Sex-dependent association of DNA methylation in the coding region of the corticotropin-releasing hormone gene and schizophrenia spectrum disorder

Abstract Background Schizophrenia spectrum disorder (SSD) is a common mental disorder causing severe and chronic disability. Epigenetic changes in genes related to the hypothalamic–pituitary–adrenal (HPA) axis are believed to play an important role in SSD pathogenesis. The methylation status of the corticotropin-releasing hormone (CRH) gene, which is central to the HPA axis, has not been investigated in patients with SSD. Aim We investigated the methylation status of the coding region of the CRH gene (hereafter, CRH methylation) using peripheral blood samples from patients with SSD. Subjects and methods We used sodium bisulphite and MethylTarget to determine CRH methylation after collecting peripheral blood samples from 70 patients with SSD who had positive symptoms and 68 healthy controls. Results CRH methylation was significantly increased in patients with SSD, especially in male patients. Conclusions Differences in CRH methylation were detectable in the peripheral blood of patients with SSD. Epigenetic abnormalities in the CRH gene were closely related to positive symptoms of SSD, suggesting that epigenetic processes may mediate the pathophysiology of SSD.


Introduction
Schizophrenia spectrum disorder (SSD) is a common mental illness characterised by severe, chronic, and disabling symptoms that impair emotion regulation, cognitive function, and volitional behaviour (Zhu et al. 2019); approximately 1% of the global population has SSD (McGrath et al. 2008).Patients with SSD exhibit dysfunction that increases the risk of aggression, suicidal behaviour, and unemployment; such patients often experience social rejection and discrimination.Schizophrenia imposes a heavy economic burden (in terms of direct/indirect health care costs and direct non-health care costs), not only on a patient's family but also on society (Kadakia et al. 2022).The aetiology of schizophrenia is complex, and disease development is affected by genetic (accounting for 80% of variance) and epigenetic factors (environmental stressors such as place of birth, cannabis use, and obstetric complications).Patients show general volume reduction of gray matter and the hippocampus, reduced oligodendrocytes, and decreased synaptic plasticity (Nishioka et al. 2012;Falkai et al. 2015;Misiak et al. 2018).In addition, SSD frequently has only a modest treatment outcome and relapse is common; the risk of relapse is associated with sex, age, and marital status (Davarinejad et al. 2021).Although there are many studies on the aetiology of schizophrenia, its exact mechanism is still unclear.
Mental disorders are closely associated with stress responses in the brain (Palma-Gudiel et al. 2015;Watkeys et al. 2018).In many organisms, appropriate responses to stress are the basis of survival and adaptation (de Kloet et al. 2005), while inappropriate and long-term stress responses may lead to the occurrence of mental disorders, including SSD and bipolar disorder (Sinclair et al. 2012).The hypothalamic-pituitary-adrenal (HPA) axis is an important stress response system that maintains homeostasis in response to environmental stressors (Russo et al. 2012).As shown in our previous studies, DNA methylation of the NR3C1 and NR3C2 genes associated with the HPA axis increases the risks of mental illness, including SSD and aggressive behaviour (Liu, Wu, et al. 2020;Liu, Li, et al. 2020;Qing et al. 2020;Qing et al. 2021).Given its importance to the HPA axis (Zhou and Fang 2018), we focussed on the corticotropin-releasing hormone (CRH) gene.
CRH plays a key role in the regulation of neuroendocrine, autonomic, and behavioural adaptations to stress (de Kloet et al. 2005;Ishiwata et al. 2020).The CRH gene consists of two exons and one intron, and the coding region is located in the second exon (Figure 1) (Cramer et al. 2019).CRH is mainly expressed by the parvocellular neurons of the hypothalamic paraventricular nucleus (PVN), which drives the activation of the HPA axis, thus regulating the stress response (Zhou and Fang 2018;Ishiwata et al. 2020).Elevated CRH levels in the central nervous system are associated with stress-related physiological and behavioural disorders (Linthorst et al. 1997).Numerous studies have confirmed that CRH levels are upregulated in patients with depressive disorders and those who have attempted suicide and that CRH concentrations return to normal after recovery from depression (Naughton et al. 2014).CRH concentration is also increased in individuals with post-traumatic stress disorder (Sautter et al. 2003).CRH also plays an important role in anxiety (Dedic et al. 2018), addiction (Carboni et al. 2018), and sleep (Kimura et al. 2010).The most recent study on this topic showed that the CRH concentration in the cerebrospinal fluid (CSF) of patients with SSD was lower than that in subjects in the control group and that a lower CRH concentration was associated with negative symptoms of SSD (Ishiwata et al. 2020).This evidence suggests that CRH expression is complex and may be related to the pathogenesis of SSD and other mental disorders.
DNA methylation mediates gene expression in response to environmental factors and thus plays a particularly important role in mental disorders.Studies of methylation of the CRH gene (hereafter, CRH methylation) support this hypothesis.Under conditions of prenatal stress (Mueller and Bale 2008;Xu et al. 2014), chronic variable mild stress (CVMS) (Sterrenburg et al. 2011), social stress (Elliott et al. 2010), and lack of maternal affection (Chen et al. 2012), CRH methylation decreases and CRH expression increases, leading to the development of corresponding mental disorders or symptoms.Two CRH-related CpG sites (cg19035496 and cg23409074) are significantly demethylated in patients who have attempted suicide.In subsequent adolescent cohort studies, the cg19035496 site was shown to be hypermethylated in subjects with higher general psychiatric risk scores (Jokinen et al. 2018).These results suggest that CRH methylation may be involved in the pathogenesis of mental disorders.
Previous studies on CRH methylation have focussed on the promoter region (Mueller and Bale 2008;Elliott et al. 2010;Sterrenburg et al. 2011;Chen et al. 2012;Xu et al. 2014;Jokinen et al. 2018).With ongoing, increasingly in-depth research, some CpG sites have been identified in exons or coding regions; these sites potentially affect promoter activity by modulating transcriptional elongation.In particular, DNA methylation of the coding sequence (CDS) is more effective than that of the promoter in inhibiting gene expression (Hisano et al. 2003;Zhu et al. 2005).For example, CpG methylation in the MCT3 coding region inhibits its expression (Zhu et al. 2005).Testicular germ cell-specific demethylation of the CDSs of the TACT1/actl7b and PDHA2 genes inhibits their expression (Hisano et al. 2003;Pinheiro et al. 2012).Methylation of +10 and +88 CpG sites in the CDS of the MAT1A gene reduces promoter activity by 60% (Tomasi et al. 2012).Based on these results, methylation of CpG sites or CpG islands in the CDS is highly important for the regulation of gene expression.CRH methylation in the CDS of SSD patients has not previously been analysed.The objectives of this study were to determine the methylation status of CpG islands in the CDS of the CRH gene in the SSD population and to explore the relationship between CRH methylation (in the CDS) and SSD.  1. Patients with SSD who were admitted to the Psychiatric Hospital of Yunnan from December 2012 to August 2013 and agreed to participate were included.All patients were diagnosed by at least two professional attending psychiatrists according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV), and clinical records.The healthy controls were unrelated individuals from the Centre for Disease Prevention and Control of Kunming.They were recruited and interviewed by psychiatrists and were selected according to the following inclusion criteria: healthy individuals with no history of mental disorders or brain injury who matched the study group in age, sex, and body mass index (BMI).The exclusion criteria were as follows: subjects (or their close relatives) with a history of mental disorders according to DSM-IV diagnostic criteria, severe physiological diseases, a history of craniocerebral trauma and epilepsy, or dependence on mental substances or other addictive substances (Qing et al. 2020).The study was approved by the Ethics Committee of Kunming Medical University.All subjects signed informed consent forms and donated peripheral blood.

