INTRODUCTION
The unprecedented coronavirus disease-2019 (COVID-19) pandemic has forced the world to close doors and maintain social and personal isolation. In particular, isolation by self-quarantine and physical distancing has adversely affected mental health conditions, such as depression and anxiety, which are expected to continue to have an impact even now despite the waning impact of COVID-19 [1-3]. According to the World Health Organization, depression and anxiety increased by 25% worldwide in the first year of the COVID-19 outbreak, with social isolation, loneliness, sadness, and economic problems contributing to unprecedented stress, mental health challenges, and exposure to existing health inequalities [4-6]. Mental health challenges attributed to COVID-19 are particularly pronounced among women [6], adolescents [7], and populations with medical comorbidities [7]. Such mental health problems appeared not only during the COVID-19 pandemic, but also during the Spanish flu and H5N1 pandemics [8,9]. Therefore, understanding the impact of social and personal isolation due to pandemics on mental health is crucial for managing and intervening in stress factors at both individual and societal levels.
Social isolation is a well-known cause of stress [10], which affects not only humans but also animals forming social groups, such as mice [1,11]. Social isolation is not only associated with depression and anxiety but is also known to exacerbate age-related cognitive deficits, hostility, impulse control, and sleep fragmentation. Severe social deprivation during early life is linked to various neurotransmitters and multiple profound, long-lasting neural structural changes, including corticotropin-releasing hormone (CRH), dopamine, norepinephrine, serotonin, glutamate, and gamma-aminobutyric acid (GABA). In rodents study, changes in neurotransmitters and hormones observed during social isolation were similarly observed in the experimental group that experienced maternal separation immediately after birth [1].
Stress has a strong relationship not only with mental health but also with medical diseases [12]. It is also associated with chronic diseases, such as diabetes, hypertension, and cancer. In particular, since stress and toxic substances present in the environment can affect somatic mutation, somatic mutation research has recently attracted attention in cancer research [13-15]. Currently, studying somatic mutations in the brains of individuals with psychiatric conditions other than Alzheimer’s disease, in which hypothalamic-pituitary-adrenal (HPA) axis-related factors are induced by stress, is warranted [16,17].
In this study, we aimed to examine the behavioral changes associated with social isolation in humans and mice. To assess the functional neural changes induced by social isolation, we observed the hippocampal protein expression in mice, and investigated potential genetic variations at the genetic level using single-cell RNA and whole-genome sequencing (WGS) of the hippocampus.
METHODS
Animals and experimental design
C57BL/6J male mice were purchased from DBL Co. (Eumseong, Korea) and provided with free access to food and water and four to five animals were housed per cage under a 12-hour light-dark cycle. We included only male mice in this study because demographic studies showed more negative mental health in males. All the mice were bred and maintained in accordance with the requirements of the Animal Research at National Center for Mental Health (NCMH), and all procedures and methods were approved by the Committee of Animal Research at NCMH (IACUC No: NCMH-2005-001-003-01).
C57BL/6J pregnant mice were supplied on the 12th day of time pregnant, acclimatized for about a week, and then used in the experiment. The maternal separation group (MS) was separated from the mother and other cages for 3 hours each for 3 weeks from the 2nd day after birth, and entire litters were separated from each mother without thermoregulation. The sex of the littermates was confirmed at the time of weaning, and only male mice were selected for the experiment. The social isolation group was isolated individually at 8 weeks of age for 2 weeks. Mice were randomly assigned to either the maternal separation or control groups (CTL), and after reaching adulthood, they were randomly assigned to the social isolation group.
Open-field test
Mice were placed in the center region of an open-field box (45×45×45 cm), and the locomotor activity in the open field arena was measured for the first 5 minutes. Behavioral tests were recorded and analyzed using SMART 3.0® tracking software (Panlab Harvard Apparatus, Barcelona, Spain).
Elevated plus-maze test
An elevated-plus maze made of gray acrylic with four arms, each 25-cm long and 2.5-cm wide, positioned 16-cm above the ground, was used for measuring anxiety-like behavior. Light intensity in the closed arms was approximately 0 lux. The test was initiated by placing the mouse in the center of the maze at the junction of the four arms and then allowing the mouse to freely explore the maze for 10 minutes.
