Introduction

Sex is a major risk factor for many neurological diseases and disorders, including Alzheimer’s disease (AD) [1, 2], multiple sclerosis [3, 4], autism [5], attention-deficit/hyperactivity disorder (ADHD) [6], depression [7], and age-related cognitive decline [8, 9]. Of particular importance in modulating the cognitive effects seen in many sex-biased diseases is the hippocampus, the learning center of the brain [10]. Females tend to outperform males on hippocampal-dependent learning tasks and are more impacted by many diseases/disorders of hippocampal dysfunction (i.e. AD, depression) [11]. Understanding basal sex differences and their regulation in the hippocampus can help gain insight into the etiology of sex differences in hippocampal dysfunction in common neurological diseases. The goal of this study is to separate the effects of gonadal and chromosomal sex on the sex differential regulation of mouse hippocampal gene expression through epigenomic mechanisms.

Sex characteristics can be influenced by genotypic sex, gonadal sex, and gender. Generally, in mammals, a genotypic female has two X-chromosomes (and no Y-chromosome), while a genotypic male has one X-chromosome and one Y-chromosome. Gonadal sex is classified based on the individual’s genitalia, which can influence expression of secondary sex characteristics [12]. Gonadal sex determination is driven by the presence or absence of the Y-chromosome encoded sex-determining region of Y (Sry) gene. Sry is necessary and sufficient for development of testes, and in the absence of Sry mammals develop ovaries [13]. As such, genotypic and gonadal sex are causally linked and the relative contributions of chromosomal and gonadal sex to sex-biases in health and disease are difficult to disentangle.

On the other hand, gender is a societal construct that can be molded by an individual’s perception of their sex, as well as influences from their social and physical environments. Although gender, including behavioral and societal influences, likely impacts health and disease outcomes [14], it is not possible to discern gender in non-human animal models. Here, we focus on the relative contributions of genotypic and gonadal sex to sex differences in the adult mouse hippocampus.

After gonad differentiation, hormonal secretions influence the organism’s sexual phenotype. Gonadal hormonal secretions lead to organizational effects that cause sex differentiation during development, as well as activational effects that may be temporary and reversible and can occur at any stage of life [15]. In a seminal paper, Phoenix et al. (1959) [16] described the organizational-activational theory of sexual differentiation in which during early development, hormones have an organizational effect on neural tissue development and circuitry that mediates mating (and likely other behaviors). After the organizational framework is established during development, activational effects are mediated by levels of gonadal and non-gonadal sex hormones. As such, the direct contributors to phenotypic sex effects are: (1) activational effects of gonadal hormones, (2) organizational effects of gonadal hormones, and (3) sex chromosomal effects [17].

Since the onset of developmental and age-related brain diseases correlate to critical windows of hormonal transition, gonadal hormone effects have been widely studied in brain diseases [18,19,20,21,22]. As a result, studies have interrogated the therapeutic potential of hormonal replacement therapies (HRT) in brain diseases (i.e., AD [23,24,25]). However, potential adverse outcomes [26, 27] and the feminizing/masculinizing effects of gonadal hormones have diminished enthusiasm for HRT approaches. On the other hand, several studies have also pointed to distinct contributions of sex chromosomes to brain development [28, 29] and pathology [30,31,32], opening new lines of investigation for the therapeutic target development for the treatment of brain diseases. Thus, establishing the contributions of sex chromosomes, independent of gonadal hormones, to brain diseases have come to the forefront of the neurobiology of sex differences.

Despite being the largest genomic difference between humans, the role of sex chromosomes in regulating sex effects is complex and still mostly unclear. Although the human Y-chromosome only contains 568 genes (71 protein-coding) [33], mosaic loss of Y with age has been linked to cancer [34] as well as AD [35]. In comparison, the X-chromosome has between 900 and 1500 genes, but its expression profile is complicated by the random inactivation in females of one X-chromosome on a cell-by-cell basis due to dosage compensation [36]. In fact, the complex nature of sex chromosome genomic regulation led to standardized exclusion of sex chromosomes from genome wide association studies (GWAS) [37], even those specifically interested in sex effects with brain disease [38]. X-chromosome gene dosage effects are evident in human sex chromosome aneuploidies, including Turner syndrome (XO), Triple X Syndrome (XXX), and Klinefelter syndrome (XXY), each with a wide array of CNS symptomology [39,40,41]. Thus, in humans it appears that genotypic sex, especially the number of X-chromosomes, plays an important role in proper brain development and function. However, sex chromosomal aneuploidies in humans are often confounded by different levels of gonadal hormones. Whereas, in mouse models the levels of hormones are more easily controlled and can be dissociated from the effects of different numbers of sex chromosomes.

To disentangle the effects of gonadal sex (testes v. ovaries; M v. F) and sex chromosome complement (XX v. XY) on the steady state gene expression and DNA modification patterning of the hippocampus, we use the Four Core Genotypes (FCG) mouse model [42]. The FCG male (XYM) originated through two sequential genetic changes [43]: 1) a spontaneous deletion of the testis-determining Sry gene from the Y-chromosome (Y−Sry) [44] and 2) transgenic insertion of the Sry gene onto an autosome (ASry) [45] resulting in XY−SryASry (XYM). Crossing the XYM with a wild-type C57BL/6 J XX female (XXF), results in the uncoupling of gonadal and chromosomal sex: XX and XY−Sry mice with ovaries (XXF/XYF) and XXASry and XY−SryASry mice with testes (XXM/XYM) (Fig. 1A). This allows for 2-way statistical comparisons to assess the contributions of gonadal and chromosomal sex, as well as interactive effects, on molecular and phenotypic outcomes (Fig. 1B).

Fig. 1
figure 1

Sry copy number and localization in FCG hippocampus. A) Breeding strategy used to generate the Four Core Genotypes (FCG). B) FCG mouse model two-way design to study gonadal sex and sex chromosomal contributions to sex effects. DNA isolated from FCG and wild type hippocampi (n = 3/group) was used for digital PCR Sry copy number and 10X genomics linked read sequencing. C) Wildtype (WT) C57BL/6 J mice have one copy of the testis-determining Sry gene, while FCG males (both XX and XY) have 12–14 copies of Sry. Sry was not detected in the FCG females. D) Linked read sequencing of FCG XYM shows strong linkage of the Sry gene to itself, but not to adjacent regions of the Y-chromosome (ChrY) indicating Sry is no longer on ChrY. E) Sry gene shows strong linkage to an intergenic region on Chromosome 3 (Chr3). F) Adult mouse (3–4 months old) hippocampal gene expression from previously published studies [Bundy et al., 2017 (GSE83931), Chucair-Elliott et al., 2019 (GSE135752); and Hadad et al., 2019 (PRJNA523985)] were compared to FCG hippocampal gene expression in the present study, to determine potential alterations in gene expression adjacent (± 10 Mbp) to the Sry insertion site on Chr3 in the FCG XXM/XYM. Tracks represent genes that are detected as expressed in previously published datasets (blue) and the present FCG study (red)

Full size image

Early FCG mouse studies were aimed at determining if sex chromosome complement (XX v. XY) contributed to development of well-established sexually dimorphic phenotypes [43, 46,47,48] [49] [50, 51]. Later FCG studies have brought to light distinct contributions of genotypic sex to disease-associated phenotypes seen in Experimental Autoimmune Encephalomyelitis (EAE) and pristane-induced Systemic Lupus Erythematosus (SLE) mouse models [30]. Additionally, XX mice have longer lifespans than their XY counterparts, regardless of gonadal sex [31], and XX mice show resiliency to death in an AD mouse model [32].

