Gene Dysregulation in the Adult Rat Paraventricular Nucleus and Amygdala by Prenatal Exposure to Dexamethasone

Fetal programming is the concept that maternal stressors during critical periods of fetal development can alter offspring phenotypes postnatally. Excess glucocorticoids can interact with the fetus to effect genetic and epigenetic changes implicated in adverse developmental outcomes. The present study investigates how chronic exposure to the synthetic glucocorticoid dexamethasone during late gestation alters the expression of genes related to behavior in brain areas relevant to the regulation and function of the hypothalamic–pituitary–adrenal axis. Pregnant Wistar Kyoto rats received subcutaneous injections of dexamethasone (100 μg/kg) daily from gestational day 15–21 or vehicle only as sham controls. The amygdala and paraventricular nucleus (PVN) were micro-punched to extract mRNA for reverse transcription and quantitative polymerase chain reaction for the analysis of the expression of specific genes. In the PVN, the expression of the glucocorticoid receptor NR3C1 was downregulated in female rats in response to programming. The expression of CACNA1C encoding the Cav1.2 pore subunit of L-type voltage-gated calcium channels was downregulated in male and female rats prenatally exposed to dexamethasone. Collectively, the results suggest that prenatal exposure to elevated levels of glucocorticoids plays a role in the dysregulation of the hypothalamic–pituitary–adrenal axis and potentially learning and memory by altering the expression of specific genes within the amygdala and PVN.


Introduction
The thrifty phenotype hypothesis put forth by Barker and Hales in 1992 attempted to explain the development of impaired glucose tolerance and later metabolic syndrome with respect to impaired growth in fetal and infantile stages of development [1][2][3]. Their research was among the first to describe the consequences of fetal programming. In fetal programming, stressful conditions in the external environment producing maternal distress are effectors of phenotypic alterations to progeny postnatally [3][4][5]. These alterations are not necessarily adaptive and tend to contribute to the development of disease in later life.
Other sources of fetal programming include fetal glucocorticoid (GC) exposure altering hypothalamic-pituitary-adrenal (HPA) axis activity [6]. These GCs may be endogenous or synthetic, as with dexamethasone (Dex) administered to mothers at risk for preterm labor.
The HPA axis is the physiological system responsible for the secretion of GCs such as cortisol in humans. Adrenocorticotropic hormone (ACTH) released from the anterior pituitary gland stimulates cellular activity, promoting the synthesis of cortisol from cholesterol in the adrenal cortex [7]. During an acute response to an immediate stressor, the amygdala processes the initial threatening stimuli, signaling the hypothalamus, which subsequently activates the sympathetic nervous system, keeping the organism in an alert and energized state for a reaction to the initial stressor [8]. If it is necessary to maintain this state over Figure 1. Diagrammatic representation of the HPA axis and modulation by the amygdala. While endogenous GCs provide a negative feedback mechanism after their production in the adrenal gland and the occupation of GC receptors, the amygdala is known to stimulate the axis upstream of the adrenal glands. Created with BioRender.com accessed on 23 June 2022.
We have previously shown that Dex administration to pregnant Wistar-Kyoto rats as a proxy for late gestational stress can elicit changes in the gene expression in the offspring prefrontal cortex (PFC), as well as in behavior [18]. Behavior observed on the Porsolt swim test included the greater immobility of the Dex-programmed offspring relative to controls on the first day of testing, while on the second day, female rats in both treatment and control conditions displayed increased immobility relative to males. The elevated plus maze (EPM) is a well-validated test whereby anxiety-like behaviors, or the lack of these, can be measured based on the typical rodent's proclivity for dark, enclosed spaces and locations away from observable heights [45]. Dex-programmed animals entered the open arms with 4 paws more often than the sham-treated offspring, while entering the enclosed arms with 2 paws less often, reflecting a reversal of expected normal thigmotaxic behaviors [18,46]). In this study, we measured changes in gene expression in the PVN and amygdala of the same Dex-programmed animals from the original study.