DNA sample preparation and bisulphite modification
Blood samples were collected from all patients with SSD and the controls.Genomic DNA was isolated from whole-blood samples using a QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany).Genomic DNA integrity was evaluated using agarose gel electrophoresis.Genomic DNA was quantified with a NanoDrop 2000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA), and the absorbance ratios at 260/280 nm and 260/230 nm were determined to ensure DNA purity.The samples were treated with sodium bisulphite and processed using the EZ DNA Methylation Gold Kit (Zymo, CA, USA) according to the manufacturer's instructions.Cytosine (C) residues without genomic DNA methylation were transformed into uracil (U) residues.
DNA methylation levels were analysed using MethylTarget® (Genesky Biotechnologies Inc., Shanghai, China) based on next-generation sequencing (NGS).An optimised multiplex PCR primer panel was used to amplify the target fragment from the sample with the sulphite-transformed sample genome as the template.After the application of quality control procedures, the amplification products of all multiplex PCR primers in the panel from the same genomic DNA sample template were combined; an equal quantity of amplification products was produced from each primer.Primers with index sequences were used to amplify the specific tag sequences compatible with the Illumina platform.All samples were mixed with the same amount of index PCR products, and the final MethylTarget library was obtained by tapping and recycling.The fragment length distribution of the library was verified with an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA).After the accurate quantification of library molarity, high-throughput sequencing was performed using the Illumina MiSeq (Illumina, CA, USA) platform in 2 × 150-bp double-ended sequencing mode to obtain FASTQ data.