Fear conditioning
Fear conditioning tests were performed for 2 days (Panlab Harvard Apparatus). On the first day, the mouse was placed in the center of the chamber and subjected to a foot shock of 7 mA for 1 second for fear-conditioning. At 24 hours after fear learning, the mice were placed in the same chamber without a foot shock for 5 minutes to measure contextual fear retrieval.
Western blots
The sample extracted from hippocampus of control and maternal separation mice. The cells were homogenized in lysis buffer containing 25 mM Tris·HCl pH7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), and protease inhibitors. The cell lysates (20 μg) were electrophoresis using SDS gels and transferred to nitrocellulose membranes and the incubated with tropomyosin receptor kinase B (TrkB) antibody (1:1,000, goat, AF1494; R&D systems, Minneapolis, MN, USA), corticotropin-releasing hormone receptor 1 (CRHR1) antibody (1:1,000, goat, GRX88961; GeneTex, Irvine, CA, USA), CRHR2 antibody (1:1,000, rabbit, AP17244PU-N; OriGene Technologies Inc., Rockville, MD, USA), Grin2B antibody (1:1,000, rabbit, 21920-1-AP; Proteintech, Rosemont, IL, USA), HTR1B antibody (1:1,000, rabbit, OAAF02797; Aviva Systems Biology, San Diego, CA, USA), doublecortin (DCX) antibody (1:1,000, rabbit, #4604; Cell Signaling Technology, Danvers, MA, USA), MAP2 antibody (1:1,000, rabbit, #4542; Cell Signaling Technology), SOX2 antibody (1:1,000, rabbit, #23064; Cell Signaling Technology), NeuN antibody (1:1,000, rabbit, #24307; Cell Signaling Technology), Erk antibody (1:1,000, rabbit, #4695; Cell Signaling Technology), p-Erk (thy202/204) antibody (1:1,000, rabbit, #9101; Cell Signaling Technology), mTOR antibody (1:1,000, rabbit, #2983; Cell Signaling Technology), p-mTOR antibody (1:1,000, rabbit, #5536; Cell Signaling Technology) and β-actin antibody (1:10,000, mouse, Sc-47778 HRP; Santa Cruz Biotechnology, Dallas, TX, USA) for 16 hours at 4°C. After washing with 1X TBST buffer, the blots were incubated with antigoat or anti-rabbit IgG (Thermo Fisher Scientific, Waltham, CA, USA). Band images were obtained by using a Molecular Imager ChemiDoc XRS+ (Bio-Rad, Hercules, CA, USA), and band intensity was analyzed using Image LabTM software version 2.0.1 (Bio-Rad).
Single-nucleus RNA sequencing
Briefly, frozen hippocampus of 3 mice in each sample group were dissociated using the Adult Brain Dissociation kit (Miltenyi Biotec, Bergisch Gladbach, Germany), and after dead cell removal, 10x Chromium Single Cell 3' Protocol (10x Genomics, Pleasanton, CA, USA) was used to prepare a cDNA library with a target of 10,000 cells per sample. And the prepared library was analyzed using an Illumina HiSeq X ten sequencing system (Illumina Inc, San Diego, CA, USA).
For bioinformatic analysis, raw base call files from the sequencer were generated into FASTQ files using “mkfastq,” and alignment, filtering, barcode counting, and unique molecular identifier counting were performed with “cellranger.” Seurat [18] 4.0.5 R package was used for the following analyses: data preprocessing, quality check, dimensionality reduction, graphbased cell clustering, the top expressed biomarker finding, gene-set enrichment (using gProfiler [19]), finding differentially expressed genes (DEGs) among groups, identifying subtypes of cell composition, and cell population analysis of single nucleus RNA sequencing data.
Brain dissociation, single nucleus RNA sequencing and bioinformatic analyses were performed by Macrogen (Seoul, Korea).
WGS
To investigate somatic mutations, we compared the genomes of the hippocampus (100× depth coverage) and the genome of the mouse heart (30× depth coverage).