Although hippocampal sex differences in the transcriptome and epigenome are well-established across development, aging, and disease in mice and humans, the relative contributions of sex chromosome complement (XX v. XY) and gonadal sex (M v. F) to the steady state and stimulus responsive transcriptome and epigenome are not fully defined. Given its influence on X-chromosome inactivation (XCI), genome accessibility, and regulation of gene expression, specific methylation of the fifth carbon of a cytosine residue, resulting in the DNA modification 5-methyl-cytosine (mC) is a particularly interesting epigenetic factor. In this study, we use transcriptomic and epigenetic approaches to examine the hippocampal transcriptome and methylome in adult FCG mice. We then compare our findings to previously identified hippocampal sex differences to begin to separate contributions of sex chromosome complement (XX v. XY) and gonadal sex (testes v. ovaries; M v. F) to sex differences in transcriptional programming.

Results

Sry Copy number and Localization in Adult FCG Hippocampi

The testis-determining Sry gene is considered the “master switch” in mammalian gonadal sex determination [52]. In the FCG model, Sry is absent from the Y-chromosome and inserted onto an autosome, uncoupling gonadal and genetic sex. Previous initial reports [53] used fluorescence in situ hybridization (FISH) and PCR amplicon sequencing to localize a concatemer (12–14 copies) of Sry on chromosome 3 (Chr3: 70,673,749–70,673,824) in FCG XX and XY males. To verify Sry copy number, we designed a digital PCR Sry copy number assay and confirmed 12–14 copies of Sry in FCG males (XXM/XYM), as compared to one copy in WT males (Fig. 1C). To confirm the localization of Sry, a form of linked-read sequencing was used, which barcodes sequence reads that come from the long DNA fragments with the same oligo tag. Within the visualization, a darker amber color indicates more linked reads between the regions on the X and Y axes (Fig. 1D). There was no linkage between Sry and the adjacent region of the Y-chromosome, indicating deletion of Sry from the Y-chromosome. There was a strong linkage of the Sry gene with the previously-identified [53] region of chromosome 3, within a region with no known gene annotation (Fig. 1E). No other linked regions were identified indicating that this is the only autosomal insertion site of Sry. To examine if Sry insertion alters expression of genes on Chr3, hippocampal RNA-Seq data from and wild-type C57Bl/6 J mice was compared for genes adjacent (± 10Mbp) to the Sry insertion site (Fig. 1F). Forty genes were expressed at detectable levels (> 20 reads) in male and female hippocampi in all three previous sex difference studies examined [54,55,56], and each of these 40 genes were also expressed across all four groups in the FCG hippocampi (Supplemental Table 1). No sex differences were identified across the three previously published studies. Only a single gene (Fam198b) ~ 9.2 Mbp from the Sry insertion was differentially expressed by sex in the FCG hippocampi. Due to the distance from the insertion site and variability in autosomal sex differences, this sex difference is unlikely a result of the Sry insertion. Further, there was no ectopic expression of Sry in the hippocampi of male FCG mice.

Table 1 Digital PCR copy number and BSAS primers (5’- > 3’) for targeted methylation analysis
Full size table

Transcriptomic Analysis of Sex Chromosomal (X/Y) Differential Expression from Adult FCG Hippocampi

Despite X-inactivation compensatory mechanisms, there are a number of X-chromosome genes whose expression is imbalanced between males and females in the mouse hippocampus. Previous reports have established differentially expressed sex chromosomally-encoded genes in the mouse hippocampus throughout development and aging [54,55,56,57] (Supplemental Table 2; GEO Accession: GSE83931, GSE135752, GSE76567; SRA bioProject: PRJNA523985). Intersecting sex chromosomally-encoded (X/Y) differentially expressed genes by sex across studies identified eight common genes (Fig. 2A, Supplemental Table 2), including X-chromosome genes (** [80]. On the other hand, SINE elements seem tightly correlated with proximate gene regulatory factors, including promoters and transcription factor binding sites. We hypothesize that, in addition to serving as boundary for heterochromatic domains, SINE-Alu elements serve as a template to recruit epigenome modifiers to regulate gene transcription (especially in genes that escape X-inactivation) in response to gonadal hormone signals. Motif analysis of X-chromosomal SINE-Alu repeat sequences identified SP1 as a potential transcription factor related to these sequences (Fig. 4C). STRING protein network analysis [81] identified estrogen receptor 1 (ESR1) and histone deacetylase 1 (HDAC1) as predicted interactors to SP1 (Fig. 4D).

In the adult FCG hippocampus, Aff2 was differentially expressed by gonadal sex (M v. F), with higher expression in gonadal males as compared to females (Fig. 1F), regardless of sex chromosome complement. Investigation of the Aff2 promoter region revealed TSS flanking by SINE-Alu repeats, as well as transcription factor binding sites for SP1, ESR1, ESR2, and JUN (Fig. 4E). To better understand the epigenomic regulation of Aff2, we examined representative genome tracks of FCG hippocampal methylation alongside publicly available methylation and chromatin data, as well as positioning of CpG islands (CGI) and L1/Alu repeats (Fig. 4F). Aff2 has a CGI-containing promoter that coincides with active histone marks (H3K4me2, H3K4me3, H3K27ac, H3K9ac) and open chromatin (ATAC-Seq) peaks from P0 forebrain [82]. Flanking the Aff2 TSS are Alu repeats (both up and down stream), with L1 elements completely absent from the promoter region, but densely populating the region up and downstream of the flanking Alu repeats. Considering only CpG sites between the two Alu repeats flanking the TSS, we quantified the site-specific methylation within each of the FCG groups. Within this region, mCG was higher in XX compared to XY (Fig. 4G; Two-way ANOVA, main effect sex chromosome complement, ***p < 0.001) and higher in gonadal females compared to males (Fig. 4F; Two-way ANOVA, main effect gonadal sex, #p < 0.001), as well as a significant interaction between chromosomal and gonadal sex (p = 0.001). Consistent with these results, the Aff2 promoter was hypomethylated in WT C57Bl/6 males (compared to female) [83] mirroring the patterning seen in the FCG hippocampus. These data suggest that gonadal sex (and potentially gonadal hormones) may contribute to X-chromosomal DNA methylation and have implications in escape from X-inactivation. Based on the associations observed here, we propose that circulating estradiol binds to ESR1 and in the nucleus complexes with SP1 and HDAC1. The complex interacts with the SINE-Alu repeats flanking active promoters to induce changes in the chromatin landscape, including hypermethylation and heterochromatization (Fig. 4H).