Statistical analyses
FastQC software was used for quality control of the sequencing data.The FASTX Toolkit (http://hannonlab.cshl.edu/fastx_toolkit/index.html) was used for the FASTQ data formatting steps.FASTA format reads were mapped to the target bisulphite genome (hg19) using BLAST (Camacho et al. 2009).The unmapped reads were filtered, and the mapping reads with coverage greater than 90% and identification greater than 95% were reserved as valid reads for subsequent analysis.The sequencing depth of each amplicon of each sample was calculated by effectively reading the target genome region.Reads appearing fewer than 10 times were removed, and the overall sequencing depth of each sample was evaluated.The conversion efficiency of bisulphite was determined by calculating the percentage of C residues outside of CpG sites; a conversion efficiency of over 99% was required (Merkel et al. 2019).The methylation analysis was conducted with Perl scripts.T-tests were used for statistical analyses, and differences with p values <.05 were considered statistically significant.All statistical analyses were performed using SPSS 22.0; all graphical analyses were performed using GraphPad Prism version 8.

Risk factors for schizophrenia
In this study, 68 healthy controls (47 men and 21 women) and 70 patients with SSD (46 men and 24 women) were included in the analysis; all 70 patients with SSD presented positive symptoms.A logistic regression analysis was first performed to assess the environmental factors associated with schizophrenia.The environmental factors included educational level (low: middle school or below; high: high school or above), marital status (single or married), and smoking status.The risk of schizophrenia was higher in individuals with a low educational level than in individuals with a high educational level (p = .022).Individuals who were single had a higher risk of developing schizophrenia than those who were married (p = .002).A significant association between smoking status and SSD was not identified (p = .914),as shown in Table 2.

Overall analysis of CpG island methylation
In this study, the average methylation sequencing depth was 1172×, and the percentage of samples with a sequencing depth greater than 400× was 91.98%.The sequencing depth and quality of all samples met the standards required for subsequent analysis.The mean sulphite conversion efficiency was 99.28%.Analysis of the DNA methylation status of CpG islands in the CDS of the CRH gene showed that compared with the control group (mean = 0.0204, SD = 0.0062), methylation was significantly increased in the SSD group (mean = 0.0242, SD = 0.0105) (p = .0092).When the subgroups were further divided according to sex, this significant difference was maintained in the male SSD subgroup (p = .0105).Although a trend of hypermethylation was observed in the female SSD subgroup, it was not significant (p = .4632)(Figure 2 and Table S1).

Analysis of methylation at CpG sites
Seventeen CpG sites were detected in the CpG island (Figure 1 and Table S2).The methylation levels of all CpG sites were less than 6% in both the SSD group and the control group.
For individual CpG sites, we also grouped the results by sex and compared the DNA methylation levels of all CpG sites between the sexes.Compared with the control group, the methylation levels of 6 CpG sites (CpG9, p = .0083;CpG11, p = .0163;CpG12, p = .0233;CpG14, p = .0154;CpG16, p = .0081;and CpG17, p = .0031)were significantly higher in the male SSD subgroup than in the male control group, while the CpG3 methylation level was not significantly different (Figure 3(B) and Table S2).Surprisingly, compared with the female control group, the female SSD group exhibited a significant difference in methylation at only the CpG4 site (p = .0383);no significant differences were found at the other sites (Figure 3(C) and Table S2).
In addition, we compared and analysed the methylation levels of all male (n = 93) and female (n = 45) samples.No significant differences in the methylation levels of whole fragments or single CpG sites were observed (Figure 3(D)).

Discussion
Numerous studies have confirmed that CRH drives HPA axis activity (Zhou and Fang 2018) and therefore plays a particularly important role in the stress response.CRH promoter methylation can cause excessive activation of the HPA axis by regulating CRH gene and protein expression and thus participates in the pathogenesis of SSD and other mental disorders (Mueller and Bale 2008;Elliott et al. 2010;Sterrenburg et al. 2011;Chen et al. 2012;Xu et al. 2014;Jokinen et al. 2018;Ishiwata et al. 2020).Notably, methylation of the CDS (CDS methylation) is more effective than methylation of the promoter at inhibiting gene expression (Hisano et al. 2003;Zhu et al. 2005).However, CDS methylation of the CRH gene in SSD patients has not been previously analysed.This study is the first to examine the CDS methylation status of the CRH gene in the peripheral blood of patients with SSD; the present study aimed to determine the association of CDS methylation of the CRH gene and SSD.As mentioned above, we selected CpG islands located in the CDS of the CRH gene, including 17 CpG sites (Figure 1), and analysed them using sodium bisulphite treatment combined with the MethylTarget method.CDS methylation of the CRH gene was significantly increased in patients with SSD (Figure 2 and

