Early Life Stress Alters Gene Expression and Cytoarchitecture in the Prefrontal Cortex Leading to Social Impairment and Increased Anxiety

Early life stress (ELS), such as abuse, neglect, and maltreatment, exhibits a strong impact on the brain and mental development of children. However, it is not fully understood how ELS affects social behaviors and social-associated behaviors as well as developing prefrontal cortex (PFC). In this study, we performed social isolation on weaned pre-adolescent mice until adolescence and investigated these behaviors and PFC characteristics in adolescent mice. We found the ELS induced social impairments in social novelty, social interaction, and social preference in adolescent mice. We also observed increases of anxiety-like behaviors in ELS mice. In histological analysis, we found a reduced number of neurons and an increased number of microglia in the PFC of ELS mice. To identify the gene associated with behavioral and histological features, we analyzed transcriptome in the PFC of ELS mice and identified 15 differentially expressed genes involved in transcriptional regulation, stress, and synaptic signaling. Our study demonstrates that ELS influences social behaviors, anxiety-like behaviors through cytoarchitectural and transcriptomic alterations in the PFC of adolescent mice.


INTRODUCTION
The environment is a crucial factor for providing optimal growth and health conditions for children, including social, cognitive, or immune system-related aspects (Ferguson et al., 2013;Sonuga-Barke et al., 2017;Consiglio and Brodin, 2020;Mackes et al., 2020). Particularly, childhood environments have great impacts on the brain architecture, synaptic plasticity, and mental development (Takesian and Hensch, 2013;Miguel et al., 2019).
For example, adverse childhood environments and experiences such as abuse, neglect, and maltreatment have profound effects on the brain and mental development, and constitutes major risk factors for brain structural changes and adult psychopathology (Teicher et al., 2016;Jaffee, 2017). Human brain imaging studies have reported that maltreatment reduces specific brain region volumes such as anterior cingulate cortex, ventromedial and dorsomedial PFC, hippocampus, and striatum, and these brain regions are highly involved in emotional control and the reward system (Chugani et al., 2001;Cohen et al., 2006;Edmiston et al., 2011;Dannlowski et al., 2012;Teicher et al., 2012;Kelly et al., 2013;Teicher et al., 2016). Importantly, specific brain regions are affected by specific types of abuse, neglect or maltreatment. Exposure to harsh corporal punishment caused a reduced gray matter volume in the prefrontal cortex of young adults (Tomoda et al., 2009). In addition, exposure to parental verbal aggression in childhood reduced fractional anisotropy in the arcuate fasciculus, and affected the development of the auditory association cortex involved in language processing (Tomoda et al., 2011). The child brain development has plasticity, and the effects of the childhood environments and experiences lead to structural and functional changes in the brain, as well as behavioral alterations.
Social isolation, social defeat stress, and electric shock are widely used as animal models to reproduce ELS and traumatic experiences in childhood (Schöner et al., 2017;Aspesi and Pinna, 2019;Pinna, 2019). A previous study of animal models reported that socially isolated mice during postnatal days (P) 21-35 displayed reduced sociality, working memory, and myelination in the PFC (Makinodan et al., 2012). In addition, reduced synapses and endogenous excitatory subtypes of the layer 5 pyramidal neurons were reported in the PFC of mice socially isolated for the same period (Yamamuro et al., 2018). We have also reported that social isolation during P21-53 induced the alteration of gut microbiota in mice (Usui et al., 2021b). These studies indicate that social isolation causes a lack of social experience and isolation stress, resulting in serious impacts on both PFC architecture and physical health. However, it is not fully understood how ELS affects various social behaviors and itsassociated behaviors as well as transcriptome in the PFC of stressed individuals.
To address these questions, we performed social isolation on weaned pre-adolescent mice until adolescence for analyzing the impacts of ELS. In this study, we investigated various social behaviors (interaction, novelty, and preference), anxiety-like behaviors as well as cytoarchitecture and gene expressions by RNA sequencing (RNA-seq) of the PFC in adolescent ELS mice.

Mice
Wild-type C57BL/6N (Japan SLC Inc., Shizuoka, Japan) mouse was used. Mice were housed in the barrier facilities of University of Fukui under 50% humidity and a 12:12 h light:dark cycle (8:00 to 20:00) and given free access to water and food. Social isolationinduced ELS was performed as described previously (Usui et al., 2021b). Briefly, the used experimental and control mice came from the litters of the same mother. For ELS (via social isolation), male mice were individually housed using a filter cap (#CL-4150, CLEA Japan, Inc., Tokyo, Japan) immediately after the weaning at P21 until P50. For controls, 4 male mice were co-housed in the same cage without filter cap. Mice behaviors were assessed from P50 and kept in each housing condition until dissection at P67. A total of 39 mice were used in this study, using littermates that gave birth to 6 or more at a time. The number of mice used in each experiment is shown in corresponding figure legend. All procedures were performed according to the ARRIVE guidelines and relevant official guidelines under the approval (#27-010) of the Animal Research Committee of the University of Fukui. All behavioral tests were performed between 10am and 5pm. Experimenters blind to housing information performed all behavioral tests.

