Translating the Transcriptome: Sex Differences in the Mechanisms of Depression and Stress, Revisited

Abstract

The past decade has produced a plethora of studies examining sex differences in the transcriptional profiles of stress and mood disorders. As we move forward from accepting the existence of extensive molecular sex differences in the brain to exploring the purpose of these sex differences, our approach must become more systemic and less reductionist. Earlier studies have examined specific brain regions and/or cell types. To use this knowledge to develop the next generation of personalized medicine, we need to comprehend how transcriptional changes across the brain and/or the body relate to each other. We provide an overview of the relationships between baseline and depression/stress-related transcriptional sex differences and explore contributions of preclinically identified mechanisms and their impacts on behavior.

Explorations of sex differences in the brain began in the 19th century. Initial studies examined sex differences in region, size, and neuronal complexity of postmortem tissue. Many emphasized the smaller size of the female brain and mistakenly interpreted this difference as evidence of reduced cognitive capacity (1,2). As women already know, size does not matter, and the 10% smaller size of the female brain is a function of having a smaller body (3). As imaging technology developed, investigators examined sex differences in gray matter and white matter volumes, localized activity, and connectivity throughout the brain (3–5). Questions have been raised about the legitimacy of some of these data, given that the authors often did not correct for brain volume or state the effect size (6). The validity of emphasizing differences between the sexes, rather than similarities, was also raised as an issue (6). Furthermore, some argued that studies failed to capture the multivariate nature of social, psychological, and cultural factors contributing to the behavioral and neural differences of the sexes (6,7). A new era of studies that include trans-as well as cisgender people (8–10) and a growing body of work examining individuals with sex chromosome aneuploidy (11–16) may provide a greater understanding of the interactions between societal factors, sex similarities, and sex differences in humans.

While it is vital for studies on sex differences to be contextualized and for findings to be framed in a conscientious manner, these considerations do not delegitimize the need for understanding the biological impact of sex differences in the brain and body. Taking the study of sex differences outside of societal contexts by utilizing preclinical models has revealed a vast array of sex differences, and recent studies have demonstrated the impact of sex on cellular and molecular function (17–24). Several articles have demonstrated striking preclinical sex differences in transcriptional and epigenetic signatures across development and in response to steroid hormones or stress (25–30). These sex differences in transcriptional profiles extended to humans, with studies of postmortem tissue providing support for convergent sex differences that result in the same end points via different mechanisms (17,31–33). The role of these sex differences in the transcriptional signature of mood disorders in humans (34,35) is covered in depth by Seney et al. (36). Because of a policy shift by the National Institutes of Health to include sex as a biological variable (32), our understanding has broadened to include the myriad ways in which sex factors into the responses of our brains and bodies. This review addresses our growing knowledge of how baseline sex differences and those altered by depression in humans overlap with preclinical data to identify mechanisms that may contribute to sex-specific aspects of mood disorders.

BASELINE TRANSCRIPTIONAL SEX DIFFERENCES

There is a great deal of variability in estimates of the number of genes that are differentially regulated in men and women under typical conditions. Reports range from 0.2% to 33% of the genes of the human body and brain (37–39). This wide range is in part because of the use of different methods, tissue types, and number or age of subjects in studies. In a gene expression study across human tissues, 47% of the variance was due to tissue type, compared with 4% due to the individual (40). Although the specific differentially expressed genes in the brain vary by region, sex differences in expression are reported for all regions and tissues, both at baseline and in people with depression (11,34,35,39,40). This variability suggests that we cannot make assumptions that the differences seen in one region of the brain or the body are applicable to other regions and tissue types.