Epigenomic Patterning in Genes that Consistently Escape X-Chromosome Inactivation in Mouse Hippocampus

There were four X-chromosome genes (Kdm6a, Ddx3x, Eif2s3x, ** mouse hippocampus. BMC Genomics 18(1):237" href="#ref-CR54" id="ref-link-section-d406529999e2564">54,55,56,57] or present FCG studies) (Supplemental Table 2–3) matched that of consistent escape genes, we assessed the epigenomic marks associated with variably esca** genes (Supplemental Table 4). We found that only three variable escape genes (Med14, Magt1, BC065397) shared the same chromatin signature as the common escape genes (Kdm6a, Eif2s3x, and Ddx3x). However, the promoter regions of these genes (Supplemental Table 4), are hypermethylated in XX genotypes compared to XY, suggesting that the epigenomic landscape (including histone marks) are likely different between the active and inactive X.

Of the variable escape genes: 27% had large CGI promoters, 39% had active promoter histone marks, 17% had gene body H3K36me3, 55% had SINE-Alu promoter flanks, and 49% had low L1 gene body density (< 10%) (Fig. 5E). Thus, it seems likely that the epigenomic regulation of variably esca** genes is distinct from that of common escape genes. As previously suggested, we believe that X-chromosomal DNA methylation could be modulated by gonadal hormone levels. ATAC-Seq profiling of the female brain across the estrus cycle identified 238 genes in proximity to estrus-responsive chromatin [90]. Intersecting the 238 estrus-responsive chromatin genes with the 160 variable escapee genes and the consistent escape chromatin signature, identified 48 estrus-responsive genes with variable escape and a chromatin signature distinct from the consistent escape genes (Fig. 5F, Supplemental Table 4). ** windows and minimum average difference of 10% between at least two groups (Chisq-test, sliding linear model (SLIM) q < 0.05), followed by a two-way ANOVA to determine main effects of chromosomal and gonadal sex, as well as potential interactions (Bonferroni correction for six pairwise comparisons, p < α = 0.0083). Using these criteria, we identified 2,552 DMRs: 2,456 by sex chromosome (XX v. XY), 145 by gonadal sex (M v. F), and 87 interactive effects (Fig. 6A, Supplemental Table 5) on the X-chromosome. Consistent with our previous findings (Fig. 3–5), sex-chromosomally regulated DMRs were: 1) enriched in gene regulatory features (gene body, promoter, TFBS, enhancer, CGI, CTCF) and CNS histone marks for active promoters (H3K4me2, H3K4me3, H3K27ac, H3K9ac), poised promoters (H3K4me1, H3K27me3) and transcriptional regulation (H3K36me3), and 2) depleted in repetitive elements (LINEs, SINEs, LTRs) (Fisher’s exact test, p < 0.05) (Fig. 6B). Correlation of Jaccard distances of the assessed genomic features with sex chromosomally-driven DMRs, show strong correlation to active promoter regions, again suggesting strong sex chromosome autonomous regulation of X-inactivation (Fig. 6C). Within unexpressed genes in the FCG hippocampi there was a small but significant difference in promoter methylation between XX (~ 68%) and XY (~ 64%) (Fig. 6D, Two-way ANOVA, main effect sex chromosome complement, ***p < 0.001), with no effect of gonadal sex. Within expressed gene promoters, there was a large difference between XX (~ 40%) and XY (~ 17%) genotypes (Fig. 6D, Two-way ANOVA, main effect sex chromosome complement, ***p < 0.001), with no effect of gonadal sex. The incidence of CGI-containing promoters was much higher within expressed genes (~ 61%) as compared to unexpressed genes (~ 11%) (Chi-sq test, p < 0.05) (Fig. 6E). There was a strong, negative association between the difference in promoter mCG (XX-XY) and log(FC(XX/XY)) gene expression (Pearson r, p < 0.05) with no difference in slope or intercept of the linear fit between gonadal males and females. X-chromosomal genes with XX-biased expression (XX > XY) had smaller differences in promoter methylation (XX-XY), while genes with XY-biased genes had larger differences in promoter methylation (XX-XY). These results are consistent with the escape signature of Kdm6a, Ddx3x, and Eif2s3x (Fig. 5A).

Fig. 6
figure 6

Epigenomic analysis of X-chromosomal differentially methylated regions (DMRs) from the adult FCG hippocampus. DNA was isolated from FCG hippocampi (n = 3/group). Methylation in CG context was assessed by WGoxBS (n = 3/group) and analyzed in methylKit to identify ChrX DMRs (w = 1000). ChrX DMRs with minimum mCG difference of 10% between at least two groups (Chisq-test, sliding linear model (SLIM) q < 0.05) were assessed for main effects of sex chromosome (XX v. XY) and gonadal sex (M v. F) (2-way ANOVA, Bonferonni α < 0.0083). A) Venn diagram of DMRs by sex chromosome (XX v. XY), gonadal sex (M v. F), and interactive effect of gonadal and chromosomal sex. B) Relative enrichment and depletion of ChrX DMRs by sex chromosome (XX v. XY) within various genomic features. C) BedSect (Mishra et al., 2020) correlation plot of ChrX genomic regulatory features (Ward hierarchical clustering, Jaccard similarity index). D) Distribution of unexpressed and expressed gene promoter mCG (%). Unexpressed gene promoters (± 2 kb from TSS) had higher mCG in XX genotypes (68%) as compared to XY genotypes (54%), regardless of gonadal sex. Expressed gene promoters had higher mCG in XX genotypes (40%) as compared to XY genotypes (17%), regardless of gonadal sex (2-way ANOVA, main effect sex chromosome complement, ***p < 0.001). E) Only 11% of unexpressed ChrX gene promoters contain CGIs, as compared to 61% of expressed ChrX gene promoters. F) Correlation of log(FC) gene expression (XX/XY) and difference in gene promoter mCG (XX-XY) for 19 ChrX DEGs identified in Fig. 2E (Pearson r, p < 0.05). ** mouse hippocampus. BMC Genomics 18(1):237" href="/article/10.1007/s12035-022-02860-0#ref-CR54" id="ref-link-section-d406529999e3249">54]) suggests that hippocampal sex differences observed during postnatal development are distinct from those seen in adulthood and with aging.