Given what is known about GCs and their role in the fetal programming of the brain and behavior, the question arises as to how GC programming may contribute to HPA axis dysregulation and the production of maladaptive behaviors, [38,[47][48][49]. Two brain areas, the PVN and amygdala, were selected for their richness in GRs, relevance to regulating the HPA axis, and roles in behavior. Considering the evidence at hand, we hypothesize that exposure to synthetic GCs prenatally programs the dysregulation of gene expression in the amygdala and PVN, with potential implications for behavioral outcomes.

Materials and Methods
Experimental protocols involving the use of animals were approved by the Animal Figure 1. Diagrammatic representation of the HPA axis and modulation by the amygdala. While endogenous GCs provide a negative feedback mechanism after their production in the adrenal gland and the occupation of GC receptors, the amygdala is known to stimulate the axis upstream of the adrenal glands. Created with BioRender.com accessed on 23 June 2022.
We have previously shown that Dex administration to pregnant Wistar-Kyoto rats as a proxy for late gestational stress can elicit changes in the gene expression in the offspring prefrontal cortex (PFC), as well as in behavior [18]. Behavior observed on the Porsolt swim test included the greater immobility of the Dex-programmed offspring relative to controls on the first day of testing, while on the second day, female rats in both treatment and control conditions displayed increased immobility relative to males. The elevated plus maze (EPM) is a well-validated test whereby anxiety-like behaviors, or the lack of these, can be measured based on the typical rodent's proclivity for dark, enclosed spaces and locations away from observable heights [45]. Dex-programmed animals entered the open arms with 4 paws more often than the sham-treated offspring, while entering the enclosed arms with 2 paws less often, reflecting a reversal of expected normal thigmotaxic behaviors [18,46]). In this study, we measured changes in gene expression in the PVN and amygdala of the same Dex-programmed animals from the original study.
Given what is known about GCs and their role in the fetal programming of the brain and behavior, the question arises as to how GC programming may contribute to HPA axis dysregulation and the production of maladaptive behaviors, [38,[47][48][49]. Two brain areas, the PVN and amygdala, were selected for their richness in GRs, relevance to regulating the HPA axis, and roles in behavior. Considering the evidence at hand, we hypothesize that exposure to synthetic GCs prenatally programs the dysregulation of gene expression in the amygdala and PVN, with potential implications for behavioral outcomes.

Materials and Methods
Experimental protocols involving the use of animals were approved by the Animal Care Committee of Laurentian University (AUP: 6013917) and followed guidelines established by the Canadian Council on Animal Care. Eight-week-old Wistar-Kyoto rats purchased from Charles River Laboratories (Saint-Constant, Québec, Canada) were housed in groups of two or three in Innocage ® IVC disposable cages (Innovive Inc., San Diego, CA, Life 2022, 12, 1077 4 of 14 USA). Cob bedding (Harklan, Madison, WI, USA) as well as Bio-serv enrichment tubes (Frenchtown, NJ, USA) were present in the cages. Cages were kept in a HEPA-filtered Innorack ® rat airflow system (Innovive Inc., San Diego, CA, USA). Both food (Teklad 22/5 rodent diet, Harklan, KY, USA) and water were made available to the animals ad libitum. A 12-h day-night cycle was instituted, with the former beginning at 6:00 a.m. Ambient room temperature was maintained at 25 • C with 53% relative humidity.
Male rats were introduced singly per three females over a period spanning five days, after which female rats were checked for the presence of a vaginal plug to confirm the occurrence of mating. Pregnant females were housed alone and weighed daily for the duration of their gestational span. Upon gestational day fifteen, and every day thereafter, pregnant rats received one of two treatments. Four dams were selected for the exposure group and were subjected to daily subcutaneous injections of Dex (Sigma-Aldrich, Milwaukee, WI, USA; 0.9% sodium chloride, 4% ethanol, and 100 µg per kg body mass Dex), while another dam received vehicle only to produce sham controls. Injections were given during the light cycle, as previously described [17,18,20].
