatally stressed control and prenatally stressed NAC embryos (P < 0.01). The Kruskal–Wallis test also showed a significant difference in MFI among the prenatally stressed and nonstressed AST embryos and the prenatally stressed and nonstressed control embryos (H[3] = 19.99, P < 0.01, Fig. 5H), with a mean rank MFI value of 19.5 for nonstressed control (Fig. 5A), 34.3 for prenatally stressed control (Fig. 5B), 15.1 for nonstressed AST (Fig. 5E), and 13.8 for prenatally stressed NAC (Fig. 5F). Post hoc tests of Dunn’s multiple comparisons test found differences in the MFI of nonstressed and prenatally stressed embryos (P < 0.05), prenatally stressed control and nonstressed AST embryos (P < 0.01), and prenatally stressed control and prenatally stressed AST embryos (P < 0.01).

Figure 5.

Prenatal stress (PS) increased levels of DHE oxidation in the ganglionic eminence (GE) of the embryonic brain. (A) All analyses were normalized to mean fluorescent intensity, MFI = 1.00 of control (Con) tissue. Measurements of MFI were taken with ×40 objective images. (B) PS increased MFI (*P < 0.05 compared with NS). In the absence of PS, MFI in NS NAC (C) and NS AST (E) were reduced below the levels of PS control (**P < 0.01). PS NAC (D) and PS AST (F) decreased levels of MFI (**P < 0.01 compared with PS control). (G) Antimycin-A (AA) was used as the positive control and showed the highest DHE oxidation (^$$P < 0.01 by a priori t-test compared with NS control). (Kruskal–Wallis and Dunn’s tests run for NAC and AST exposure effects separately).

Delays in GABAergic Progenitor Cell Migration After Prenatal Stress Normalized by Maternal Antioxidant

As reported previously (Stevens et al. 2013), prenatal stress delayed the migration of GABAergic progenitor cells into the developing neocortex at E13 and E14 (Fig. 6). Here, we replicated a delay after prenatal stress, compared with both nonstressed (P < 0.05, Fig. 6A,D) and maternal saline control (P < 0.01, Fig. 6G) conditions.

Figure 6.

Delays in GABAergic progenitor cell migration after PS normalized by maternal antioxidants. (A) NS control migration, (B) NS NAC migration, and (C) NS AST migration. (D) PS control migration was significantly delayed compared with NS control (^$P < 0.05 by a priori t-test) and NS saline (micrograph not shown; quantitative data shown in G). (E) PS NAC migration was significantly increased compared with PS control (**P < 0.01) and a main effect of NAC (^##P < 0.01) was revealed. (F) PS AST migration was significantly increased compared with PS control (**P < 0.01) and a main effect of AST (^##P < 0.01) was found. A main effect of stress (^αP < 0.05) was also unveiled between PS embryos (D–F) and NS embryos compared with (A–C). The interaction of stress and NAC treatment (^ϕϕP < 0.01) was also significant (G). (H) Quantitative PCR of the Cxcr4 gene, coding for a receptor expressed by migrating inhibitory progenitor cells (P = 0.06) (2-way ANOVAs run for NAC and AST exposure effects separately). For migration: NS: n = 10, Saline: n = 10, PS: n = 8, NS NAC: n = 10, PS NAC: n = 10, NS AST: n = 8, and PS AST: n = 9. For Cxcr4 gene expression: NS: n = 17 and PS: n = 19.

Since redox dysregulation disrupted GABAergic progenitor migration and prenatal stress induced redox dysregulation, we tested the hypothesis that the delay after prenatal stress was due to redox shifts. We increased maternal antioxidant status and found advanced GABAergic progenitor migration. Maternal NAC (F[1, 34] = 10.72, P < 0.01; Fig. 6B,E) and AST (F[1, 34] = 18.60, P < 0.01; Fig. 6C,F) significantly increased migration. NAC also interacted with prenatal stress in a way that resulted in normalized migration (NAC: F[1, 34] = 8.96, P < 0.01). Indeed, the extent of migration after prenatal stress differed significantly with and without either maternal NAC or AST (post hoc, P < 0.01).

Prenatal stress also trend-wise decreased expression of a chemoreceptive gene involved in migration, Cxcr4, (Sanchez-Alcaniz et al. 2011) in GABAergic progenitors (P = 0.06; Fig. 6H). Cxcr4 expression in the presence of maternal NAC was not affected by prenatal stress, although expression after prenatal stress was not significantly different dependent on NAC condition.