Interpreting an increase in transcriptional activation outside the context of behavior is difficult and can be misleading. An excellent recent study upends assumptions about the relationship between transcriptional sex differences, gray matter volume, and behavior (3). The authors report that cortical regions with high expression of sex-linked genes had increased gray matter volume in men, whereas regions with low numbers of sex-linked genes had higher gray matter volume in women. In men, increased gray matter volume was identified in areas of the brain associated with recognition of faces or emotion. However, women generally outperform men in the behavioral tasks that utilize these brain areas. In particular, women are better than men at recognition of faces and identifying emotions (41,42). Therefore, these data suggest an inverse relationship between gray matter volume and ability. It is likely that increased pruning in these areas results in decreased gray matter volume and produces a better signal-to-noise ratio in females, similar to the improved performance that occurs with cortical pruning during adolescence (43). These effects may also depend on the cellular makeup of the gray matter. In the brain areas that had increased gray matter volume in men, differentially regulated gene signatures were associated with neuronal identity, including axonal growth, neuronal synapses, synapse organization, and GPCR (G protein–coupled receptor) signaling (3). Areas with higher gray matter volume in women included regions important for executive function and working memory. Differentially expressed genes in these areas were enriched for terms related to the regulation of the extracellular matrix, organelles, immune processes, regulation of cell proliferation, axon guidance, and neuron recognition (3). In behavioral tasks associated with these areas, women either show no difference from men (44,45) or perform better in tasks such as working memory (46–48), suggesting that increased gray matter volume may be beneficial when the transcriptional sex differences primarily influence non-neuronal cells or processes.

As sequencing methods are advancing, sex differences within cell types are becoming apparent (49). A recent study utilizing single-cell sequencing data from endothelial cells of male and female mice found that sex-specific subclustering occurred more frequently in some types of tissue (brain, lung, aorta) than others (adipose tissue, heart, kidney) (49). Given that sex-specific subclusters of endothelial cells occurred in the brain and given the emerging role for endothelial cell–dependent blood-brain barrier permeability in depression/stress effects (50,51), these data raise the possibility that stressful experiences have different transcriptional impacts in male subjects than in female subjects, depending on cell type as well as region/organ. A recent study in male mice on transcriptional analysis of nucleus accumbens (NAc) endothelial cells revealed signatures of stress susceptibility and resilience (51). A transcriptional repressor of claudin 5 and Hdac1 (histone deacetylase 1) were reduced in the NAc endothelial cells of resilient compared with stress-susceptible mice, and increased expression was verified in the NAc of humans with major depressive disorder (MDD) (51). Pharmacologically suppressing HDAC promoted behavioral resilience, indicating that an epigenetic mechanism contributed to the behavioral response to stress by increasing permeability of the vasculature in some individuals but not others. Little is known about sex differences in stress-induced changes in endothelial cells; however, there are known sex differences in the function of endothelial cells and the vasculature of humans that change across the life span (52).

Baseline sex differences may exist to permit the brain to function similarly in male and female individuals despite the very different peripheral environments produced by the hormonal, metabolic, and immune milieus (32). Although knowledge of baseline and disease-based sex differences is becoming increasingly intricate, thereby further complicating interpretation, progress in the field highlights the importance of distinguishing pathologically relevant sex differences from changes that are due to typical sexual differentiation.

SEX DIFFERENCES IN STRESS-BASED ANIMAL MODELS

The concept of stress originated from physics to describe how an external force acts on material and was used by Cannon to describe the disruption of homeostasis within the brain or body of an individual from external and internal stimuli (53). The term “stress” was further adapted by Selye to describe the biological responses of an individual to emotional or physical threats (54). With the discovery of the hypothalamic-pituitary-adrenal (HPA) axis and the subsequent negative feedback effects on glucocorticoid receptors in brain structures, stress became synonymous with the hormones secreted during HPA activation (55,56). HPA axis activation during acute stress is a dynamic adaptive response providing energy to the individual and a return to homeostasis via negative feedback. It was proposed that chronic stress exposure alters sensitivity of the system through a shift in the allostatic load leading to dysregulation and an altered baseline, thus contributing to the development of depression (53). It has been recognized since Selye’s groundbreaking work in the 1930s that exposure to noxious stimuli impacts many aspects of biology, including the immune, digestive, and gonadal systems, all of which intertwine in their functions (57). Subsequent studies have examined the effects of acute or chronic exposure to stressors on various regions of the brain and their transcriptional responses to the stimuli (28,34,58–61).

However, stressful experiences do not model mental illness. In rodents, the effects of stress can recapitulate some of the patterns of transcriptional sex differences identified for MDD and have been used to identify brain areas or circuits functionally implicated in behavioral susceptibility to stressors (28,34,61–63). Many studies of stress-induced changes in transcription are performed on areas of the brain initially identified as being part of a stress susceptibility/resilience circuit (Figure 1); however, it should be noted that all initial preclinical studies were performed on male subjects (58,61,64–70). Recently, the functional roles of some of these brain regions have been explored in female subjects using electrophysiology, neuronal morphology measures, and optogenetics, identifying some sex differences and similarities (71–78) (Figure 1).