Genomic DNA extracted from the heart and the hippocampus was used to construct a next-generation sequencing (NGS) library using a TruSeq DNA polymerase chain reaction-free kit (Illumina Inc). WGS was performed using the Illumina NovaSeq 6000 instrument. Thereafter, the bioinfomatic analysis continued through the following process: aligning using the GRCm39 reference (bwa-mem [20]), removing redundant reads based on align and coordinate (SortSam [21]), (MarkDuplicate [21]), analyzing variations (single nucleotide polymorphism [SNP], insertion and deletion [indel] [bcftools [22]], structure variant [Delly [23]]), copy number variation (CVNcaller [24]), and somatic mutation (VarScan2 [25]), and annotating each variation (SnpEff, ANNOVAR [26]).
Demographic data
This study was approved by Institutional Review Board of Inha University Hospital (2022-09-044); it was conducted between April and June 2022.
A web-based self-reported assessment was conducted for the self-quarantined community population who was not infected after contact with patients with COVID-19, and a webbased survey was conducted upon obtaining consent. Immediately after self-isolation by the public health center, a link to the self-report evaluation webpage was sent. The information obtained from the study participants included sex, age, marital status, education, residence type, degree of depression (Patient Health Questionnaire-9 [PHQ-9], cut-off value 10), degree of anxiety (General Anxiety Disorder-7 [GAD-7], cut-off value 5), somatic symptoms (Patient Health Questionnaire-15 [PHQ-15], cut-off value 10), suicidal ideation (P4 screener), stress perception (Perceived Stress Scale [PSS], cut-off value 14), post-traumatic stress symptoms (Korean version of the Impact of Event Scale [IES-K], cut-off value 33), sleep disorder symptoms (Korean version of the insomnia severity index, [ISI-K], cut-off value 8), and self-esteem status (Rosenberg Self-Esteem Scale [RSES], cut-off value under 15).
Quantification and statistical analysis
Statistical data analysis was performed using R (version 3.5.1; R Foundation for Statistical Computing, Vienna, Austria) and Prism 8 (GraphPad, San Diego, CA, USA). Data normality was determined using a Shapiro-Wilk normality test. Normally distributed data were analyzed using Student’s t-test and analysis of variance (ANOVA), followed by post-hoc tests. Data that failed the normality test were analyzed using the Mann-Whitney U test. The ROUT method was used to exclude outliers with a Q factor of 5%, considering the differences among wild-type mouse populations, such as resilience.
RESULTS
Demographic factors of self-quarantined individuals owing to COVID-19
Statistical analysis was conducted based on the demographic factors of the uninfected population who were self-quarantined owing to COVID-19 (Table 1). We compared the population according to sex, marital status, and residential status (Table 2). There were differences in depression (female [F] 3.2 vs. male [M] 4.2, p=0.002), anxiety (F 2.0 vs. M 2.6, p=0.02), somatic symptoms (F 5.3 vs. M 8.0, p<0.001), suicide (F 2.5 vs. M 2.2, p=0.004), sleep (F 6.2 vs. M 7.6, p=0.003), and stress perception (F 14.8 vs. M 15.9, p=0.009), and post-traumatic stress disorder (PTSD) symptoms (F 8.3 vs. M 10.5, p=0.04). Stress perception showed scores above clinical significance in all participants.
Marital status was divided into single (single, separated, and divorced) and married (married and cohabitating). There were differences in the age (single [S] 32.2 vs. married [M] 45.0 years, p<0.001), year of education (S 13.8 vs. M 14.5, p=0.008), depression (S 4.2 vs. M 3.4, p=0.02), suicide (S 2.1 vs. M 2.5, p=0.001), and self-esteem (S 30.8 vs. M 32.0, p=0.006).
According to the residence status, the status of living alone (A) and living together (T) (including both family and nonfamily) were compared. There were differences in the age (A 35.6 vs. T 40.7 years, p<0.001), year of education (A 14.6 vs. T 14.1, p=0.03), sleep (A 8.1 vs. T 6.6, p=0.009), and self-esteem (A 30.6 vs. T 31.8, p=0.03). There were no statistical differences between variables such as sex and marriage, sex and residence, and marriage and residence. Scores for sleep were above clinical significance in participants living alone.
Overall, male, being single, and living alone tended to have negative mental health outcomes in self-quarantined people; we further investigated the effects of social isolation on the brain and mental health through animal and genomic studies.