Three-Chamber Social Interaction Test
Three-chamber social interaction test was performed as described previously (Usui et al., 2021b). Sociability and social novelty behaviors were assessed using a three-chamber box (W610 × D400 × H220 mm) with an infrared video camera system (O'Hara & Co., Ltd, Tokyo, Japan) at P53 after collecting fecal samples. For the first trial, empty wired cages were placed into both chambers for habituation. For the second trial, a stranger male mouse (mouse A) was placed into a wired cage of the right chamber to examine the sociability. For the third trial, mouse A was kept in the same wired cage as a familiar mouse, and a novel stranger male mouse (mouse B) was placed into a wired cage of the left chamber to examine the social novelty. Each trial was examined for 5 min after which the interactions with the targets were scored using an infrared video camera system (O'Hara & Co., Ltd, Tokyo, Japan).

Social Interaction Test
Two mice were placed in the diagonal corners of a novel chamber (W480 × D480 × H330 mm) and allowed to examine the social interactions for 10 min. Time spent in social interaction was measured and tracked by digital counters with infrared sensors (SCANET MV-40, MELQUEST Ltd., Toyama, Japan).

Social Preference Test
Mice were placed in one of the corners of a novel chamber (W480 × D480 × H330 mm) and allowed to examine the social preference for 10 min. A stranger male mouse was placed into a wired cage (80 × 80 mm) of the center of arena to examine the social preference. Time spent around the wired cage area (250 × 250 mm) and locomotor activity were measured and tracked by digital counters with infrared sensors (SCANET MV-40, MELQUEST Ltd., Toyama, Japan).

Light-Dark Box Test
Mice were placed in the dark room of a light -dark box (W300 × D150 × H150 mm) and allowed to freely explore for 10 min. Time spent in each box (W150 × D150 × H150 mm), transition, and locomotor activity were measured and tracked by digital counters with infrared sensors (SCANET MV-40, MELQUEST Ltd., Toyama, Japan).

Open Field Test
Mice were placed in one of the corners of a novel chamber (W480 × D480 × H330 mm) and allowed to freely explore for 120 min. Time spent in the center of arena (80 × 80 mm) and locomotor activity were measured and tracked by digital counters with Frontiers in Genetics | www.frontiersin.org November 2021 | Volume 12 | Article 754198 infrared sensors (SCANET MV-40, MELQUEST Ltd., Toyama, Japan).

Marble-Burying Test
Marble-burying test was performed as described previously (Usui et al., 2021a). Mice were placed in the corner of a novel home cage evenly placed eighteen marbles and allowed to freely explore for 20 min. After 20 min, the number of marbles buried was recorded. A marble was defined as buried when less than onethird of the marble was visible.

Immunohistochemistry
Immunohistochemistry was performed as described previously (Usui et al., 2021a ; Nacalai Tesque, Kyoto, Japan) was also used to stain nuclei. Images were collected using an all-in-one fluorescence microscope (BZ-X700, KEYENCE Corporation). Immunopositive cells in the PFC (cingulate and prelimbic cortical regions) were manually quantified within the area (4.8 × 10 5 μm) at bregma 1.98 to 1.78 by an experimenter blind to housing information.

RNA-Seq Data Analysis
RNA-seq data analysis was performed as described previously (Usui et al., 2017a). Raw reads were first filtered for quality and trimmed for adapters using FASQC (http://www. bioinformatics.babraham.ac.uk/projects/fastqc) and Trimmomatic (Bolger et al., 2014). Filtered reads were then aligned to the mouse genome mm10 (https://genome.ucsc. edu) using STAR v2.5.2b (Dobin et al., 2013) allowing three mismatches. Uniquely mapped reads (bam flag NH:i:1) were used to obtain the gene counts using HTSeq package (Anders et al., 2015), and the read counts were normalized to the CPM (counts per million) implemented in the edgeR package (Robinson et al., 2010;McCarthy et al., 2012). For further analysis, we performed a sample-specific CPM filtering considering genes with CPM values of 1 in all replicates for treatments or controls. DESeq (Anders and Huber, 2010;Love et al., 2014) was then used to detect the differentially expressed genes (DEGs). We applied a filter of FDR (adjusted p-value) of <0.05 and absolute log fold change of >0.3 to identify significantly changed genes. Gene ontology (GO) analysis with the significant DEGs was carried out using ToppGene (https://toppgene.cchmc.org), and these GO terms were consolidated using REVIGO (Supek et al., 2011). The list of autism spectrum disorder (ASD) genes was derived from the SFARI database (1010 genes; https://gene.sfari.org/database/ human-gene). The NCBI Gene Expression Omnibus (GEO) accession number for the RNA-seq data reported in this manuscript is GSE180055 (token: ivureikyjdwthuf).