Figure 1.

Circuitry of susceptibility and resilience in male and female mice. The social defeat model was used to identify stress susceptibility/resilience circuits, including connections between the PFC, NAc, VTA, ILT, amygdala, and hippocampus (58). In male mice exposed to SD, increased activation of glutamatergic afferents from the PFC and amygdala to the NAc promote resilience (65–67,69). Increased activation of the ILT, ventral hippocampus, or burst firing of the VTA increased susceptibility (65,66,142). In female mice, exposure to 6 days of SCVS increased activation of a pathway from the LHb to the VTA, but burst firing between the VTA and NAc was not increased (143). Similar to stress-susceptible males, females had increased activation of glutamatergic projections from the ventral hippocampus to the NAc, and testosterone was found to reduce firing, promoting resilience in males (71). Although it has not been directly tested in females, the glutamatergic projections from the PFC to the NAc may also contribute to stress susceptibility or resilience as VGLUT1, a presynaptic marker for forebrain projections, was decreased in the NAc of female mice susceptible to SCVS (78). VGLUT2, which represents projections from the ILT, was increased, similar to susceptible males after SD (66,78). ARC, arcuate nucleus of the hypothalamus; C-P, caudate putamen; DMT, dorsomedial thalamus; DR, dorsal raphe nucleus; GABA, gamma-aminobutyric acid; HYP, hypothalamus; IC, inferior colliculus; ILT, intralaminar thalamus; LC, locus coeruleus; LH, lateral hypothalamus; LHb, lateral habenula; NAc, nucleus accumbens; PAG, periaqueductal gray; PFC, prefrontal cortex; SC, superior colliculus; SCVS, subchronic variable stress; SD, social defeat stress; SNr, pars reticulata of substantia nigra; VP, ventral pallidum; VTA, ventral tegmental area. Figure adapted from (58).

Sex differences in transcriptional signatures of multiple forms of stress in rodents have been examined and compared with humans with MDD. These forms of stress in rodents include prenatal/early-life stress [see Parel and Peña (79) for more on early-life stress], adult variable stress/chronic mild stress, adult social isolation, and social defeat stress (34,62,63,80,81). A comparison of stress models in mice with each other and with humans with MDD found that variable stress and social isolation each replicated approximately 20% of the transcriptional changes due to depression in humans for the prefrontal cortex (PFC) and NAc (62). Social defeat stress recapitulated approximately 4% of the changes in gene expression, but social stress or social isolation both mapped the relationship of genes to each other associated with depression better than variable stress. In this study, only variable stress included female subjects, even though the human participants included both sexes. This study demonstrates that specific patterns of transcription are regulated by the different forms of stress indicating that not all stress models have the same impact on the brain (Figure 2). Forms of social stress (including isolation) that engage repeated exposure to the same stimuli are likely activating different processes than variable stress. Even within models of social stress, exposure to a novel aggressor has a very different transcriptional signature than isolation from conspecifics. Yet, all of these forms of stress in animals have some relevance to the changes we see in the brains of people who have experienced depression (Figure 2).

Figure 2.

Different types of stress reproduce different transcriptional/brain region–specific signatures of depression. (A) Based on published data (62), different forms of stress in mice account for different aspects of depression (combined men and women) as indicated by significant overlap in genes and gene ontogeny terms. (B) When transcriptional signatures of depression are segregated by sex, different gene ontogeny patterns arise with some overlap to the combined set [from PFC data (35)]. (C). Six days of variable stress produces changes in immune pathways in the NAc of female mice (pink) compared with males (blue). Removing Dnmt3a before stress exposure, which makes females behaviorally resilient, produces a more male-like transcriptional pattern (purple) (28). cAMP, cyclic adenosine monophosphate; CRF, corticotropin-releasing factor; GCE, germ cell–expressed protein; GPCR, G protein–coupled receptor; IL, interleukin; ILK, integrin-linked kinase; KO, knockout; MAPK, mitogen-activated protein kinase; NAc, nucleus accumbens; PFC, prefrontal cortex; RA, rheumatoid arthritis; TSP, thrombospondin.