S1
).After division into subgroups according to sex, we found that this increase was only significant in the male SSD subgroup.Among the individual CpG sites, 7 (CpG3, CpG9, CpG11, CpG12, CpG14, CpG16, and CpG17) were significantly associated with SSD accompanied by positive symptoms.When the samples were divided into male and female subgroups, the CpG4 site showed significant differences in methylation in only the female subgroup, while the other 6 CpG sites showed significant differences in only the male subgroup (Figure 3 and Table S2).
DNA methylation is an important link between environmental factors and genetic background, and differences in DNA methylation may lead to the development of SSD and other mental disorders (Popov et al. 2012;Gürel et al. 2020).Previous studies have focussed mainly on promoter methylation and less on CDS methylation.This study confirmed that the CDS methylation level of the CRH gene is associated with SSD.This phenomenon has been observed not only in human and mouse studies (Hisano et al. 2003;Zhu et al. 2005;Pinheiro et al. 2012;Tomasi et al. 2012) but also in studies on fungi, viruses, and cells.In fungi, the effect of CDS methylation on transcript extension is greater than that on transcription initiation (Barry et al. 1993;Rountree and Selker 1997).CDS methylation of the herpes simplex virus TK gene has been shown to inhibit TK expression (Graessmann et al. 1994).At the cellular level, Hsieh (Hsieh 1997) and Irvine (Irvine et al. 2002) documented that CDS methylation of the luciferase gene inhibited transcription with an efficacy 5 times greater than methylation of the long terminal repeat promoter of Rous sarcoma virus in 293 cells.CDS methylation may control gene expression by inducing local inhibitory chromatin structures via CpG methylation and thereby preventing promoter activity (Pinheiro et al. 2012).Some studies have suggested that CDS methylation can initiate chromatin structure formation, reduce the efficiency of Pol II extension, and lead to transcriptional silencing (lorincz et al. 2004).Although the mechanism by which CDS methylation affects gene expression is not completely clear, CDS methylation plays an important role in gene expression and merits greater attention.
Most past evidence has indicated that CRH methylation decreases in response to prenatal stress and the lack of maternal affection, leading to upregulation of CRH gene or protein expression, which drives the activation of the HPA axis and eventually leads to mental or stress disorders (Mueller and Bale, 2008;Elliott et al. 2010;Sterrenburg et al. 2011;Chen et al. 2012;Xu et al. 2014).However, we observed a higher methylation level in the SSD group than in the control group, suggesting that CRH expression might be lower in the SSD group.Similarly, a recent study reported that the CRH concentration in the CSF of patients with SSD was lower than that in the control group, and a lower CRH concentration was associated with negative symptoms of SSD (Ishiwata et al. 2020).Additionally, in an adolescent cohort study, the cg19035496 site in the CRH gene was hypermethylated in subjects with higher psychiatric risk scores (Jokinen et al. 2018).Previous studies have reported inconsistent results regarding the expression of CRH in patients with SSD.The methylation and expression of the CRH gene are inconsistent among patients with SSD and other mental disorders or stress disorders, possibly because the pathogenesis of mental disorders is complex and CRH plays different roles in each disorder.
In the present study, we found that the CDS methylation of the CRH gene was associated with positive symptoms in patients with SSD.Notably, this association was specific to males.Previous studies have reported sex differences in SSD, males more likely to be diagnosed and develop cognitive impairment than females (Gershon 2002;Mendrek and Mancini-Marïe 2016).Sex differences in CRH function have also been reported in previous studies, although the results are inconsistent.After adrenalectomy (ADX), the potent androgen dihydrotestosterone promotes a rapid change in the level of CRH mRNA in male rats (Heck and Handa 2019).Male mice are more susceptible than female mice to memory impairment caused by CRH in the medial septum (Wiersielis et al. 2019).Negative feedback from PVN-expressed CRH, mediated by the glucocorticoid receptor (GR), shows significant sex differences, especially in female mice (Heck et al. 2020).Thus, the effects of CRH on stress responses and mental disorders may differ between the sexes.We speculate that data on sex differences in DNA methylation may enhance understanding of sex differences in CRH gene expression, HPA axis activation and the pathogenesis of SSD.
Our research has some limitations.The sample size was relatively small, and the sample consisted of only individuals of Han Chinese ethnicity.Further research on different groups and larger sample sizes is needed to confirm this finding.The number of females was also far lower than that of males.Future studies should recruit a sample of a balanced sex ratio to verify sex differences in DNA methylation.Similar to most previous studies on human methylation, this study used DNA extracted from peripheral blood for methylation analysis.Some studies have found that CRH methylation detected in peripheral blood reflects methylation in the central nervous system and brain (Jokinen et al. 2018).Future research should directly compare the methylation levels in brain tissue and peripheral blood in post-mortem samples to verify this hypothesis.Many external factors affect methylation levels, but in this study, educational level, marital status and smoking status had no significant impact on methylation levels (Table S3).No participants had a history of mental illness or a family history of mental illness and none used sedatives or psychotropic drugs in the past 6 months.Therefore, we assume that these factors did not affect the results of this study.We did not investigate exposure to environmental pollutants or trauma in childhood or adulthood, and the collection of participant clinical information was not perfect.However, to prevent the interference of external factors on methylation results, future studies should recruit similar samples.Additionally, gene hypermethylation is often associated with decreased expression; we used the JASPAR database (http://jaspar.genereg.net/;profile score threshold of 80% and score > 5) to predict whether transcription factors could bind to these CpG sites.We found that TFAP2A, PAX5, KLF4, and NHLH1 could bind to these CpG sites (Table S4), but the correlation between CDS methylation of the CRH gene and its expression was not analysed in this study; this topic merits further research.Future studies should include functional experiments to clarify how CDS methylation of the CRH gene affects the occurrence and development of SSD.
Our previous studies on the methylation of the NR3C1 and NR3C2 genes (associated with the HPA axis) in patients with SSD have yielded very significant results.The methylation status of the NR3C1 and NR3C2 genes is significantly associated with SSD and is sex-dependent to some extent (Liu, Wu, et al. 2020;Liu, Li, et al. 2020;Qing et al. 2020;Qing et al. 2021).Based on the results of the present study, we strongly believe that methylation of genes in the HPA axis mediates gene-environment interactions and affects the pathogenesis of SSD.
In conclusion, this study revealed differences in the CDS methylation of the CRH gene in the peripheral blood of patients with SSD who presented positive symptoms.Although the functional significance of this methylation difference is not clear, epigenetic abnormalities in the CRH gene are presumed to be related to the pathophysiology of SSD.This study is the first to show an association between SSD and CpG island methylation in the coding region of the CRH gene in peripheral blood.These results improve our understanding of the role of the HPA axis in the pathological development of mental disorders (especially SSD) and address a gap in research on CDS methylation of the CRH gene in patients with SSD.Based on our results, epigenetic processes may mediate psychopathology by influencing HPA axis activity.The effects of differences in CRH methylation on gene expression and function require further study and may help to clarify the pathophysiology of SSD.