Pups were raised normally and weaned at three weeks of age. Eight pups were selected per sex from the four Dex-exposed dams, while eight pups per sex were selected from the sham exposure mother. Upon reaching nineteen weeks of age, young adult rats were euthanized by intraperitoneal injection of 75 mg ketamine (100 mg/mL, Ketalean, CDMV Inc., Brampton, ON, Canada) and 5 mg xylazine (100 mg/mL, Sigma, Milwaukee, WI, USA) per kilogram of body mass. Tissues of interest were collected immediately and subsequently frozen using dry ice. Whole brains were collected and stored at −80 • C for future genetic and morphological analyses. Coronal sections of rat brains with widths of 5-20 µm were made using a CM3050 S cryostat (Leica Biosystems, Deer Park, IL, USA) maintained at −20 • C. Using the rat brain atlas of Paxinos & Franklin (1997), areas of interest were identified, and anterior-posterior measurements from bregma were used to delineate their locations in the sections [50]. Punctures in the sections were made using a sterile pipette tip to retrieve relevant tissues within the Cryostat.
The tissues retrieved from the right amygdala and the paraventricular nucleus of each rat were placed in TRI reagent (Sigma-Aldrich, Oakville, ON, Canada; 1 mL/50 mg of tissue) and subjected to two 2-min cycles at 30 hertz in a TissueLyser (Qiagen, Hilden, Germany) for the mechanical homogenization of the sample. mRNA was then extracted from the samples using a TRIzol extraction method, as previously described [18].
Using Primer3 and BLAST, primers for use in real-time quantitative polymerase chain reaction (PCR) were designed for the experiment (Table A1). As per Livak and Schmittgen (2001), primers were validated using serial dilutions of cDNA and amplified in PCR [51]. The specificity of primers was evaluated using melt curves post amplification. Genes of interest were selected based on a whole-genome microarray analysis of the brain [52] and included those relevant to the metabotropic glutamate receptor pathway, calcium signaling, neural differentiation and growth, and glucocorticoid receptors. Two reference genes, CycA and Ywhaz, were selected based on Bonefeld et al. [53].
Relative gene expression between treatments (n = 8 per sex per treatment) was analyzed using the QuantStudio5 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Amounts of 15 µL reaction volume per sample in 96-well plates contained 0.4 ng/µL cDNA, DEPC-treated water, 1.2 µM/µL forward and reverse primers for each respective gene, and SYBR green master-mix (SensiFAST SYBR Lo-ROX, Bioline, FroggaBio, Toronto, ON, Canada). Changes in relative gene expression between treatments with respect to reference genes, with the sham male group as a calibrator, were measured using the ∆∆C T method, followed by melt curve analysis [51].
Statistical analyses for genetic comparisons between groups were conducted using Jamovi v1.6.23 and R version 4.1.3 [54,55]. The parametric assumptions of normality and homoscedasticity in analysis of variance (ANOVA) were tested using the Shapiro-Wilk and Levene's tests, respectively. Where the assumption of homoscedasticity was rejected between treatments, Welch's test was conducted. Where both assumptions were met, two-factor ANOVAs with interaction were conducted. Where primary effects of treatment were determined in ANOVA, Dunnett's post hoc test was employed to assess differences in gene expression between rats born to Dex-exposed mothers and their respective controls. Where primary effects were a result of sex rather than treatment, Tukey's post hoc test was employed to assess differences in gene expression between male and female control rats as well as male and female rats born to Dex-exposed mothers.