MECHANISMS CONTRIBUTING TO SEX DIFFERENCES

The greatest benefit of using animal models is the ability to directly manipulate potential mechanisms of psychiatric illness and examine the biological and behavioral results. A number of differentially expressed genes and processes in males and females have been identified through overlap between human and rodent sequencing data. Overexpressing, blocking, and mimicking the function of these genes has identified a series of potential sex-specific treatment targets (28,34,81–84). Many of these individual genes have widespread effects through their transcriptional interactions and impacts on development. Furthermore, many of these mechanisms involve interactions between signals from the body and the brain and will be detailed below.

Hormonal Sex

Cells in the brains and bodies of both male and female subjects express multiple progesterone, androgen, and estrogen receptors (ERs), which are active regulators of transcription throughout development and into adulthood (85–90). These steroid hormones can be produced by the gonads as well as locally synthesized within the brain and other tissues (91–95). Transcriptional effects of steroids occur either directly by nuclear receptor activation or indirectly through membrane-localized receptor-initiated intracellular signaling pathways that modulate a variety of transcription factors and regulators (96). Actions of steroids via their nuclear receptors are necessary and sufficient for sexual differentiation of the brain during critical periods of development (97). In male rodents, this differentiation is due to the aromatization of testosterone to estradiol in the brain following the perinatal androgen surge (98,99). The organizational effects of testosterone and estrogens during early development and puberty result in permanent epigenetic changes that dramatically affect adult behaviors (100,101). Adult women have much higher circulating concentrations of estrogens than men, while men express higher levels of testosterone, which contribute to some differences in gene transcription (102). In addition, sex differences in steroid hormone receptor expression in various brain nuclei further increase the divergence in the transcriptome between sexes (103). It has been suggested that differential transcriptional regulation by ERα in particular may contribute to the increased incidence of mood disorders in women and developmental psychiatric disorders in men (104).

Chromosomal Sex

Chromosomal sex is the most obvious, yet understudied, mechanism of baseline and mental health–associated sex differences. Using publicly available sequencing data from previously published papers (3,38–40), we generated a list of genes (Table 1) that were regulated by depression within sex (34) and were consistently sexually differentiated at baseline in men and women across brain regions in at least two separate datasets. One possibility is that depression is shifting transcription in specific brain regions toward male or female extremes. Alternatively, depression may be neutralizing transcriptional sex differences. Using this filter, 24 genes across different brain regions were regulated by depression and also demonstrated baseline sex differences. Overall, depression drove male and female transcriptional patterns toward each other rather than emphasizing a more male or female pattern of expression. However, there were some gene- and region-specific differences that became more sex extreme in brain tissue from people with depression. More than half of the genes (57%) resided on the sex chromosomes. The majority (41%) resided on the X chromosome, whereas 16% were Y linked (Table 1). Half of the X-linked genes were differentially regulated by depression in men, 20% were relevant to both sexes, and 30% were relevant only to female depression. In humans, sex chromosome aneuploidies contribute to the development of cognitive and mood-related issues and structural differences in the brain (13–15,105–107). People with sex chromosome aneuploidies also have an increased risk of developing autoimmune diseases, which in turn increase the risk of developing depression (15). The X and Y chromosomes share conserved homologous sequences via the two pseudoautosomal regions (15). However, the X and Y chromosomes differ in the number of genes that reside on them, and there are potential dosing effects owing to genes that escape X inactivation (108–111). Of the 10 genes listed that reside on the X chromosome, 7 (70%) reportedly escape X inactivation in humans (112), and 2 (Kdm5C, Kdm6A) also escape X inactivation in mice (113). It should be noted as a caveat that there are differences between individuals, tissues, cell types, and even neighboring cells of the same type in whether a gene undergoes X inactivation (113–115). Much future research is needed to delineate the importance of X inactivation or parent-of-origin effects with respect to stress and depression in a variety of cell types and brain nuclei.

Table 1.

