Introduction

Identifying molecular mechanisms that mediate individual differences in stress vulnerability and resilience has potential to advance our understanding of the pathogenesis of mental illness and reveal novel targets for the treatment or prevention of stress-related disorders. Abnormalities in glucocorticoid signaling have consistently been linked to the pathophysiology of psychiatric disorders, including depression and anxiety [1]. However, attempts to treat or prevent these disorders by directly targeting the glucocorticoid receptor (GR) have shown inconsistent efficacy [2], possibly due to the wide-ranging effects of GR activation on signaling pathways with diverse functions. Refining our understanding of how specific components of the glucocorticoid signaling cascade contribute to stress vulnerability and depression risk will thus be crucial for identifying improved molecular targets for intervention.

We previously found that expression of the GR target gene, Serum and Glucocorticoid-regulated Kinase 1 (SGK1), is increased in peripheral blood of unmedicated depressed patients and in the hippocampus of rats following prenatal stress [3]. However, whether SGK1 mechanistically contributes to stress-induced abnormalities in brain function and psychopathology is unknown.

SGK1 is a serine/threonine kinase that has been implicated in ion transport regulation [4, 5], neuronal excitability [6], cell proliferation and differentiation [3], and synaptic plasticity [7]. SGK1 expression is also increased in vitro following treatment of human hippocampal neurons with glucocorticoid hormones, and glucocorticoids suppress neurogenesis via an SGK1-dependent mechanism, suggesting a potential role for SGK1 in shaping abnormal neurodevelopmental trajectories under stress conditions [3].

The hippocampus plays a crucial role in determining individual differences in stress vulnerability [8,9,10] and antidepressant responses [11,12,13]. In rodents, chronic stress decreases adult hippocampal neurogenesis and increases the activity of granule cells in the ventral dentate gyrus (DG) region of the hippocampus – two mechanisms that increase susceptibility to stress-induced psychopathology in mice [8, 10]. Despite the role of SGK1 in mediating glucocorticoid effects on hippocampal development [3], it remains to be determined whether SGK1 is part of the mechanism by which stress reduces neurogenesis and increases DG activity in vivo.

Early life adversity (ELA) is a major risk factor for psychiatric disorders that we and others have found to be associated with long-term changes in hippocampal neurogenesis, neuroendocrine function, and heightened vulnerability to future stressors [14, 15]. While SGK1 levels in peripheral blood are associated with depression diagnosis [3], it remains elusive whether ELA may lead to long-lasting increases in SGK1 expression and whether such effects would indeed occur in the human hippocampus.

In addition to environmental factors, such as stress and ELA, genetic predisposition contributes to illness vulnerability and interacts with environmental exposures to determine an individual’s risk for developing a psychiatric disorder [16]. Whether genetic variants that determine differences in SGK1 expression in the brain contribute to stress susceptibility or depression risk has never been investigated. Identifying such genetic contributions could refine our understanding of differences in susceptibility to psychopathology and potentially serve as an early screening tool to predict individual risk for mental illness.

In this study, we used a translational approach to evaluate the effects of ELA and genetic vulnerability on hippocampal SGK1 expression and depression risk in humans, and we tested a functional role for SGK1 as a mechanism underlying stress effects on DG activity, neurogenesis, and behavior abnormalities using mouse models. By integrating data from human postmortem brain tissue, genetic modeling, hippocampal-specific knockdown, and pharmacological inhibition in mice, our study reveals a novel role for SGK1 as both a vulnerability factor for psychiatric disorders and as a novel target for therapeutic prevention of stress-induced disorders.

Materials and methods

Postmortem brain tissue

Postmortem hippocampus tissue from 50 male brains was obtained from the Douglas-Bell Canada Brain Bank with ethics approval from the institutional review board of the Douglas Mental Health University Institute and written informed consent from patients or family members, as appropriate. ELA was characterized based on adapted Childhood Experiences of Care and Abuse interviews assessing experiences of sexual and physical abuse, as well as neglect before age 16 (Supplementary Table 1) [17]. Please also see Supplementary Information.

Adolescent brain and cognitive development (ABCD) study

For analysis of expression-based polygenic risk scores (ePRS), we performed secondary analysis on data from the Adolescent Brain Cognitive DevelopmentSM Study (ABCD Study®, Release 4.0) [18, 19], approved by the institutional review board of Columbia University.