Figure 1 .
Figure 1.schematic diagram of Cpg islands in the CDs of CRH.upper panel: CRH gene structure.The CRH gene is composed of 2 coding exons.The coding region is completely located within the second exon and is represented by the dotted area.The areas filled with diagonal lines represent the regions sequenced in this study.lower panel: The box surrounded by a dotted line contains the genomic sequences analysed to identify Cpg islands.The examined Cpg loci are underlined, presented in bold and numbered (nCBi reference sequence: ng_016127.1).

Figure 2 .
Figure 2. DnA methylation of the CDs of CRH in patients with ssD and matched healthy controls.Comparisons of the methylation status (mean ± sD) among all samples (left panel) and the male (middle panel) and female (right panel) subgroups (when samples were separated according to sex).*p < .05 and **p < .01.

Figure 3 .
Figure 3.Comparison of Cpg site methylation status between patients with ssD and matched healthy controls.methylation status (mean) of all samples (A) and male (B) and female (C) subgroups (when samples were separated according to sex).(D) methylation status (mean) of men and women.*p < .05. *p < .05,and **p < .01.

Table 1 .
Demographic characteristics of participants., and c represent the p values derived by student t test comparisons of male controls vs. male patients with ssD, female controls vs. female patients with ssD, and controls vs. patients with ssD, respectively.

Table 2 .
Differences in environmental factors between controls and patients with ssD.
Education level: low: middle school and below; high: high school and above.