Genes Identified as Being Consistently Significantly Regulated by Depression Within Sex and at Baseline Between Controls Across Brain Regions and Datasets

Gene Name	Women MDD vs. Women Control Log2 Fold Change; Region	Men MDD vs. Men Control Log2 Fold Change; Region	Baseline Sex Differences Women vs. Men Log2 Fold Change; Region	Chromosome
CCDC85C	−0.304361892; BA 25 (shift male)	0.261311924; AI (shift female)	0.400825381; BA 25 F > M 0.341141733; AI F > M	14
EIF1AX * XI Escape Human (112)	0.110018247; BA 11 MDD > C (shift male)		−0.013466896; BA 11 (ns) F < Ma	X
FRG1B	0.452924542; BA 25 (shift male) 0.446263541; NAc (shift male) 0.385279314; AI (shift male)		−1.079547872; BA 25 F < M −0.83569602; NAc F < M −1.0142453; AI F < M	20
GYG2P1		−0.332080232; BA 11 (shift female) 0.340555426; NAc (shift male)	−5.894129326; BA 11 F < M −6.153769643; NAc F < M	Y
HAUS7	0.510195991; AI (shift male)		−0.164991303; AI (ns) F < Ma	X
JPX * XI Escape Human (112)		−0.256263298; BA 25 (shift female)	0.433801925; BA 25 F > M	X
KBTBD4	0.172606953; BA 11 (shift male)		−0.075256756; BA 11 (ns) F > Ma	11
KDM5C * XI Escape Human (112) Mouse Brain (113) T Cells (144)	−0.11464424; BA 8/9 (shift male)	0.102395131; BA 25 (shift female)	0.589381813; BA 8/9 F > M 0.544810211; BA25 F > M	X
KDM6A * XI Escape Human (112) Mouse Brain (113) T Cells (144,145)		−0.164522189; BA 8/9 (shift male) −0.119496151; BA 11 (shift male) −0.217563793; BA 25 (shift male)	0.188654978; BA 8/9 F > M 0.171845366; BA 11 F > M 0.091158594; BA 25 (ns) F > Ma	X
MAP3K2		−0.158048477; BA 8/9 (shift female) −0.193175326; BA 25 (shift female)	−0.217563793; BA 8/9 F < M −0.119496151; BA 25 F < M	2
NLRP2	−1.211819093; AI (shift male)		1.909302773; AI F > M	19
NOX5		0.87301923; NAc (shift male)	−0.126404288; NAc (ns) F < Ma	15
PRKY		0.464557516; sub (shift male)	−6.15334242; sub F < M	Y
RHCG	−0.642745043; NAc (shift male)	0.454371835; BA 25 (shift female)	0.574813858; NAc F > M 0.676497199; BA 25 F > M	15
RP4-610C12.4	0.907619344; BA 8/9 (shift male) 0.872637689; BA 11 (shift male)		−1.966238413; BA 8/9 F < M −2.152374342; BA 11 F < M	20
RPSRX	0.212638844; BA 11(shift female)	0.18985862; BA 8/9 (shift female) 0.249427979; BA 11 (shift female)	0.456647396; BA 8/9 F > M 0.112628114; BA11 (ns) F > Ma	X
SESN1	0.200747709; BA 8/9 (shift female) 0.204249013; BA 11 shift female)		−0.056358668; BA 8/9 (ns) F < Ma −0.069093794; BA 11 (ns) F < Ma	6
SH3RF1	−0.130754587; NAc (shift female)	−0.160976932; BA 11 (shift female)	−0.037127458; NAc (ns) F < Ma −0.088250501; BA 11 (ns) F < Ma	4
SMC1A * XI Escape Human (112)	−0.313368068; BA 11 (shift male) 0.121498044; NAc (shift male)		0.348078257, BA 11 F > M 0.006943817; NAc (ns) F > Ma	X
TXLNG2P		−0.307930074; BA 25 (shift female)	−9.91100221; BA 25 F < M	Y
UTY		−0.257859356; BA 25 (shift female)	−9.302126143; BA 25 F < M	Y
XIST * XI Escape Human (112) Mouse Brain (113)		1.454434539; BA 8/9 (shift female) 1.572860015; NAc (shift female) 1.455407087; AI (shift female)	10.85002337; BA 8/9 F > M 10.17245709; NAc F > M 10.89616806; AI F > M	X
ZFX * XI Escape Human (112)		−0.184663563; BA 25 (shift male)	0.37969668; BA 25 F > M	X
ZRSR2 * XI Escape Human (112)		0.240140657; NAc (shift female)	0.283932842; NAc F > M	X

We identified 24 genes differentially regulated by MDD compared with same-sex control subjects (34) who were also regulated by sex at baseline in women vs. men (control subjects) in brain tissue from one of the following datasets (3,34,38–40).