We included N = 8588 of 9 to 10-year old male and female children with child behavior checklist (CBCL) score outcomes (Table 1). Children reported on psychopathology including suicide attempts during a computerized version of the Kiddie Schedule for Affective Disorders and Schizophrenia (KSADS) which includes a Trauma “Life Events Checklist” [20] consisting of 17 early life adversity events that the child reports to have experienced or not. To calculate an ELA score, we added up the number of ELAs from the KSADS Life Events Checklist that the child reported and constructed a 5-category variable (0= no ELA, 1 = 1x ELA, 2 = 2xELAs, 3 = 3xELAs, and 4 = 4 or more ELAs). Please also see Supplementary Information.

Table 1 Demographics of ABCD study subjects.
Full size table

SGK1 expression-based polygenic risk score (ePRS)

To compute a hippocampus-specific ePRS for the SGK1 gene, we used biomaRt R package [21, 22] to define the location of the SGK1 gene from Ensembl. We then used the Genotype-Tissue Expression (GTEx Analysis V7) [23] data and filtered it to select all available SNPs located on the SGK1 gene and their association with SGK1 mRNA expression in hippocampus tissue based on the GTEx database. There were 473 SNPs associated with SGK1 gene expression that were located within the SGK1 gene. Among these 473 SNPs, 289 were available in the ABCD cohort. This list of 289 SNPs formed the base file to calculate polygenic scores for the ABCD sample using PRSice-2 [24]. SNPs were subjected to linkage disequilibrium clumping at 500 kb radius and r2 > 0.2. SNPs with missing genotypes were excluded from the score calculation, resulting in a final list of 23 SNPs. For each of the final 23 SNPs, we obtained the SNP weight, which is the estimated effect of the genotype on SGK1 mRNA expression in hippocampus tissue based on GTEx. These effects represent the estimated change in SGK1 expression in the hippocampus per additional copy of the effect allele. For every subject j, the SGK1 ePRS was calculated as the sum across all SNPs of genotypes weighted by the corresponding SNP effect:

$${{{\rm{SGK}}}}1\,{{{{\rm{ePRS}}}}}^{{{{\rm{subject}}}}\,{{{\rm{j}}}}}={\sum }_{{{{\rm{i}}}}=1}^{23}=({{{{{\rm{genotype}}}}}^{{{{\rm{subject}}}}\,{{{\rm{j}}}}}}_{{{{\rm{i}}}}}\times {{{\rm{SNP}}}}\,{{{{\rm{weight}}}}}_{{{{\rm{i}}}}})$$

This score reflects the estimated gene expression of SGK1 in the hippocampus for each subject.

Experimental mice

All animal procedures were conducted in accordance with the US National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals, and with the Institutional Animal Care and Use Committee (IACUC) at the New York State Psychiatric Institute (NYSPI). Mice were housed in groups of 3–5 per cage with free access to food and water on a 12:12-h light/dark cycle. Male C57BL6/J mice were obtained from JAX at 7 weeks of age, allowed to adapt to the animal testing facility for one week, and exposed to chronic CORT treatment or chronic social defeat stress (CSDS) at 8 weeks of age. Mice that underwent limited bedding and nesting (LBN) were bred in house and LBN and control litters were generated from the same breeding cohort. For hippocampal deletion of SGK1, SGK1-flox mice were bred homozygous for the floxed allele [25].

Limited bedding and nesting (LBN)

Litters were exposed to the limited bedding and nesting (LBN) paradigm, which causes fragmented and unpredictable maternal care as a form of ELA [14, 26]. Briefly, dams and their newborn litters were placed on a wire mesh grid and provided with 1/3 of nesting and bedding material from postnatal day (P)3-10. Mice were returned to standard-rearing conditions on P10. Standard-reared control cages were changed at P3 and P10 but otherwise left undisturbed. All litters were weaned at P21 and male and female brains extracted at P56.

Chronic corticosterone (CORT) treatment

CORT (C2505, Sigma Lot #SLCC6498, SLBT 2966) was administered at a concentration of 35ug/ml CORT per cage of five C57BL/6 J male mice ad libitum in drinking water with 0.45% beta-cyclodextrin (A405837, AmBeed, lot# A405837-005). Control mice received only the vehicle solution of 0.45% b-cyclodextrin in drinking water. Mice were treated with CORT or vehicle for 5 consecutive weeks.

Chronic social defeat stress (CSDS)

CSDS experiments were conducted as previously described [8, 10]. Please see Supplementary Information.