Fig. 9
figure 9

Transcriptomic analysis of autosomal differential expression in the adult FCG hippocampus. DNA and RNA were isolated from FCG hippocampi (n = 10–16/group). mRNA expression was assessed by stranded RNA-Seq (n = 5–6/group) and RT-qPCR (n = 10–16/group). Results were compared to previously published hippocampal transcriptomic sex differences. Boxplots represent median, interquartile range, and minimum/maximum normalized RQ. A) Comparison of four previous hippocampal transcriptomic studies (Supplemental Table 4), shows 2896 autosomal-encoded sex differences in wild-type mice across all studies with no genes in common between all studies. B) Pairwise correlations of the fold-change (M v. F) for the 2896 autosomal DEGs were conducted for genes that were expressed in all studies. Hierarchical clustering of the correlation coefficients (Pearson r) shows grou** of studies. C) WebGestalt over-representation analysis of the 2896 autosomal-encoded sex differences in WT hippocampus identified KEGG pathways, transcription factors, and GO biological processes enriched in hippocampal sex difference genes (hypergeometric test, BHMTC, FDR < 0.05). D) Previously identified estrus-cycle responsive genes (Jaric et al., 2019) were compared to hippocampal sex differences identified in previous WT (Fig. 9A) and the present FCG study. The number of estrus-responsive genes (diestrus female v. male – red, proestrus female v. male – blue) were plotted for each dataset. E) Differential expression analysis of FCG hippocampal RNA-Seq libraries identified 212 differentially expressed autosomal genes, 62 had a main effect of chromosome only (XX v. XY), 123 had a main effect of gonadal sex only (M v. F) and 26 genes had main effects of chromosome and gonadal sex. There was only one gene (Tm7sf2) which had an interactive effect of chromosomal and gonadal sex (Two-way ANOVA, BHMTC, FC >  = 1.25, FDR < 0.1). F) Normalized gene expression of Tm7sf2 by RNA-Seq (Two-way ANOVA, BHMTC, FC >  = 1.25, FDR < 0.1, interactive effect). Bar chart is mean ± SEM. G) Principal component analysis of differentially expressed autosomal genes showed separation of gonadal sex (M v. F) in the first component (33.9%) and separation of the chromosome (XX v. XY) in the second component (20.2%). H) Hierarchical clustering of differentially expressed autosomal genes shows separation of genotypes by gonadal sex and sex chromosome complement. I) ORA of the autosomal sex differences identified in the FCG hippocampus revealed biological differentially regulated by sex chromosome complement and/or gonadal sex. J-N) Differential expression of select genes was confirmed by RT-qPCR (n = 10–16/group, Two-Way ANOVA, main effect of sex chromosome complement (XX v. XY) *p < 0.05, **p < 0.01, and ***p < 0.001 or gonadal sex (M v. F) #p < 0.05, ##p < 0.01, and ###p < 0.001). Only Irf7 had a significant interaction between gonadal and chromosomal sex (p = 0.012)

Full size image

Next, we ran GO Biological Process, KEGG Pathway, and transcription factor over-representation analysis (ORA) using WEB-based GEne SeT AnaLysis Toolkit (WebGestalt, www.webgestalt.org) on autosomal genes that were differentially expressed by sex in at least one outside study (Supplemental Table 2 (Union)) (Fig. 9C, Supplemental Table 6, hypergeometric test, BHMTC, FDR < 0.05). Enriched biological processes included lipid homeostasis, CNS neuron differentiation, regulation of leukocyte activation, pattern specification process, and reproductive system development, among others. Differentially expressed genes shared transcription factor regulators for important functions, including immune regulators (IRF1, SRF, and MEF2), metabolic regulators (FOXO4), and vesicular trafficking (AP4), among others. Two significant KEGG pathways were over-represented in the hippocampal sex differences: neuroactive ligand-receptor interaction and systemic lupus erythematosus. Together, these data suggest that hippocampal sex differences alter neuronal differentiation/signaling and immune-related processes, which both could be partially mediated by differences in metabolic processes.

Of note is the fact that only one of the previous sex difference studies [54] mentioned estrus-cycle staging females, with collection of 2 estrus, 2 metestrus, and 1 diestrus females from each age group (1, 2, 4 mo). As such, this study [54] identified fewer estrus-dependent transcriptomic changes in 2 mo (6 genes) and 4 mo (22 genes) compared to age-matched 2 mo [57] (78 genes) and 4 mo [55] (90 genes) from other studies that did not estrus-cycle stage (Fig. 9D, Table 2). The 24 mo group [56] had few estrus-responsive DEGs, while the 2 mo [57], 3 mo [101], and 4 mo [55] that were not estrus cycle staged showed the largest number of estrus-responsive DEGs (Fig. 9D).

Next, we assessed autosomal sex differences in FCG hippocampus by sex chromosome complement (XX v. XY) and gonadal sex (M v. F) using directional RNA-Seq. After aligning, quantifying, and calling 212 differentially expressed genes: 62 genes differentially expressed by sex chromosome complement alone (XX v. XY), 123 genes differentially expressed by gonadal sex (M v. F), 26 genes differentially expressed by both sex chromosome complement and gonadal sex, and 1 gene with a significant interaction of gonadal and chromosomal sex (Fig. 9E). Principal component analysis of differentially expressed autosomal genes showed separation by gonadal sex in component 1 (33.9%) and separation by sex chromosome complement in component 2 (20.2%) (Fig. 9G). Hierarchical clustering of autosomal-encoded DEG shows proper clustering of samples by genotype (Fig. 9H). ORA of the 88 chromosomally-driven (XX v. XY) autosomal sex differences in the FCG hippocampus revealed four enriched biological processes: antigen processing and presentation, regulation of innate immune response, response to virus, and response to protozoan (Hypergeometric test, BH FDR < 0.05). ORA of the 149 gonadal sex-driven (M v. F) autosomal sex differences in the FCG hippocampus revealed three enriched biological processes: leukocyte mediated immunity, response to virus, and response to protozoan (Hypergeometric test, BH FDR < 0.05) (Fig. 9I).

Since we identified response to viruses and interferon-beta as pathways enriched in sex differences in the FCG hippocampus, we further examined interferon-associated genes IRF-7 and IFIT-3, both of which were differentially expressed by sex in other studies (Supplemental Table 3 (Union)). Interferon (IFN), as part of the innate immune system, triggers the expression of IFN-stimulated genes (ISGs) in response to viral recognition. IFN-beta is a type I IFN that is activated through PRRs [102], like Tlr7, is primarily in microglia [97]. In mouse models of AD, IFN was found to activate microglia leading to neuroinflammation and synaptic degradation. Blocking IFN signaling decreased microglia activation and concomitant synapse loss. Activation of IFN pathway was also observed in human AD [103]. Transcription factor IRF-7 is considered a “master regulator” in type-I IFN responses [104]. Irf7 was differentially expressed in our study by sex chromosome complement (XX v. XY) and gonadal sex (M v. F) in FCG hippocampus as evidenced by RNA-Seq (Supplemental Table 4) and RT-qPCR confirmation (Fig. 9J). Irf7 RT-qPCR also identified an interactive effect of gonadal and chromosomal sex (p = 0.012). Interferon-induced protein with tetratricopeptide repeats 3 (IFIT3) is an antiviral RNA-binding protein which acts an intermediary in the activation of IRF-3 and upregulation of IFN-beta [105]. Ifit3 was differentially expressed by sex chromosome complement (XX v. XY) and gonadal sex (M v. F) in FCG hippocampus as evidenced by RNA-Seq (Supplemental Table 4) and RT-qPCR confirmation (Fig. 9K).

Antigen processing and presentation pathways were also over-represented in our analysis of genes differentially expressed by chromosomal sex in the FCG hippocampus. Class I major histocompatibility (MHC-I) complexes function in innate viral antigen presentation and detection. Given that previous studies from our group identified sexually divergent induction of MHC-I with aging in the mouse and rat hippocampus [106, 107], we examined expression of MHC-I components B2m, H2-D1, and H2-K1. Here we show differential expression of B2m by gonadal sex (Fig. 9L) and H2-D1/H2-K1 by sex chromosome complement and gonadal sex (Fig. 9M-N) in the FCG hippocampus (n = 10–16/group, Two-Way ANOVA, main effect of sex chromosome complement (XX v. XY) *p < 0.05, **p < 0.01, ***p < 0.001 and/or gonadal sex (M v. F) #p < 0.05, ##p < 0.01, ###p < 0.001).