Discussion

Redox dysregulation is a mediator of the impact of multiple environmental exposures, suggesting that it may also play a role in mediating the effects of in utero environmental changes like prenatal stress (Wells et al. 2009). The balance between oxidative and reductive processes is critical during rapid periods of cell growth and differentiation, so even small changes in ROS during embryonic brain development could have significant effects (Dennery 2007). We show here that redox dysregulation influences interneuron migration and that prenatal stress leads to redox dysregulation in the embryonic brain. We also show that manipulation of redox balance (in this case, maternal administration of antioxidants) interacts with prenatal stress effects on embryonic brain and can prevent both the increased level of ROS and migration delay. These findings suggest a biochemical pathway that could be targeted to prevent altered neurodevelopment exemplified by prenatal stress effects.

Direct exposure of embryonic brain to the pro-oxidant, hydrogen peroxide (H₂O₂), results in deviations of cellular direction and increased velocity (Fig. 1L,M). Although H₂O₂ is not a free radical by definition, it can form hydroxyl radicals through the Fenton reaction with Fe²⁺ and is detoxified by classic antioxidants (Burton and Jauniaux 2011). This increased rate of movement in multidirectional, nontangential pathways appears somewhat different from the overall delay in migration seen in embryonic brain after prenatal stress described here and previously (Stevens et al. 2013). This difference may be the result of multiple phenomena. First, in vivo experimental assessments focused on the leading edge of migration but not other cell movements; aberrant directionality of movement within in vivo experiments therefore may be present but not measured. The cross-sectional analysis of fixed embryonic brain also focused on migration within the superficial, pial route, while live-imaging analyses evaluated movement of cells in both superficial and deep pathways which may differ. Future work will examine these questions. Second, effects of repetitive prenatal restraint stress may result from multiple acute ROS exposure episodes during which the movement we measured in vitro may occur. In this case, once redox balance is restored in the embryonic brain, migrating interneurons may resume their normal migration patterns, however, delayed overall due to the “detours” taken.

ROS, such as H₂O₂, may be strongly connected to mitochondrial dysfunction including oxidative phosphorylation (Burton and Jauniaux 2011), disruptions of which alter migratory trajectory (Lin-Hendel et al. 2016). (Lin-Hendel et al. 2016) found migration patterns of cortical interneurons with dysfunctional mitochondria exhibited more frequent and aberrant directional changes and slower migratory rates. The increase in multidirectional migration patterns that we found here may suggest a mitochondrial redox imbalance induced by H₂O₂. Our findings that migration was also delayed by in utero intracerebroventricular injection of rotenone (Fig. 2) also support a central role for mitochondrial redox regulation in the movements of this progenitor cell population.

The impact of prenatal stress on interneuron migration (Stevens et al. 2013) may involve redox effects on the embryonic brain. In support of this hypothesis, we found changes in the expression of redox-related genes in embryonic brain after prenatal stress (Fig. 3). Increased expression of apoptosis-inducing factor 1, mitochondrial (Aifm1) is critical for normal mitochondrial function and may indicate that a subset of the embryonic progenitors assessed had greater energy metabolism after prenatal stress and may have generated more ROS in the process. Increased expression of thioredoxin reductase 1 (Tr1) suggests an attempted compensatory response to oxidative stress. Elevated nuclear factor erythroid-derived 2 (Nrf2) levels demonstrate that embryonic neuronal precursors in vivo have a concerted canonical response to maternal stress to induce genes with antioxidant response elements to which Nrf2 binds. While we were surprised to find few changes in antioxidant gene expression, this may reflect that prenatally stressed embryonic brain reaches a state of redox dysregulation through other nontranscriptional mechanisms such as increased functioning of pro-oxidant pathways and reduced activity of antioxidants.

These gene expression findings suggest that the embryonic brain was in a more pro-oxidant state after prenatal stress. However, glutathione redox balance did not demonstrate increases in GSSG. Reduced glutathione (GSH) is an important antioxidant that in combination with GPx1 activity directly reduces hydroperoxides, in turn producing glutathione disulfide (GSSG). As such, the ratio of oxidized to reduced form (GSH:GSSG) is used as a marker of redox dysregulation, or oxidative stress (Zitka et al. 2012). The decreased levels of GSSG we found after prenatal stress suggest that embryonic brain had lower levels of oxidative stress. When taken with our data that shows trend lower GPx1 enzyme activity level as a result of prenatal stress (Fig. 4D), there may be a deficiency of this antioxidant pathway induced by prenatal stress. In the absence of sufficient GPx1 activity, glutathione levels (GSH and GSSG) may shift to a more reduced state. Other enzyme activity not assessed here may also contribute to this shift including glutathione peroxidase 4 (GPx4), glutathione transferase (GST), and nonselenium-dependent peroxidases (Forman et al. 2009). Alternatively, stress may induce an adaptive response that reduces the amount of oxidized glutathione or reduces transport of nutrients like selenium that influence glutathione. All are intriguing mechanisms that will require additional work to parse their contribution. The influence of NAC on glutathione peroxidase activity corrected the trend deficit of activity found with prenatal stress and increased overall glutathione levels as expected. In the presence of maternal NAC, prenatal stress did not alter the GSH:GSSG ratio as it trended to do in the absence of NAC (Fig. 4).