Behavioral assays

Please see Supplementary Information.

Virus injections

AAV8-CamKII-Cre and AAV8-CamKII-mCherry virus were bilaterally injected to target the full dorsal-ventral extent of the dentate gyrus. Please see Supplementary Information.

Immunohistochemistry and fluorescent microscopy

Please see Supplementary Information.

Drugs

Mice were injected intraperitoneally (i.p.) with the SGK1 inhibitor, GSK650394 (5 mg/kg; MedChem Express), 30 min before each defeat on each day of the 10-day CSDS paradigm. Please see Supplementary Information for details on GSK650394 quantitation in tissue.

Statistics

Data was assessed for normality using the Shapiro–Wilk test. Gene expression in human- and mouse brain tissue and mouse behavior data were analyzed using GraphPad Prism 9. SGK1 mRNA expression in control vs suicide hippocampus, in CORT- vs vehicle treated mice, and in SGK1 knockdown vs mCherry control mice was assessed using unpaired student’s t-test. Comparisons between SGK1 mRNA expression in human hippocampus of control, suicide without ELA, and suicide with ELA groups were analyzed using One-Way ANOVA with Tukey’s posthoc test. SGK1 mRNA in CSDS-exposed mouse hippocampus, and susceptibility/ resilience phenotypes were analyzed using One-Way ANOVA with Tukey’s posthoc test. Behavioral effects of LBN on SGK1 mRNA in adult male and female mice, and SGK1 knockdown and inhibition effects on behavior in CSDS-exposed mice were analyzed using Two-Way ANOVA with Tukey’s posthoc test.

Statistical analyses for SGK1 ePRS were performed using R (version 1.4.1717). Linear mixed effects models were applied using R-package “lme4” with a random effects term for family and site, since the ABCD study included siblings and twins. All continuous variables were standardized. All models were weighted by the American Community Survey 2011–2015 (ACS) weights provided by ABCD to ensure national representativeness of the sample. Child sex and age at interview were added as covariates to all models and all models included first 10 ancestry PCs to adjust for ancestry stratification.

Results

SGK1 expression is increased in the hippocampus of suicide completers

To determine how psychopathology affects SGK1 expression in the human hippocampus, we used postmortem brain tissue of male suicide decedents from the brain collection of the Douglas-Bell Canada Brain Bank at McGill University to compare hippocampal SGK1 mRNA levels between subjects who died by suicide and control subjects who died of natural causes (see Supplementary Table 1). We found higher SGK1 mRNA in the hippocampus of subjects who died by suicide (n = 36) compared to control subjects (n = 14) (**p = 0.0068; Fig. 1A). To test whether ELA contributes to SGK1 expression, we then separated our suicide group into subjects with and without reported ELA before age 16 [15]. Subjects who died of suicide with a reported history of ELA had the highest levels of SGK1 expression (F2,47 = 5.29, **p = 0.0085; Control vs Suicide/no ELA, p = 0.075; Control vs Suicide/ELA, **p = 0.002; Fig. 1B), indicating that SGK1 expression in the hippocampus of suicide decedents is amplified by a history of ELA.

Fig. 1: Hippocampal SGK1 is increased in suicide decedents with a reported history of ELA.
figure 1

A SGK1 mRNA is higher in postmortem hippocampus of subjects who died by suicide compared to non-suicide controls. B Hippocampal SGK1 is highest in suicide decedents with a history of ELA before age 16.

Full size image

Genetic risk alleles for increased hippocampal SGK1 expression predict depression severity

Because we found an association between SGK1 expression and ELA in the hippocampus of suicide decedents, we asked whether individuals with genetic variants linked to high hippocampal SGK1 expression might be more vulnerable to the effects of ELA on early-onset depression. To this end, we used a large population sample of children from the Adolescent Brain and Cognitive Development (ABCD) study [18, 19] (n = 8588) for which we generated expression-based polygenic risk scores (ePRS) based on single nucleotide polymorphisms (SNPs) in the SGK1 gene (see Table 1, Supplementary Table 2) [27]. This ePRS sums each subject’s number of effect alleles for the SNPs in the SGK1 gene, weighted by each SNP’s association with hippocampal SGK1 expression derived from the GTEx database [23]. A higher ePRS therefore indicates greater predicted hippocampal SGK1 mRNA levels (Fig. 2A). Using this method, we found that depressed children have higher hippocampal SGK1 ePRS compared to children without depression (β = 0.03, t(8397) = 2.76, **p = 0.0059, n = 8,130 without KSADS depression diagnosis, n = 458 with KSADS depression; Fig. 2B) and that higher hippocampal SGK1 ePRS is associated with more depressive symptoms (β = 0.03, t(8555) = 2.58, *p = 0.01, Fig. 2C) as measured by the Child Behavior Checklist (CBCL). Sex did not moderate these associations (p > 0.14), indicating that the direction of effect was similar for both boys and girls. When we stratified the data by sex as a biological variable, we found that the association between hippocampal SGK1 ePRS and depression diagnosis was marginally significant for boys (p = 0.07), and significant for girls (*p = 0.01); while the association between SGK1 ePRS and depressive symptoms was only significant for boys (*p = 0.02; Supplementary Table 3).