Based on the meta-analysis of previous studies, it appears that autosomal sex differences in the hippocampus are context-dependent and likely change throughout development and with aging. In the adult FCG hippocampus, sex differences were primarily enriched for in immune-related pathways and are regulated by both chromosomal and gonadal sex.

Autosomal Chromosome Levels of Methylation in FCG Hippocampus by WGoxBS

After analyzing autosomal sex differences in FCG hippocampal gene expression, we assessed autosomal methylation in CG and CH context by WGoxBS. Overall, there were no differences in total autosomal mCG (Fig. 10A), as well as when split into in repetitive (Fig. 10B) and non-repetitive (Fig. 10C) element autosomal mCG. No differences were found in average autosomal mCH (Fig. 10D), repetitive element mCH (Fig. 10E) or non-repetitive element mCH (Fig. 10F) levels. There were also no apparent differences in autosomal mCG patterning across CGI, shores, and shelves (Fig. 10G), gene bodies/flanking regions (Fig. 10H), or CTCF-binding sites/flanking regions (Fig. 10I). These data are consistent with previous findings that sex differences in mouse hippocampal autosomal methylation are site-specific and are a mix of both hyper- and hypomethylation in females compared to males [108].

Fig. 10
figure 10

Autosomal levels of DNA methylation in the FCG hippocampus by WGoxBS. DNA and RNA were isolated from FCG hippocampi (n = 10–16/group). Autosomal methylation in CpG (CG) and non-CpG (CH) contexts were assessed by WGoxBS (n = 3/group). A) There is no difference in overall mCG on autosomes, in either (B) Repetitive or (C) Non-repetitive elements. D) There is no difference in overall mCH on autosomes in either (E) Repetitive or (F) Non-repetitive elements (Two-way ANOVA). G-I) There is no difference in autosomal mCG in: (G) CpG Islands (CGI), shores, and shelves; (H) gene bodies and flanking regions (± 4 kb); or (I) CTCF binding sites and flanking regions (± 4 kb)

Full size image

Autosome Differentially Methylated Regions (DMRs) by Whole Genome Oxidative Bisulfite Sequencing (WGoxBS) in FCG Hippocampi

CG methylation was examined in the context of autosomal: CNS-associated histone marks [82] (H3K9ac, H3K4me2, H3K4me3, H3K27ac, H3K9me3, H3K36me3, H3K4me1, H3K27me3) (Fig. 11A), specific repetitive elements (SINE: Alu, B2, B4, MIR; LINE: L1, L2) (Fig. 11B), and proximal gene regulatory regions (enhancer, promoter, CGI, gene body) (Fig. 11C). There were no main effect differences by gonadal or chromosomal sex in average mCG across any of these genomic regions. In LINE elements (L1 & L2) there was a significant interactive effect of gonadal and chromosomal sex on autosomal LINE elements (L1, L2) mCG (Two-way ANOVA, p < 0.05) but with a minimal methylation difference (< 1%). While there were no sex effects in average mCG across these features, the levels of mCG varied among different features (i.e. Enhancers ~ 48% and Promoters ~ 3%).

Fig. 11
figure 11

Epigenomic analyses of autosomal differentially methylated regions (DMRs) from the adult FCG hippocampus. Autosomal methylation in CpG (CG) context was assessed by WGoxBS (n = 3/group). First, average mCG levels in various genome contexts across autosomes were compared and no overall significant sex effects (gonadal or chromosomal) in mCG within: (A) autosomal P0 forebrain histone modification foci, (B) repetitive elements (LINE-L1, LINE-L2, SINE-Alu, SINE-B2, SINE-B4, SINE-MIR) or (C) autosomal proximal gene regulatory regions (Enhancer, promoter, CGI, Gene Body) (Two-way ANOVA). Differentially methylated regions (DMRs) (w = 1000 bp) were called in methylKit (post-hoc Two-Way ANOVA with Bonferonni correction p < α = 0.0083). D) GO Biological process over-representation analysis of autosomal genes (± 2 kb) containing DMRs by sex chromosome, gonadal sex, or an interaction of gonadal and chromosomal sex. E) Correlation of promoter mCG (-2 kb from TSS (XX-XY) and gene expression (log(FC)(XX/XY)) in autosomal DEGs by sex chromosome (XX v. XY) was not significant in XXF v. XYF or XXM v. XYM comparisons (Pearson r, p > 0.05). F) Correlation of promoter mCG (-2 kb from TSS (XX-XY)) and gene expression (log(FC)(XX/XY)) (Pearson r) in autosomal DEGs by sex chromosome (XX v. XY). Promoter methylation and gene expression were negatively correlated in XYF v. XYM comparison (Pearson r = -0.167, *p = 0.045) but this was not observed in the XXF v. XXM comparison. G) Genes closest to autosomal DMRs (± 2 kb gene body) with main effects of sex chromosome (XX v. XY) and/or gonadal sex (M v. F) were intersected with FCG autosomal DEGs (Fig. 9E). The majority (177/211 = 84%) of DEGs did not coincide with differentially methylated regions. H) Genome tracks of Esr1 gene body (chr10) with FCG DMRs by chromosomal (1 & 2) and gonadal (3) sex, FCG mCG (%), WT male and female neuronal mCG (Lister et al., 2013), and male (testis) and female (placenta) specific histone peaks (H3K27ac, H3K4me1, H3K4me3) (The ENCODE Project Consortium et al. 2020). I) Esr1 DMR (1, 2, & 3) mCG. DMRs 1 & 2 were differentially methylated by chromosomal sex (XX v. XY), while DMR 3 was differentially methylated by gonadal sex (M v. F) (Two-way ANOVA, main effect of sex chromosome **p < 0.01, main effect gonadal sex, ##p < 0.01). No significant interactions between gonadal and chromosomal sex were observed. J) DMR mCG levels within select autosomal FCG DEGs (Gabrb1, Oas2, Setbd2). DMRs within Gabrb1 and Oas2 were differentially methylated by sex chromosome complement (Two-way ANOVA, main effect sex chromosome complement, **p < 0.01). The DMR within the Setdb2 gene body was differentially methylated by chromosomal and gonadal sex (Two-way ANOVA, main effect sex chromosome complement, **p < 0.01, main effect of gonadal sex, ##p < 0.01). No interactive effect of gonadal and chromosomal sex was observed

Full size image

After exploring the overall levels and patterning of DNA methylation on the autosomes, autosomal mCG DMRs were assessed using 1 kb non-overlap** windows (Chisq-test, sliding linear model (SLIM) q < 0.05) and post-hoc two-way ANOVA with Bonferonni correction (p < α = 0.0083) to assess main and interactive effects of chromosomal and gonadal sex. Using these criteria, we identified 4505 DMRs by sex chromosome (XX v. XY), 1385 DMRs by gonadal sex (M v. F), and 487 significant interactive effects of gonadal and chromosomal sex (Supplemental Table 7). We identified 2615 genic regions intersecting DMRs by chromosomal sex, 911 genic regions intersecting DMRs by gonadal sex, and 332 genic regions intersecting DMRs with an interactive effect of chromosomal and gonadal sex. These gene lists were assessed by GO Biological process over-representation analysis to identify significant pathways (Supplemental Table 7, hypergeometric test, BHMTC < 0.2). ClusterProfiler [109] functional profiler assessment identified enrichment in neuronal-specific pathways (synapse organization, dendrite development, glutamatergic synaptic transmission) within the sex chromosomal and/or gonadal DMR- related genes (Fig. 11D).