Behavioral and protein expression measurements to investigate the effects of early maternal separation and social isolation in mice
Based on the demographic results, we conducted an isolation test on mice to evaluate the biological mechanism of the effect of social isolation, and early maternal separation was used to hypothesize the factors for individual vulnerability along with social isolation in adult mice (Figure 1 and Supplementary Figure 1).
Locomotion observed through the open field test did not demonstrate a difference between the control and MS (Figure 1A). In the elevated plus maze test to measure anxiety behavior (Figure 1C), the time spent in the open arms in the MS increased compared to the CTL. In the fear conditioning test using foot shock (Figure 1D and Supplementary Figure 1D), the MS demonstrated increased freezing compared to the CTL, and the sound cue-induced freezing behavior was different from that of the CTL (control vs. control following social isolation) and between the control and maternal separation groups after social isolation (Figure 1E and Supplementary Figure 1E).
We observed the effects of maternal separation and social isolation on the expression of proteins in the hippocampi of the mice (Figure 1F and G). Immunoblotting was performed, focusing on the HPA axis and neurogenesis abnormalities caused by stress. CRHR1 expression was increased in the MS compared to that in the CTL, and maternal separation and social isolation demonstrated an increased pattern compared to that in the CTL. CRHR2 expression was higher in the maternal separation and social isolation group than in the CTL. There was no difference between the groups in the TrkB that brain derived neurotrophic factor binds to; SOX2, a neural stem cell marker, and DCX, a neuro progenitor cell marker, were increased in the MS group compared to the CTL. Except for MAP2, there were no differences in the major downstream signaling pathways and associated proteins of CRHR1, such as GRIN2B, mTOR, p-mTOR, ERK, and p-ERK (Supplementary Figure 2).
Single-nucleus RNA sequencing was performed to analyze single-cell transcriptome changes in the hippocampus of mice subjected to maternal separation and social isolation
We performed genomic analysis to identify single-cell-level transcriptome changes in the mouse brain, specifically in the hippocampus, which is directly affected by stressors, such as social isolation and maternal separation. Single-nucleus RNA sequencing (SnRNAseq) analysis was performed to observe somatic mutations and transcriptional changes resulting from the effect of stress in each cell.
snRNAseq and analysis were performed on the CTL, MS, and social isolation after maternal separation group (SS) (three mice in each group). Each group showed a similar cell distribution (Supplementary Figure 3A and B). In the CTL; neurons accounted for 92.1%, astrocytes for 4.9%, and oligodendrocytes for 2.3% (Figure 2A, B and Supplementary Figure 3C). Each cluster was classified according to the transcript expression (Supplementary Figure 3D and E).
Differences in DEGs analyses were compared by group, and the expression of transthyretin (Ttr) increased in the MS compared to that in the CTL; increased expression was observed in all clusters, including neurons, astrocytes, and oligodendrocytes (Figure 2C and H). Between MS and SS, an increase in Ttr was observed in the MS (Figure 2D). In the SS compared to the CTL, decreases in Dipeptidyl peptidase 10 (Dpp10), Netrin G1 (Ntng1), and Protocadherin 15 (Pchd15) were observed (Figure 2E-G).
Somatic mutations revealed by WGS in the hippocampus of mice with maternal separation and social isolation
We observed stress-induced DNA-level somatic mutations in the hippocampi of mice subjected to stressors, such as maternal separation and social isolation. WGS was performed in the heart of each mouse at approximately 30× depth and hippocampus at approximately 100× depth. In addition, the WGS results for the heart and hippocampus were compared, and the results for each individual were compared according to the group.
First, in the WGS results of the heart and hippocampus, a difference in the total variants, including insertions, deletions, and SNPs was observed; however, no difference was observed in each group (Figure 3A-C). This could be attributed to the differences in the organs; however, the effect of sequencing sensitivity according to sequencing depth coverage should also be considered.