Paraventricular Nucleus
Of the fifteen genes selected for analysis, two of them displayed dysregulation as an effect of treatment in the PVN (Table 1). A significant effect of treatment was demonstrated by ANOVA for the expression of the Nuclear Receptor Subfamily 3 Group C Member 1 (NR3C1; F 1,17 = 6.81, p = 0.018, η 2 = 0.263). No significant effect of animal sex (F = 0.522, p = 0.480, η 2 = 0.020) or interaction was noted (F = 1.509, p = 0.236, η 2 = 0.058) Post hoc Dunnett's test reveals a significant decrease in NR3C1 expression in Dex condition female rats relative to respective sham control (p = 0.031; Figure 2a). The gene encoding Glutamate Ionotropic Receptor AMPA Type Subunit 2 (GRIA2) shows a significant effect of treatment in ANOVA (F 1,19 = 5.40, p = 0.031, η 2 = 0.212). However, Dunnett's test indicates that differences in expression between Dex condition animals showed a trend but were not statistically significant at p < 0.05, compared with sham controls. Solute Carrier Family 1 Member 2 (SLC1A2) displayed a significant effect of sex (F 1,19 = 5.24, p = 0.034, η 2 = 0.187) rather than treatment (F = 0.446, p = 0.512, η 2 = 0.016). In this case, a sex-treatment interaction trend is noted but is not significant, at F = 3.282, p = 0.086 and η 2 = 0.117. Sham females appear to have higher baseline expression of SLC1A2 that is attenuated in the Dex exposure condition, whereas lower baseline expressing males appear to display slight upregulation in the Dex exposure condition. Post hoc Tukey's test indicates that sham control females have higher expression of this gene compared with male rats (p = 0.036; Figure 2b). Notably, this difference between sham female and male rats is attenuated in both of the Dex exposure conditions.

Discussion
Traditionally known for its role in the HPA axis, recently the PVN is receiving further attention for its roles in modulating the emotional and anxiety networks as well as anxi-

Discussion
Traditionally known for its role in the HPA axis, recently the PVN is receiving further attention for its roles in modulating the emotional and anxiety networks as well as anxietyand depression-like behaviors [56]. Another key player in these networks is the amygdala, with specific nuclei having excitatory roles in increasing HPA axis activity [9]. With normal neural development, these areas of the brain are integral to adaptive behaviors in response to stressful stimuli. It is expected that alterations in these areas as a result of fetal programming would alter key features in adaptive behaviors.
Differential expression of genes was observed in the PVN as an effect of treatment. NR3C1, encoding the GR, was downregulated in Dex condition female rats specifically. This effect is thought to reflect a decrease in negative feedback to the PVN by GCs that would facilitate and maintain the HPA axis response to stressors [57]. Chronic perinatal exposure to GCs has previously been shown to decrease GR expression in the PVN, raising CRH and endogenous GC production with implications for the development of mood and anxiety disorders [58][59][60]. Another study using similar conditions to those described here reported GR downregulation in the hippocampus of WKY rats that also involved negative feedback of the HPA axis, as suggested by elevated levels of corticosterone in animals born to Dex-treated mothers [61]. Results such as these tend to correspond with increases in circulating endogenous GCs, indicating overactivity of the HPA axis [62]. For example, Dex-programmed vervet monkeys demonstrate elevated blood cortisol levels following mild stress compared with controls [63]. A similar presentation is observed in prenatally stressed rhesus monkeys by Schneider et al. (2002), accompanied by reduced locomotion and exploratory activity [64]. A more robust activation of the HPA axis is often reported in female animals compared with males, with estrogens implicated as modulating HPA axis activity by decreasing negative inhibition of the axis in the central nervous system, including the PVN [65,66]. It is thought that females may be more sensitive to the effects of gestational exposure to stress hormones as a result [67]. The results of this study may indicate that fetal programming by GCs causes dysregulation of the HPA axis in female rats selectively, with estrogens contributing to observed differences in sex. While sex differences associated with Dex programming were not apparent in behavioral testing in the previous study, behaviors described as adaptive in response to stress were observed in general for animals born to Dex-exposed mothers in the PST and EPM [18]. Alterations in HPA axis activity and regulation may support adaptive stress-related behavior in response to adverse conditions, such as the PST, though the current results do not support a mechanism explaining altered behavior in programmed male rats.