To determine if there was a relationship between autosomal gene expression and promoter methylation, we correlated the difference in promoter methylation (mCG %) and differential gene expression (log(FC)) by sex chromosome (Fig. 11E, XX v. XY) and gonadal sex (Fig. 11F, M v. F). There was a weak negative correlation between differential gene expression and promoter methylation among the XYF v. XYM comparison (Fig. 11F, Pearson r = -0.167, *p = 0.045). No other significant correlations were identified.

Most (84%) of the autosomal FCG DEGs from Fig. 9E did not overlap with a DMR (Fig. 11G). The autosomal DEGs that did overlap with DMRs appear to be primarily related to neuronal functions (Gabrb1, Syt2, Cacna1c, Kcnj6). The 177 DEGs not overlap** with DMRs, were enriched in immune-related genes (i.e., Irf7, Ifit1/3, B2m, H2-D1, H2-K1). Since immune cells, like microglia, are minority populations compared to neurons, the epigenomic signatures of microglia are not well represented in the bulk epigenomic data and require epigenomic analysis of these specific cell types to reveal methylation differences not observed in whole tissue analyses [110]. Further cell type-specific epigenomic data will need to be collected from the FCG model to determine the cell type-specific contributions to sex differential gene expression in the hippocampus.

Esr1 was identified as a gene that variably escapes X-inactivation in the mouse hippocampus. Within the FCG mCG data, we identified three DMRs within the Esr1 gene body: two DMRs by chromosomal sex (DMRs 1 & 2) and one DMR by gonadal sex(DMR 3) (Fig. 11H-I). Esr1 DMRs by chromosomal sex may suggest a sex differential imprinting mechanism used to regulate Esr1 expression before the induction of gonadal hormones. The gonadal DMR indicates a potential feedback domain to modulate estrogen-receptor expression in response to gonadal hormones.

There were also examples of differential methylation within DEGs in the FCG hippocampus, including: GABA receptor Gabrb1, interferon response gene Oas2, and methyl-CG dependent histone modifier Setdb2 (Fig. 11J). These results suggest that DNA methylation is dynamic and reducing the signal-to-noise-ratio with cell type-specific data will be critical for determining the true nature of DNA methylation dependent gene regulation for varying genomic contexts.

Despite a strong immune-related transcriptomic signature, differential methylation appears to be mostly involved in neuron-related pathways. This could be expected given that neurons are the predominant cell type in the hippocampus. While the association of DMRs to DEGs is variable, the expression analysis examined steady state gene expression and these DMRs may cause differential gene expression in acute, stimulus-responsive situations. Future studies will also need to specifically examine specific glial cell populations (i.e. microglia) to uncover methylation-gene expression relationships in these cells.

Discussion

The study of sex effects in brain health and disease have begun receiving needed experimental attention in neuroscience studies. Not only do the sexual dimorphisms, differences, and divergences [111] need to be characterized but also the regulatory mechanisms giving rise to these sex effects need to be addressed. While hormonal mechanisms (both organizational and activational) have been the most studied, the potential regulation of sex effects by sex chromosomes, either independently or in concert with gonadal hormones, has received relatively limited attention. Symptomology associated with human sex chromosome aneuploidies [39,40,41] and experimental evidence obtained from FCG mouse studies [17, 31, 32, 48, 53] have provided critical insight into the importance of sex chromosomes (independent of gonadal sex) to sex effects seen in development, aging, and disease. Previous reports have established sex differences in the hippocampal epigenome and transcriptome with brain aging [108, 125] and disease [57, 126, 127], further suggesting that the intersection of epigenetic alterations and sex effects requires further investigation. Here we use the FCG mouse model to delineate the contributions of gonadal and chromosomal sex to the epigenomic regulation of hippocampal gene expression in adult mice.

In the present study, we first validated the genetics of the FCG model, confirming prior imaging and PCR studies [53] and providing a precise location of Sry insertion to an intergenic region of Chr3. While there are ~ 13 copies of Sry, the insertion is not within an annotated gene. Since the concatemer of Sry could disrupt the chromatin accessibility within and around the insertion site on chromosome 3, we assayed for ectopic Sry expression and changes to gene expression genes adjacent to the insertion site. However, we found no evidence of Sry expression in the hippocampus (as expected) and no disruption of gene expression in adjacent genes. Additionally, tissue-specific expression of Y-encoded genes (Kdm5d, Eif2s3y, Ddx3y, Uty) was maintained in the FCG, despite deletion of Sry. Together these data provide confidence that the transgenic manipulations used to generate the FCG are not confounding the analysis of sex differences in the hippocampus.

XCI, the random silencing of one X-chromosome (** was performed using the primers: Sry (5′-AGC CCT ACA GCC ACA TGA TA-3′, 5′-GTC TTG CCT GTA TGT GAT GG-3′), Ymt (Y chromosome-specific sequence, 5′-CTG GAG CTC TAC AGT GAT GA-3′, 5′-CAG TTA CCA ATC AAC ACA TCA C-3′), and myogenin (5′-TTA CGT CCA TCG TGG ACA GCA T-3′, 5′-TGG GCT GGG TGT TAG TCT TAT-3′) [53], following Jackson Laboratory’s standard PCR assay genoty** protocol #5590.

High Molecular Weight (HWM) DNA Isolation for Pseudo Long-Read Genomic Sequencing

Hippocampi were dissected from a FCG XYM mouse, snap frozen in microfuge tubes with liquid nitrogen, and stored at -80ºC prior to DNA isolation. HMW DNA was extracted from fresh-frozen tissue according to 10X Genomics sample preparation protocol (https://support.10xgenomics.com/genome-exome/sample-prep/doc/demonstrated-protocol-hmw-dna-extraction-from-fresh-frozen-tissue). Frozen tissue was thawed on ice and minced with a razor blade. The tissue was gently Dounce homogenized in 500 µL of Nuclei Isolation Buffer (Sigma Nuclei PURE Prep Kit: Lysis Buffer, 1 mM DTT, Sigma Nuclei PURE Prep Kit: 10% Triton X-100). After a brief centrifugation, the supernatant was transferred using a wide-bore pipette tip to a 2.0 mL round-bottom tube and centrifuged at 500 × g for 5 min. The supernatant was discarded, and pelleted nuclei were then resuspended in 70 µL of ice-cold PBS. To digest the nuclei, 10 µL of Proteinase K was added to the resuspended nuclei, followed by 70 µL ice-cold Digestion Buffer (20 mM EDTA, pH 11, 2 mM Tris–HCl, pH 8.3,10 mM N-Laurylsarcosine sodium salt). Samples were rotated end-over-end for 2 h at room temperature. To purify the DNA, Tween-20 was added to the sample to a final concentration of 0.1% and then 1 × SPRISelect Reagent was added. The samples were rotated end-over end for 20 min. Tubes were placed in the DynaMag-2 magnetic rack to allow bead capture. After removing and discarding the supernatant, the beads were washed twice with 70% ethanol. The DNA was eluted from the beads with 50 µL Sample Elution Buffer (Qiagen AE Buffer, 0.1% Tween-20). Qubit dsDNA BR kit was used to quantify the DNA.