Although the genome-wide significance p-value threshold of 5×10-8 has become a standard for common-variant genomewide association studies for lower frequency variants, we interpreted our results for 1×10-9 or higher as significant for more stringent statistics. Among the SNPs (Figure 3D), a significant difference was observed in the chromosome 9 3025779 intergenic position (T → A) between the MS and SS. In indel, compared to the CTL, the intergenic somatic mutation at the chr Y 90795686 position (CCG insertion) and the somatic mutation in the intronic region of the Tmem267 gene at the chr 13 120069455 position (CTTTAAG insertion) demonstrated significant differences in SS. Between the MS and SS, the above chr Y 90795686 and Tmem267 intronic somatic mutations appeared in the SS compared to the MS, and loss of heterogeneity of the intergenic region at chr 15 74958254 (C deletion) was observed in the MS compared to SS. Somatic mutation profiles for each mouse are shown in Supplementary Table 1.
Using WGS, somatic mutations in the MS and SS were confirmed and compared with those in the CTL. In particular, Tmem267 revealed somatic mutations in the intron region that could directly affect the transcriptome. Intergenic somatic mutations were also observed.
DISCUSSION
In this study, socially isolated groups of individuals who self-quarantined owing to COVID-19 exposure demonstrated worse results in terms of depression, anxiety, and mental health measurements. In addition, anxiety-related behaviors increased in mice subjected to early maternal separation and social isolation. The protein expression of CRHR1, a stressrelated CRH receptor, and neurogenesis-related SOX2 and DCX increased in the hippocampi of socially isolated mice. SnRNAseq revealed that mice subjected to early maternal separation and social isolation showed transcriptome changes, including increased Ttr expression and decreased Dpp10, Ntng1, and Pchd15 expression. Finally, multiple somatic mutations, including the Tmem267 intronic mutation in the MS+SS, were observed in the hippocampi of the mice.
Social isolation, one of the major stressors in this study, is related to the deterioration of human mental health, such as depression and anxiety. Animal experiments confirmed changes in anxiety behavior and stress-related proteins in the hippocampi following social isolation. In addition, somatic mutations through WGS and transcriptome changes upon singlenucleus RNA sequencing could be induced by isolation stress.
In a study conducted on somatic mutations in the human brain related to psychiatric disorders [27-29], 131 postmortem brains of patients with schizophrenia, Tourette’s syndrome, autism spectrum disorder, and a healthy control group were analyzed using WGS at 200× coverage [27]. A total of 20-60 single nucleotide variants were observed, and more than 101 hypermutable brains were distributed according to age rather than disease classification. In a study using whole exome sequencing (WES) of 500× or more coverage of postmortem brains in the Alzheimer’s disease and healthy control groups, no difference was observed between the groups; however, an increase in the mutations with age was observed, and 22 somatic mutations per year were predicted in the hippocampal formation [30] (40 somatic mutations per year were predicted in the dentate gyrus in the healthy control group only in the study) [31]. In a study of patients with schizophrenia known to have a heritability of approximately 80% [28], eight genetic mutations of a total of 106 somatic deletions were reported in three patients compared to two healthy controls [32]. Another study in patients with schizophrenia demonstrated an increase in long interspersed nuclear element-1, which is known to be increased in Rett syndrome [33]. In addition, in a recent WES study, no difference was observed in the number of somatic mutations between the schizophrenia patient group and control group; however, mutations in the SCZ-associated pathways, including GRIN2B, were reported to be more prevalent in the schizophrenia group [34]. In this study, similar to previous studies, no difference was observed in the number of somatic mutations in the hippocampi of the MS, MS+SS, and CTL mice.
In particular, among results of WGS for somatic mutation, the Tmem267 intron mutation was observed in the MS+SS. The decrease in Tmem267 is reported to be related to the increase in miR-142a-3p, a microRNA, in the ventral tegmental area of mice subjected to chronic mild stress [35]. Transcriptomic findings from medial prefrontal cortex showed increased Tmem267 in a mouse model of neuropathic pain-induced depression [36]. In cancer studies, the 3’ UTR shortening of Tmem267 by loss of NUDT1 in hepatocellular carcinoma cells has been reported to increase cell proliferation [37], and genomic results from a tongue cancer cohort show that Tmem267 expression is a good prognostic factor [38].