The expression of GRIA2, encoding a subunit of the AMPA-type glutamate receptor (AMPAR), was upregulated in Dex-condition offspring non-specifically within the PVN. AMPARs are glutamate-gated heterotetrameric ion channels mediating most of the fast excitatory synaptic transmission in the brain [68,69]. AMPARs have been well studied in the processes of memory and learning, though their role elsewhere is poorly understood [70]. Hypothalamic AMPARs facilitate autonomic responses to stressors, such as increases in blood pressure, where the enrichment of PVN neurons with AMPARs lacking GRIA2 increases excitability related to hypertension [71,72]. Reductions in AMPAR transmission have been implicated in anxiety and stress, while disruption in the trafficking of the GRIA2 subunit has been associated with predisposition to stress [71,[73][74][75]. Finally, positive allosteric modulators of AMPARS produce antidepressant effects and have improved behavioral and glutamate transmission deficits in rats stressed perinatally [74,76,77]. Targeting hypothalamic AMPARs for novel therapies in diverse disorders has been proposed, though the specific roles maintained by the GRIA2 subunit in these processes are currently understudied [70].
The downregulation of CACNA1C, encoding the Ca v 1.2 pore subunit of L-type voltagegated calcium channels (L-VGCCs), occurred as an effect of treatment in both Dex-condition males and females in the amygdala. This finding contrasts a previous in vitro experiment where the direct application of GCs to basolateral amygdala neurons increased the expression of the Ca v 1.2 subunit [78]. The mechanism linking Dex exposure in gestation and the reduction of CACNA1C in the entire amygdala likely differs considering the current results.
The role of L-VGCCs has also been studied in fear conditioning and memory in the lateral amygdala. During membrane depolarization, influxes of calcium through the NMDA-type and L-VGCCs are thought to underlie the processes of long-and shortterm fear memory formation [79,80]. The inhibition of L-VGCCs, however, impairs the formation of long-term fear memories only [80][81][82]. Similarly, the activity of L-VGCCs is required in the basolateral amygdala for the occurrence of long-term fear extinction [83]. CACNA1C region-specific knockouts and heterozygous mice have displayed a variety of other impairments in different facets of memory, including spatial memory [84]. Deficits in the expression of the Ca v 1.2 subunit in the amygdala may reflect similar memory-deficit phenotypes in rats born to Dex-exposed mothers. Behavioral testing paradigms in the original study were not designed for the detection of fear-related learning and memory, though results obtained in this study indicate such inquiry may be warranted.
Two effects in gene expression by sex were observed in this study that were abolished by treatment. In the PVN, the expression of SLC1A2 was higher in female rats compared with males in the sham control conditions. The encoded membrane-bound protein removes glutamate from the synaptic cleft in the central nervous system, playing an important role in staving off neuronal excitotoxicity [85]. Sex differences in SLC1A2 mRNA expression have not yet been reported to the best of our knowledge. The other effect of sex on gene expression was found within the amygdala, where SNAP25, an integral component of neuronal vesicle exocytosis and docking, was expressed at higher levels in male rats compared with female rats [86]. Both splice variants of the SNAP25 protein, SNAP25a and SNAP25b, are mostly confined to neurons of the central nervous system [87]. The upregulation of the SNAP25 gene was previously reported in the PFC of male rats born to Dex-exposed mothers [18]. Sexually dimorphic expression of SNAP25 has been proposed as one underlying feature of sex differences in brain function [88]. For example, the incidence of attention deficit hyperactivity disorder is higher in males, which may reflect sexually dimorphic differences in frontal lobe functioning. As for the role of SNAP25 in the amygdala, further study is required to determine what function sex differences in gene While behavioral data is useful for corroborating results, the testing paradigms from our earlier study were not designed for the thorough evaluation of processes like learning, fear conditioning and extinction, and responses to stressful situations [18]. A wide array of processes possibly influenced by GC fetal programming in different areas of the brain should be considered in designing future experiments. Evaluating differences in fetal programming modified by sex and sex hormones, especially as they pertain to the HPA axis, also represents a logical target for study given the sex differences in gene dysregulation reported here.

Conclusions
The present study demonstrates that two brain areas that have not received as much attention in fetal programming studies, the PVN and amygdala, are relevant targets for studies on stress programming. Chronic prenatal exposure to GCs in WKY rats alters gene transcription in these areas of the brain that are relevant to regulating the HPA axis, behavior, emotion, and memory, among other features related to the development of mental illness. Further study of how these areas of the brain are impacted by stress programming later in life may provide further evidence to explain their roles in the development of maladaptive behavior and psychiatric illness.