10X Linked-Read Library Preparation

Linked-read genomic libraries Chromium were constructed from 1 ng of HMW DNA from an XYM using Chromium Genome Library Prep Kit (#PN-120229, 10X Genomics, Pleasanton, CA), according to manufacturer’s instructions. Briefly, following HMW gDNA extraction 1 ng of HMW DNA was loaded onto a Chromium Genome Chip Kit (#PN-120216, 10X Genomics) for Gel Bead-in-Emulsions (GEM) generation and barcoding. After SPRISelect bead cleanup and library construction, libraries were normalized to 4 nM, pooled, and sequenced NextSeq500 (High PE150) in the OMRF Clinical Genomics Center (CGC). Sequencing data was aligned to the mm10 genome using the LongRanger pipeline (10X Genomics) and visualized in the Loupe browser.

Isolation of DNA/RNA from FCG Hippocampi

Hippocampi were dissected from XXF, XXM, XYF, and XYM FCG mice (n = 10–16/group), snap frozen in microfuge tubes with liquid nitrogen, and stored at -80ºC prior to DNA isolation. Nucleic acids (DNA/RNA) were isolated from flash-frozen tissues using the All Prep DNA/RNA Mini Kit (Qiagen, Germantown, MD) as previously described [127,128,129]. Briefly, 600 µL of Buffer RLT with beta-mercaptoethanol was added to the tube containing the frozen hippocampi. A steel bead was added to the tube and homogenized for 30 s at 30 Hz using a Tissue Lyser II (Qiagen). Homogenate was loaded onto a DNA spin column and the flow through supplemented with ethanol was loaded onto an RNA spin column. Columns were washed and nucleic acids eluted by centrifugation. Total DNA/RNA were quantified by Nanodrop (Thermofisher Scientific, Madison, USA). Quality of DNA and RNA were assessed by genomic and RNA screentapes, respectively, on a Tapestation 2200 (Agilent Technologies, Frankfurt, Germany). Only samples with RNA and DNA integrity numbers > 7 were used for subsequent experiments.

Digital PCR (dPCR) Sry Copy Number Assay

Custom Sry fluorogenic copy number assays were designed and purchased from Integrated DNA Technologies (Coralville, IA) with primers (5’-GTC ACA GAG GAG TGG CAT TT-3’, 5’-AGT CTT GCC TGT ATG TGA TGG-3’) and probe (5’-/56-FAM/CAACCTTCT/ZEN/GCAGTGGGACAG GAA/3IABkFQ/-3’). Mouse Tert fluorogenic copy number assay was used as a single copy gDNA control as reference (#4,403,316, Life Technologies). DNA from FCG hippocampi (n = 3/group) was used for dPCR, as previously [129], using the QuantStudio 3D Digital PCR kit (#A26361, ThermoFisher Scientific), according to manufacturer’s instructions. After combining DNA with the Quantstudio 3D mastermix and Sry or Tert fluorogenic assay, reactions were loaded onto a Quantstudio 3D PCR chip with a Quantstudio 3D chip loader (#4,482,592, Thermofisher Scientific) and cycled on a GeneAmp PCR system 9700 with a flatblock attachment. Chips were read by the Quantstudio 3D chip reader (#4,489,084, Thermofisher Scientific) and analyzed using Quantstudio 3D AnalysisSuite cloud software V3.1.

Quantitative PCR (qPCR)

Confirmation of gene expression levels was performed with qPCR as previously described [55, 130, 131]. cDNA was synthesized with the ABI High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems Inc., Foster City, CA) from 25 ng of purified RNA. qPCR was performed with gene-specific primer probe fluorogenic exonuclease assays (TaqMan, Life Technologies, Waltham, MA, Supplemental table 5) and the QuantStudio 12 K Flex Real-Time PCR System (Applied Biosystems). Relative gene expression (RQ) was calculated with Expression Suite v 1.0.3 software using the 2−ΔΔCt analysis method with GAPDH as an endogenous control. Statistical analysis of the qPCR data was performed using GraphPad Prism 8 (San Diego, CA). Two-way ANOVA analyses were performed followed by the Tukey’s multiple comparison test (p < 0.05).

Library Construction and RNA Sequencing (RNA-seq)

Illumina’s TruSeq Stranded mRNA Library Prep Kit (#20,020,594, Illumina) was used on 500 ng of total RNA for the preparation of strand-specific sequencing libraries according to manufacturer’s guidelines. As previously described [56], rRNA depletion was performed prior to library construction.

RNA was isolated from fresh-frozen hippocampal tissue of 12 mo FCG mice (n = 5–6/group), using Qiagen AllPrep DNA/RNA Mini Kit. After verifying RNA integrity numbers (RIN) with TapeStation (Agilent) and quantifying RNA with Qubit dsDNA Broad Range Assay kit (Invitrogen), 1 ug of RNA was used to construct RNA-Seq libraries using the Illumina TruSeq Stranded RNA Library Prep Kit, following the manufacturer’s guidelines. cDNA libraries were sized by TapeStation (Agilent) and quantified by qPCR (KAPA Biosystems). Libraries were then normalized to 4 nM, pooled, denatured, and diluted for sequencing on Illumina Hiseq2500 in a 2 × 100 bp fashion.

RNA-Seq Data Analysis

Following sequencing, reads were trimmed, aligned, differential expression statistics and correlation analyses were performed in Strand NGS software package (Agilent), as previously described [55]. Reads were aligned against the Mm10 build of the mouse genome (2014.11.26). Alignment and filtering criteria included: adapter trimming, fixed 2 bp trim from 5’ and 6 bp from 3’ ends, a maximum number of one novel splice allowed per read, a minimum of 90% identity with the reference sequence, a maximum of 5% gap, trimming of 3’ end with Q < 30. Alignment was performed directionally with Read 1 aligned in reverse and Read 2 in forward orientation. Reads were filtered based on the map** status and only those reads that aligned normally (in the appropriate direction) were retained. Normalization was performed with the DESeq algorithm [132]. Transcripts with an average read count value > 20 in at least 100% of the samples in at least one group were considered expressed at a level sufficient for quantitation per tissue. Those transcripts below this level were considered not detected/not expressed and excluded, as these low levels of reads are close to background and are highly variable. A fold change >|1.25| cutoff was used to eliminate those genes which were unlikely to be biologically significant and orthogonally confirmable due to their very small magnitude of change. For statistical analysis of differential expression, a two-way ANOVA with the factors of sex chromosome complement and gonadal sex and a Benjamini–Hochberg Multiple Testing Correction (BHMTC) with an FDR of 0.1. Visualizations of hierarchical clustering and principal components analysis were performed in Strand Next Generation Analysis Software (NGS) (Version 3.1, Bangalore, India). The entirety of the sequencing data is available for download in FASTQ format from NCBI Gene Expression Omnibus (GEO) under accession number GSE184098.