AI, anterior insula; BA, Brodmann area; BA 25, subgenual anterior cingulate cortex; BA 11, orbitofrontal cortex; BA 8/9, medial prefrontal cortex; F, female; M, male; MDD, major depressive disorder; NAc, nucleus accumbens; ns, not significant; XI, X inactivation.

^aAll listed genes were significantly regulated by sex unless noted as ns.

Sex chromosomes also contribute to many of the identified alterations in pathways associated with depression and stress, even when the gene of interest does not reside on the X or Y chromosome. Pathways consistently identified with sex differences in depression and stress paradigms include immune function, mitochondria structure/function, synaptic plasticity, and epigenetic changes (Figure 2). The X chromosome contains more immune-related genes than any other chromosome (116), and the Y chromosome contains genes that epigenetically regulate genes associated with the immune system (117). The nonreproductive role of the Y chromosome has been highlighted as contributing to cardiac and immune disease states (118) that may also impact the risk of developing depression.

A number of new studies have examined the importance of the sex chromosome complement, made possible in part by the development of the four core genotypes mouse model. This model allows the dissociation of the chromosomal sex complement from gonadal development through manipulations of the Sry gene (119,120). Use of this model in combination with chronic mild stress identified effects on dopamine-, GABA (gamma-aminobutyric acid)-, and glutamate-associated genes in the PFC that were limited to XX mice, regardless of gonadal state, and were not significantly altered in XY mice (63). These data support work that demonstrates alterations in the GABA gene network specifically in women with depression (121,122). Additional studies on the transcriptional signatures of four core genotype mice following stress have identified multiple immune pathways across brain regions as being regulated by genetic sex versus gonadal sex (123). Genetic sex-regulated genes were associated with the adaptive immune system, in particular with T-cell and B-cell development, and signaling and may reflect changes in perivascular cells as well as the parenchyma. These data indicate that chromosomal effects of sex that contribute to stress-related behavior, and potentially depression, affect far more than neuronal networks in the brain. Even when examining changes in transcription within brain structures, many of the cell types affecting transcriptional function may originate outside of the brain.

Immune Mechanisms

Many of the sex-based transcriptional changes related to stress and depression throughout the body involve the immune system; both sex-linked genes and steroid hormones contribute to sex differences in the immune system (124). Similar to what has been reported in other types of tissue, immune cells cluster more by cell type than by sex (125). Identification of peripheral pan-immune sex signatures in mice produced a list of 14 genes, in which 3 resided on the X chromosome and 11 were on the Y chromosome. Only Xist overlapped with the sexually differentiated genes listed in Table 1, suggesting that baseline transcriptional sex differences in blood do not reflect transcriptional sex differences in the brain. Alternatively, this may be due to species, strain, or rearing differences (126–129). Within blood, the majority of autosomal sex differences were found in macrophages. Of the 41 genes significantly sexually differentiated, 26 (63%) were expressed at higher levels in females and 15 genes (36%) were at higher levels in males. In contrast, a study that activated male and female macrophages with interferons found far more genes significantly upregulated in female and male mice compared with same-sex unstimulated control mice. Of these, 200 were upregulated in both sexes, and 53 genes (18%) were female specific, whereas 42 (14%) were only upregulated in males (125,130).