Statistical Analysis
All behavioral data are represented as the means of biologically independent experiments with ± standard errors of the mean (SEM). The statistical analyses (independent samples t-test and Pearson's r) were performed using the GraphPad Prism 9 software. Asterisks indicate p-values (****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05) with p < 0.05 being considered as the threshold of statistical significance.

ELS Caused Adolescent Social Impairments
To investigate the influence of ELS on behaviors, cytoarchitecture, and transcriptome in the PFC of adolescent mice, we performed social isolation immediately after the weaning ( Figure 1A). After the social isolation period (P21-P50), we found that there was no difference in the weight between controls and ELS mice (control 26.89 ± 0.34, ELS 27.55 ± 0.45, 95 %Cl −0.4800 to 1.786, p 0.25) ( Figure 1B).

Neuronal Loss and Microglia Increase by ELS
Next, we examined how ELS affects the PFC cytoarchitecture. We found a slight reduction of the number of NeuN+ mature neurons in the PFC of ELS mice (control 358.00 ± 6.89, ELS 329.80 ± 9.95, 95 %Cl −56.12 to −0.2850, p 0.0482) ( Figures 4A,C,D). We found a slight increase of the number of Iba1+ microglia in the PFC of ELS mice (control 103.20 ± 2.44, ELS 115.00 ± 4.18, 95 %Cl 0.6355 to 22.96, p 0.0407) ( Figures 4B,C,E). These results indicate that ELS caused a decrease in the number of neurons and an increase in the number of microglia in the PFC of adolescent ELS mice.

ELS Altered PFC Transcriptome
To understand the molecular mechanisms underlying behavioral and histological phenotypes, we characterized the transcriptome in the PFC of ELS mice by RNA-seq. Transcriptome profiles were clearly separated between control and ELS mice ( Figure 5A). Differential expression analysis of the RNA-seq data uncovered 15 DEGs (FDR < 0.05) in PFC of ELS mice compared to control mice ( Figure 5B, Supplementary Table S1). GO analysis for the DEGs highlighted functions involved in regulation of transcription by RNA polymerase II, positive regulation of transcription, positive regulation of RNA metabolic process, response to corticosterone, response to mineralocorticoid, and trans−synaptic signaling ( Figure 5C, Supplementary Table  S2). DEGs were also involved in the pathway in glucocorticoid receptor regulatory network, and diseases such as sleep disorders, Clostridium difficile, glucocorticoid receptor deficiency, substance withdrawal syndrome, and stress, psychological ( Figure 5C, Supplementary Table S2). The expressions of DEGs were confirmed in the PFC of ELS mice by qPCR ( Figure 5D). Since ELS mice exhibited social impairments, we overlapped the DEGs with the gene list of ASD, a developmental disorder that shows impairments in social communication. Then, we found that 3 DEGs (Zbtb16, Per2, Lrfn2) overlapped with SFARI ASD genes ( Figure 5E). Together, these results indicate that ELS affected genes are involved in transcriptional regulation, stress signaling, and sleep disorders. Our results also suggest that 3 DEGs (Zbtb16, Per2, Lrfn2) are involved in sociality.