Fig. 2: Hippocampal ePRS for SGK1 predicts depression and moderates an association between ELA and depression.
figure 2

A Diagram of SGK1 ePRS calculation. B Hippocampal SGK1 ePRS is higher in depressed children compared to non-depressed children. C Higher hippocampal SGK1 ePRS is associated with more depressive symptoms. D SGK1 ePRS moderates an association between ELA and depressive symptoms, and E between ELA and suicide attempts. Number of adversities: 0, 1, 2, 3, or ≥4. ELA – Early Life Adversity. ePRS – expression-based polygenic risk score. *p < 0.05; **p < 0.01.

Full size image

We next investigated whether the association between predicted hippocampal SGK1 expression and depression was influenced by ELA. Indeed, the hippocampal SGK1 ePRS interacted with ELA severity to predict higher levels of depressive symptoms (β = 0.025, t(8227) = 2.30, *p = 0.022; n = 8269). This association was strongest in children with high hippocampal SGK1 ePRS (simple slope at +1 SD: β = 0.21, **p < 0.01) compared to those with low ePRS (−1 SD: β = 0.16, **p < 0.01; Fig. 2D). Hippocampal SGK1 ePRS also interacted with ELA severity to predict suicide attempts (β = 0.03, t(7885) = 2.36, *p = 0.013, n = 7,842 without attempt, n = 96 with attempt). More ELA exposure was associated with higher rates of suicide attempts in children with high ePRS (simple slope at +1 SD: β = 0.10, **p < 0.01) and mean ePRS (β = 0.06, **p < 0.01), but not in children with low ePRS (−1 SD; p > 0.15; Fig. 2E), suggesting low hippocampal SGK1 expression may be protective against ELA. We found no significant three-way interaction with sex. Stratifying the data by sex showed that the interaction between hippocampal SGK1 ePRS and ELA on depressive symptoms was significant in boys (*p = 0.04) and marginally significant in girls (p = 0.08), with similar effect sizes in both sexes (β = 0.03). The interaction between hippocampal SGK1 ePRS and ELA on suicide attempts was only significant in boys (*p = 0.01), but not in girls (p = 0.31; Supplementary Table 3).

To test whether the associations between predicted SGK1 expression and depression are specific to the hippocampus, we generated ePRS for four additional depression-relevant regions: amygdala, nucleus accumbens (NAc), prefrontal cortex (PFC) and anterior cingulate cortex (ACC) [28, 29]. We found a small to moderate correlation between the hippocampal SGK1 ePRS and amygdala (r = 0.24, **p < 0.01), NAc (r = 0.27, **p < 0.01), ACC (r = 0.54, **p < 0.01) and PFC (r = −0.09, **p < 0.01) ePRSs, suggesting that while correlated, these ePRSs are different from each other and brain region specific. We found no main effect of the non-hippocampal ePRSs on depressive symptoms, and only the NAc ePRS was associated with depression diagnosis, albeit with a smaller effect size and significance than the hippocampal SGK1 ePRS (β = 0.02, *p = 0.03; Supplementary Table 4).

When we tested whether ePRSs for the four non-hippocampal regions moderate associations between ELA and depression or suicide attempts, we found a significant interaction effect on depressive symptoms only for the amygdala (β = 0.04, ***p < 0.001) and ACC SGK1 ePRSs (β = 0.02, *p = 0.03). We found no interaction effect on suicide attempts for any region other than the hippocampal SGK1 ePRS (Supplementary Table 4).