Interestingly, TR1 and redox states are interdependent (Perez-Torres et al. 2017). Despite an increase in Tr1 gene expression with prenatal stress, its activity level trended downwards (Fig. 4E) which may result from glutathionylation facilitated by an oxidative state of the cell (Casagrande et al. 2002). ROS may have arisen locally but also may have been contributed from the maternal-placental unit in a diffusible form such as oxidized lipids.

N-acetylcysteine (NAC) is an aminothiol and synthetic precursor of intracellular GSH (van Zandwijk 1995). When given to pregnant dams, ROS levels abnormally increased by prenatal stress were normalized (Fig. 5D), indicated by DHE oxidation, which measures superoxide and other pro-oxidants. Under normal physiological conditions, the most ubiquitous oxygen free radical is the superoxide anion. Superoxide and other radicals are detoxified by the SOD enzyme, which converts it to hydrogen peroxide (a less reactive molecule). When given simultaneously with restraint stress, astaxanthin (AST) also normalized DHE oxidation in the brain (Fig. 5F). AST, a xanthophyll carotenoid, is a potent antioxidant that is known to cross the blood brain barrier and scavenge ROS. It possesses neuroprotective properties by enhancing the antioxidant enzymes SOD and GPx1 and reducing lipid peroxidation (Liu and Osawa 2009; Wu et al. 2014; Al-Amin, Sultana, et al. 2016). Known to cross the placenta, NAC works to restore GSH and may be able to cross the blood brain barrier (Horowitz et al. 1997; Bavarsad Shahripour et al. 2014). Nonetheless, the ability of GSH to quench superoxide radicals has been demonstrated before (Heribert and Helmut 1983). NAC and AST may both act either directly on embryonic brain by potentially crossing the placenta or through altering maternal redox balance which secondarily impacts embryonic brain. These findings demonstrate that maternal redox manipulation and maternal stress interact in previously undiscovered ways to influence embryonic brain development.

The influence of maternal states on embryonic brain redox balance is critical, given the experiments here that are the first to show that direct redox dysregulation of embryonic brain disrupts interneuron migration. Here, we replicated our previous findings showing that prenatal stress influences interneuron migration (Stevens et al. 2013) and applied the method of maternal redox manipulation to test its efficacy in preventing these effects (Fig. 6). Offspring of mothers pretreated with NAC or AST showed normalized interneuron migration despite prenatal stress (Fig. 6E,F, respectively). These findings are in line with the impact of high levels of oxidative radicals on radial migration in the developing forebrain (Narasimhaiah et al. 2005), although unlike radially migrating cells, prenatal stress does not induce apoptosis in migrating interneurons (Stevens et al. 2013). The mechanism by which pro-oxidant states impact interneuron migration may involve their chemoreceptive properties, such as Cxcr4 expression. Cxcr4 is influenced by the repressor Yin Yang 1 which is sensitive to pro-oxidant states (Hasegawa et al. 2001; Beck et al. 2010). However, there may also be nontranscriptionally mediated impacts of redox dysregulation on mitochondrial function which is critical to cell migration. These findings suggest an important mechanism by which environmental exposures may influence neurodevelopment that may converge with genetic risk such as 22q11 mutations that influence the same cellular process (Meechan et al. 2012).

Indirect interaction of maternal redox manipulation with prenatal stress, rather than direct redox impacts on embryonic brain, may be responsible for the effects we found here. However, we observed significant main effects of NAC and AST treatment, both on DHE oxidation levels in embryonic brain and on extent of migration (Figs 5 and 6), further demonstrating the sensitivity of neuronal precursors generally and cortical interneuron migration specifically to redox regulation.

To our knowledge, these experiments are the first to investigate the role of dysregulated redox states as an intermediary between the effects of prenatal stress and alterations in embryonic brain. Numerous studies have shown that redox dysregulation may be a component of physiological stressors that occur during pregnancy (Peuchant et al. 2004; Cambonie et al. 2007; Derks et al. 2010; Ziech et al. 2011). Targeting an appropriate balance of redox biology during neurodevelopment may have benefits for offspring brain and overall development. In particular, these findings may be applicable to elucidating poorly understood mechanisms of early contributions to mental illness, including genetic, biological, and environmental causes, and ultimately developing better treatments and interventions for those at risk for neuropsychiatric disorders.

Funding

Nellie Ball Trust Research Grant to H.E.S.; the Roy J. Carver Charitable Trust Junior Research Program of Excellence to H.E.S.; the University of Iowa Dean’s Graduate Research Fellowship to J.B.; and a National Institute of Health Training Grant to the University of Iowa Neuroscience Graduate Program (T32NS007421) to J.B.