DISCUSSION
In this study, we demonstrate that ELS causes the impairments of social behaviors (sociality, social novelty, and social preference) and an increased anxiety-like behavior in adolescent mice. ELS induced neuronal loss and microglia increase in the PFC of adolescent mice. As the ELS-affected genes, we identified 15 DEGs involved in transcriptional regulation, stress signaling, and sleep disorders in the PFC of adolescent mice. Previous studies have reported that social isolation at preadolescent period caused the reduction of sociality in mice (Makinodan et al., 2012;Usui et al., 2021b). Here we showed that various social impairments and increased anxiety-like behaviors in ELS mice (Figures 2, 3). Social isolation during adulthood has been used as a well-known method of generating depression models, and such mice show increased anxiety-like behaviors (Du Preez et al., 2020). In fact, increased anxiety-like behavior, repetitive behavior, memory impairment, and depression-like behavior have been reported in mice that were socially isolated earlier (P14-50) than our condition (Niwa et al., 2011). Sociality and anxiety are closely related and are controlled by neural circuits centered on PFC and amygdala (Stein and Stein, 2008;Penninx et al., 2021). These two brain regions have been reported as ones of corresponding regions involved in social behaviors (Amaral et al., 2008), and are also known as the brain regions impaired by neglect and maltreatment (Teicher et al., 2016). In the PFC of ELS mice, we observed slight changes in the numbers of NeuN+ mature neurons and Iba1+ microglia ( Figure 4). This result suggests the possibility that ELS may cause inflammation and results neuronal cell death in the brain. To support this idea, previous studies have reported neuronal cell death due to chronic stress (Bachis et al., 2008;Natarajan et al., 2017). Moreover, chronic stress has been reported to increase blood corticosterone and induce an increase in microglia (Du Preez et al., 2020). It will be interesting to analyze the effects of ELS on the amygdala as a future study.
In terms of gene expression in the PFC of ELS mice, we identify total 15 DEGs involved in transcriptional regulation, stress signaling, and sleep disorders ( Figures 5B,C). It was reasonable that there were changes in the genes related to such GOs due to ELS. Among those DEGs, there were many transcription factors, suggesting ELS causes functional changes in PFC. The alterations of genes involved in stress signals were presumed to be the effects of chronic stress due to social isolation. It is known that sleep disorders coexist in social withdrawal (Rubin et al., 2009) and psychiatric disorders such as major depressive disorder (Goldstein and Walker, 2014;Otte et al., 2016). Our data suggests that the alterations of genes associated with circadian rhythm such as Per2 and Npas4 due to the influences of ELS may trigger sleep disorders.
In order to explore the gene involved in sociality, we overwrapped the DEGs with SFARI ASD gene list, and identified Zbtb16, Per2, and Lrfn2 ( Figure 5E). Zbtb16 (Plzf) is known as a transcription factor playing key roles in many biological processes such as stem cell proliferation, differentiation, apoptosis, chromatin remodeling, metabolism and immunity (Suliman et al., 2012;Šeda et al., 2017). ZBTB16 is associated with ASD (Bacchelli et al., 2019) and skeletal defects, genital hypoplasia, and mental retardation (SGYMR) (Wieczorek et al., 2002;Fischer et al., 2008). Zbtb16 KO mice displayed ASD-like behaviors such as social impairment and increased repetitive behaviors as well as cognitive impairment (Usui et al., 2021a). However, Zbtb16 KO mice showed antianxiety-like (risk-taking) behaviors (Usui et al., 2021a). Per2 is known as a clock gene of Period family and controls circadian rhythm under the CLOCK/ARNTL regulation (Takahashi, 2017;Rijo-Ferreira and Takahashi, 2019). PER2 is also associated with ASD (Iossifov et al., 2014;RK et al., 2017), and sleep problem has been reported to coexist in developmental disorders such as ASD (Couturier et al., 2005;Krakowiak et al., 2008;Souders et al., 2009;Mazurek and Sohl, 2016). Moreover, it has been reported that Per2 expression was altered by social defeat stress (Moravcová et al., 2021). Lrfn2 (SALM1) is known as a synapse adhesion molecule interacting with PSD-95 (Ko et al., 2006). LRFN2 is associated with ASD (Voineagu et al., 2011), learning disabilities (Thevenon et al., 2016), and antisocial personality disorder (Rautiainen et al., 2016). Lrfn2 KO mice also displayed ASD-like behaviors such as social withdrawal, decreased vocal communications, increased repetitive behaviors and prepulse inhibition deficits, but not anxiety-like behaviors (Morimura et al., 2017). Lrfn2 KO mice also displayed synaptic morphological abnormalities in the CA1 hippocampal neurons, resulting LTP enhancement (Morimura et al., 2017). Together, these findings suggest that Zbtb16, Per2, and Lrfn2 are the genes involved in sociality. We acknowledge that our study has several limitations. Since only male mice are used in this study, caution must be taken in interpreting the results of behavioral analysis. In addition, despite the analysis of the effects of environmental factors, the sample size in transcriptome analysis was small and the p value in gene expressions was not significant, thus only few genes were identified as DEGs. It is expected that increasing the sample size will lead to the identification of more genes affected by ELS. Furthermore, it is also important to understand the maternal gestational stress as a one of ELS, which is known to affect the brain development and behavior of the offspring. Previous studies have reported that maternal gestational stress has an epigenetic alteration on offspring gene expression as well as impairments of motor and cognitive development, immune system, and adaptation to stress (Wadhwa et al., 2001;Kingston et al., 2012;Argyraki et al., 2019).
In conclusion, our results demonstrate that ELS impacts adolescent social behaviors, anxiety-like behaviors, and PFC characteristics in mice. Our study suggests that ELS causes the alterations in the cytoarchitecture and transcriptome of PFC, leading to social impairments and increased anxiety-like behaviors in adolescents.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.