Bisulfite Amplicon Sequencing (BSAS)

BSAS for specific gene promoters was performed as previously described [110, 128]. Briefly, FCG hippocampal DNA (200 ng) was bisulfite converted with the EZ DNA Methylation-Lightning Kit (#D5030T; Zymo Research, Irvine, CA), according to the manufacturer’s guidelines. Bisulfite-PCR was conducted using primer sets (Integrated DNA Technologies; Table 1) designed against specific genomic regions with Methyl Primer Express v1.0 software (Thermofisher Scientific). PCR amplicons were cleaned with AmpureXP beads (#A63882; Beckman Coulter Life Sciences, Indianapolis, IN) using size selection with 0.7 × bead ratio. Following clean-up, the amplicons were quantified using QubitTM dsDNA HS assay kit (#Q32851; Thermofisher Scientific) and pooled. One ng of the pooled amplicons was used for library construction with the Nextera XT DNA library preparation kit (#FC-131–1096; Illumina, San Diego, CA), according to the manufacturer’s instructions. Libraries were quantified with Qubit dsDNA HS assay kit and TapeStation HD1000, normalized to 1 nM or 4 nM, and pooled for sequencing. Pooled libraries were then sequenced on iSeq or MiSeq (Illumina) at loading concentrations 35 pM or 8 pM, respectively. Fastq files were aligned to amplicon sequences in CLC Genomics Workbench 11.0 (Qiagen) using the “Map Bisulfite Reads to Reference” feature. Site-specific CpG (CG) and CH methylation percentages were extracted for downstream analysis.

Library Construction and Oxidative Bisulfite Sequencing (WGoxBS-seq)

DNA was isolated from fresh-frozen hippocampal tissue of 12 mo FCG mice (n = 3/group), using Qiagen AllPrep DNA/RNA Mini Kit. Whole genome oxidative bisulfite sequencing libraries were prepared according to manufacturer’s guidelines (Ovation Ultralow Methyl-Seq Library System, Tecan Genomics, Inc., Redwood City, CA) as previously described [55, 110]. Briefly, 1 µg of gDNA in 50 µl 1X low-EDTA TE buffer was sheared with a Covaris E220 sonicator (Covaris, Inc., Woburn, MA) to an average of 200 base pairs. Sheared products were sized by capillary electrophoresis (DNA HSD1000, Agilent) and cleaned using an Agencourt bead-based purification protocol. After quantifying (Qubit dsDNA, Thermofisher Scientific) the cleaned DNA, 200 ng gDNA fragments were prepared in a 12 µl volume to which 1 µl of spike-in control DNA (0.08 ng/ul) with known levels of specific mC, hmC, and fC at individual sites was added. End repair, ligation of methylated adaptors (#L2V11DR-BC 1–96 adaptor plate, Tecan Genomics) and final repair were performed according to manufacturer’s instructions. Normalized DNA was oxidized and then bisulfite- converted with the True Methyl oxBS module (NuGEN) with desulfonation and purification. 22 µl of libraries were eluted from the magnetic beads. qPCR was used to determine the number (N) of PCR cycles required for library amplification. Oxidative bisulfite-converted samples were amplified for 13 cycles [95 \(^\circ\) C- 2 min, N (95 \(^\circ\) C-15 s, 60 \(^\circ\) C-1 min, 72 \(^\circ\) C-30 s)]. Amplified libraries were purified with Agencourt beads and eluted in low-EDTA TE buffer. Capillary electrophoresis (TapeStation HSD1000, Agilent) was used to validate and quantify libraries. Amplified libraries were normalized to a concentration of 4 nM and pooled, denatured, and diluted to 12 pM for sequencing on NovaSeq 6000 (Illumina) according to manufacturer’s guidelines with the exception of a custom sequencing primer (MetSeq Primer) that was spiked in with the Illumina Read 1 primer to a final concentration of 0.5 µM.

OxBS-seq Data Analysis

Global levels of mCG, hmCG, and mCH were analyzed as previously described [55, 110] Before aligning, paired-end reads were adaptor-trimmed and filtered using Trimmomatic [133] 0.35. End-trimming removed leading and trailing bases with Q-score < 25, cropped 4 bases from the start of the read, dropped reads less than 25 bases long, and dropped reads with average Q-score < 25. Unpaired reads after trimming were not considered for alignment. Alignment of trimmed OxBS-converted sequences was carried out using Bismark [134] 0.16.3 with Bowtie 2 [135] against the mouse reference genome (GRCm38/mm10). Bams were de-duplicated using Bismark. Methylation call percentages for each CpG and non-CpG (CH) site within the genome were calculated by dividing the methylated counts over the total counts for that site in the oxidative bisulfite—converted libraries (OXBS). Genome-wide CpG and CH methylation levels were calculated separately. BAM files generated during alignment were run through methylKit in R [136] to generate context-specific (CpG/CH) coverage text files. Bisulfite conversion efficiency for C, mC, and hmC was estimated using CEGX spike-in control sequences. Untrimmed fastq files were run through CEGX QC v0.2, which output a fastqc_data.txt file containing the conversion mean for C, mC, and hmC. The entirety of the sequencing data is available for download in FASTQ format from NCBI Gene Expression Omnibus (GEO) under accession number GSE184013.

DMR Analysis

CpG text files were read into methylKit [136] and converted to an object. The mouse genome was tiled in 1000 nt non-overlap** windows. Each window was filtered for a minimum count of 10. Samples were then united and compared for windows covered in at least two samples per group. Differentially methylated regions (DMRs) were called using default parameters. DMRs were filtered to differences that were > 10% different between at least two groups and had a SLIM-generated q-value less than 0.05. The methylDiff object was intersected with the methylBase object to calculate the % methylation for each window that passed the described filtering. Post-hoc assessment of main and interactive effects of chromosomal and gonadal sex was conducted using a Two-way ANOVA with Bonferonni correction for six pairwise comparisons (p < α = 0.0083).

Software Usage for Tertiary Analysis of Transcriptomic and Epigenomic Data

Over-representation analysis (ORA) for GO Biological processes, transcription factor networks, and KEGG pathways was conducted using WEB-based GEne SeT AnaLysis Toolkit (WebGestalt, www.webgestalt.org) [137]. EnrichedHeatmap [138] in R was used to visualize DNA methylation levels across ChrX and autosomal CGI, Gene body/promoter, and CTCF binding sites. Bed files for genomic features were downloaded from the UCSC table browser [139] and visualized with custom tracks using the UCSC genome browser (http://genome.ucsc.edu) [140]. BEDtools [141] was used to intersect mCG/CH text files with specific genomic features. TrawlerWeb [142] was used to identify over-represented motifs from BED files. STRING protein–protein association networks [81] were used to determine potential protein interactors with SP1. Venny [143] and BioVenn [144] were used to compare gene lists and make Venn diagrams. BedSect [145] was used to create Jaccard distance matrices comparing genomic features to differentially methylated and expressed regions. ggplot2 [146] and clusterProfiler [147] were used to generate dotplots of GO biological processes.