Epigenetic Mechanisms

Genetic and hormonal sex can interact with epigenetic mechanisms to produce lasting effects of stress and/or sex on the transcriptome (25,131–133). These mechanisms include histone modifiers, microRNAs (miRs) and DNA methyltransferases (DNMTs), enzymes that add methyl groups to cytosines resulting in either maintenance or de novo methylation (131,132,134). DNMT expression and activity during development shape adult sexual responses and play an active role in feminizing and/or masculinizing the brain (27,135). Inhibiting Dnmt3a expression or DNMT activity in females during early postnatal development was sufficient to trigger male sexual behavior in adulthood when activated by testosterone (27). Sex-specific epigenetic alterations potentially occur in the amygdala in early development, during which higher expression of DNMT3a is present only in the female amygdala on postnatal day 1, but no sex difference remains detectable by postnatal day 10 (136). There were no detectable sex differences in expression in the ventromedial hypothalamus, preoptic area, or hippocampus throughout postnatal development and into adulthood (137).

Recent studies indicate that testosterone affects DNMT activity to cause some, but not all, known sexual dimorphisms in the developing brain. Pharmacologically blocking DNMT activity with zebularine or adding testosterone produced male-like numbers of neurons in the medial preoptic area and bed nucleus of the stria terminalis of females and altered the number of ERα+ cells (135). Zebularine treatment did not block the effects of testosterone in females, supporting the concept that testosterone alters DNMT activity (135). Blocking DNMT activity did not mimic the effects of testosterone on the number of kisspeptin-positive cells in the periventricular nucleus of the hypothalamus, indicating that it is not the only mechanism leading to sexual dimorphism of brain regions. However, DNMT activity in kisspeptin-positive neurons later initiates female puberty (138). DNMT is necessary for the activation of the genes Cbx7 and Eed to permit the onset of vaginal opening and onset of estrous cycling. Blocking DNMT activity results in lasting changes in fertility (138).

In adult mice, increased expression of brain DNMT through transgenic or pharmacological manipulation results in increased behavioral vulnerability to subthreshold stress in both sexes and across different stress models (28,139). In male mice, increasing Dnmt3a expression in the NAc also produced changes in spine density that have been found following cocaine or stress exposure (139). Using the variable stress paradigm, we found that blocking Dnmt3a expression in the NAc resulted in increased behavioral resilience in females and a more male-like transcriptional profile (Figure 2) (28). Both variable stress and depression increased Dnmt3a expression in the NAc of mice and humans, respectively, albeit more so in female subjects than males (28).

The variable stress paradigm also produced sex-specific patterns of miRs in male and female mice, resulting in little to no overlap in the NAc (134). Pathway analysis of genes indicated that miRs affected immune genes to a greater extent in females exposed to stress, particularly genes related to the adaptive immune system and leukocyte migration. T- and B-cell signaling was the top conical pathway targeted by the sex-specific changes in miRs (134). Within the hypothalamus, miRs are also highly sexually dimorphic and contribute to transcriptional masculinization through estrogen response elements (133). Histone modifications may also contribute differently in males and females to the effects of stress (140). Histone posttranslational modifications directly contribute to masculinization during development by acting as a permissive switch allowing activation of the Sry gene, which in turn produces the sexual differentiation of the gonads (141). These data suggest that there is a reciprocal relationship between epigenetic modifications, sex chromosomes, hormones, and immune function that contributes to transcriptional and structural sexual mosaicism of the brain and the subsequent behavioral responses of an individual to stress.

CONCLUSIONS

In the past decade, there has been incredible growth in understanding how the transcriptional signatures of male and female individuals differ. This is revolutionizing how the field approaches treating disorders such as depression. As we develop new treatments, we should focus on a system-wide exploration of sex differences, rather than limiting work to a region-specific reductionist approach. Comprehension of how gene networks act together and impact the functions of the brain and body will provide a better framework for developing sex-specific personalized treatments for a variety of psychiatric disorders.

Recommendations for Future Research Into Exploring the Transcriptional Sex Differences of Mood Disorders

1. Inclusion of transgender people and those with sex chromosome aneuploidy, in addition to cisgender subjects, may provide additional insight into the biological factors contributing to sex differences in mood disorders.

2. Do not assume transcriptional changes are the same in both sexes, all tissue, or cell types. Effects of stress differ between cell types, brain regions, and tissue type and by sex.

3. Consider the importance and relevance of sex similarities as well as sex differences.

4. Not all stressors are the same; different stressors produce different physiological responses.

5. Drugs in humans are administered systemically; expanding our end points to include how stress or depression changes across regions of the brain and/or the relationship between the body and brain may increase the ability to translate from bench to bedside.