Among snRNAseq results, Ttr, which can inhibit Aβ fibril formation, is reported to be reduced in people with Alzheimer’s disease and is suggested to be present in the cerebrospinal fluid rather than in the brain parenchyma [39-41]. Furthermore, A study of Ttr knockout mice showed decreased depressive behavior associated with increased norepinephrine, which may explain the increased Ttr in the MS+SS group in our study. And Ntng1 or Netrin-G1, which is known to have a function to promote the growth of thalamocortical axons, showed reduced mRNA expression in the temporal lobe of people with schizophrenia and bipolar disorder [42-44]. Taken together, the results observed in our study, such as changes in Tmem267, Ttr and Ntng1, have implications for psychiatric disorders that can be induced by stress including social isolation.
In a study of stress-related somatic mutations [45-47], an RNAseq study using the blood of soldiers with PTSD demonstrated 187 and 442 cases of high and moderate severity of transcriptional abnormalities, respectively, including inflammationrelated genes [48]. Prenatal stress results in oxidative damage of the mitochondrial DNA in the hippocampus of newborn mice [49]. Post-mortem brain studies of 63 psychiatric conditions, including schizophrenia, major depressive disorder, and bipolar affective disorder, have also demonstrated mitochondrial DNA damage [50].
As described above, inflammation and oxidative stress affect the mechanisms underlying stress-related somatic mutations [51-53]. Studies have been conducted on the possibility of genetic damage attributed to inflammatory and oxidative damage caused by environmental factors, including stress in neurons, which are the main cells of the central nervous system having a very low replication rate [49,51,53]. Among these, social isolation is a well-known cause of stress, and previous studies have confirmed evidence that early-life experiences of social isolation affect central nervous system function, even in animals that form social groups, such as mice [1]. Based on this evidence, we confirmed that experiences of social isolation are associated with hippocampal protein variations related to anxiety and stress. Several studies have reported genetic variants in mental disorders, such as schizophrenia, major depressive disorder, and bipolar affective disorder, that first appear in early adulthood. However, the results are inconsistent, and genetic findings are spread across multiple diagnoses rather than a single diagnosis [28,54,55]. This could be related to the underlying mechanisms by which somatic mutations occur randomly at various developmental stages, such as increased mutations in long genes and active transcription [52,54,56,57]. In our study, we were able to observe both somatic mutations through WGS and transcriptomic changes identified by sn-RNAseq in mice that experienced early maternal separation and social isolation. In the future, research with more emphasis on the genetic changes caused by environmental factors, especially stress, is warranted.
Individual variations in stress in mice were identified upon analyzing of the results of all the experiments. Therefore, to observe a clearer difference, molecular and genetic analyses should be performed after selecting mice more susceptible to stress following behavioral experiments in future studies. In humans, it was possible to compare sociodemographic and clinical characteristics by gender; however, in the mouse study, only male mice were included based on evidence of greater vulnerability to mental health issues. This lack of information could be a limitation of the study results. Future research, including both male and female mice, is necessary to examine whether stress factors caused by social isolation could induce somatic and genomic variations, regardless of gender. In addition, this study only covered 100-fold coverage in WGS. Ideally, a coverage of 200-fold or higher is considered more sensitive to somatic mutations; thus, higher coverage is warranted. In addition, since only demographic data existed in the human data in this study, the results of this study should be reconfirmed if blood and hippocampus samples of the participants could be obtained and analyzed from a well-established biobank with clinical information, such as stress measurements.
In conclusion, the human study was based on demographic data, including depression, anxiety, and other mental health issues, among individuals self-quarantined owing to COVID-19 exposure. The study findings are valuable since the study yielded social experimental data obtained in a situation where individuals were artificially isolated because of COVID-19. In addition, early maternal separation and social isolation were objectively replicated through animal experiments, and changes in stress were observed at various levels, such as behavioral experiments, protein expression, snRNAseq, and WGS. Somatic mutations resulting from maternal separation and social isolation have also been identified. In this study, changes in mental health, such as depression and anxiety, in the general public owing to social isolation during the COVID-19 pandemic and the underlying mechanism were explored through animal experiments, snRNAseq of the hippocampus, and 100× WGS. This study provides evidence that genetic-level changes, such as somatic mutations, could be attributed to stress such as social isolation.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
Print