These data thus indicate that the collection of SGK1 SNP alleles that are associated with high hippocampal SGK1 expression predict early-onset depression and moderate interactions between ELA and both depressive symptoms and suicide attempts. While SGK1 expression in other regions, such as the amygdala and ACC, may contribute to moderating some effects of ELA on depressive symptoms, the moderation of SGK1 expression on the associations between ELA and suicide attempts was specific to the hippocampus.

Hippocampal knockdown of SGK1 confers stress resilience in mice

To determine whether SGK1 is mechanistically involved in behavioral phenotypes relevant for psychopathology, we used mouse models of ELA, chronic CORT, and adult chronic stress to test whether they would recapitulate the increased SGK1 mRNA expression that we found in the human hippocampus of suicide subjects. To disentangle ELA from adult stress effects, we modeled ELA using the limited bedding and nesting (LBN) paradigm (Fig. 3A) [26], which we previously found to predispose to heightened stress vulnerability in adulthood [14]. We found increased SGK1 expression in the DG of LBN-reared male and female mice in adulthood (at P56) compared to standard-reared control mice (main effect of ELA, F1,23 = 9.12, **p = 0.006, Fig. 3B). To test CORT effects on SGK1, we treated male mice for 5 weeks with CORT in the drinking water and found increased SGK1 expression in the DG (*p = 0.046; Fig. 3C, D). Chronic CORT treatment also decreased mobility time in the tail-suspension test (TST; *p = 0.015), which is commonly used to assess “despair-like” behavior in mice, and mobility in the TST negatively correlated with SGK1 mRNA expression (r = −0.61, *p = 0.012; Fig S1). To assess adult stress effects on SGK1 expression, we used the chronic social defeat stress (CSDS) model, which has repeatedly been shown to induce neurobiological and behavioral abnormalities in mice that resemble depression-relevant phenotypes in humans [8, 10, 28]. We exposed adult male mice to 10 days of CSDS and categorized mice as stress susceptible vs resilient based on their social avoidance score, as previously described (Fig. 3E) [8, 10, 28]. SGK1 expression was increased in the DG of stress susceptible, avoidant mice compared to resilient, non-avoidant mice and compared to undefeated controls (F2,25 = 7.35, **p = 0.003; Control vs Susceptible, **p < 0.01, Susceptible vs Resilient, **p < 0.01; n = 8–10; Fig. 3F). Moreover, SGK1 expression was inversely correlated with the social interaction (SI) score, indicating that higher levels of hippocampal SGK1 are associated with more social avoidance (r = −0.60, ***p = 0.0008; Fig. 3G). This association remained significant when we excluded the resilient mouse with the highest SI ratio (r = −0.49, *p = 0.01).

Fig. 3: Hippocampal SGK1 determines stress vulnerability in mice.
figure 3

A Timeline of LBN experiment. B SGK1 mRNA is increased in the DG of LBN-reared mice. C Timeline of CORT experiment. D SGK1 mRNA is increased in the DG of CORT-treated mice. E Timeline of CSDS experiment and subpopulations of susceptible and resilient mice following CSDS. F SGK1 expression is increased in the DG of susceptible, but not resilient mice. G SGK1 expression in the mouse DG negatively correlates with SI ratio. H Cre-virus injections into the DG of SGK1flox/flox mice. I SGK1 mRNA expression is reduced by ~85% at 5 weeks after Cre-virus injections. J Timeline of surgeries and behavior testing for SGK1 knockdown experiments. K mCherry expression indicating Cre virus expression in the DG of SGK1flox/flox mice 5 weeks after injection. Scale bars, 200μm. L DG-specific SGK1 knockdown does not affect interaction with an empty enclosure. M SGK1 knockdown prevents CSDS effects on social avoidance in the SI test, and N reduces the fraction of susceptible mice based on the SI ratio. DG – Dentate Gyrus; CORT – Corticosterone; CSDS – Chronic Social Defeat Stress; SI – Social Interaction. *p < 0.05; **p < 0.01.

Full size image

To test whether SGK1 is indeed a mechanistic factor underlying stress vulnerability, we then used a brain region specific knockdown approach to test whether CSDS would still cause social avoidance when hippocampal SGK1 is reduced. To this end, we injected CRE-expressing virus in the DG of SGK1flox/flox mice, which decreased SGK1 expression by ~85% at 5 weeks post-injection (**p < 0.01, Fig. 3H–K). CSDS did not affect interaction with an empty enclosure (Fig. 3L) but reduced the time mice interacted with a novel mouse (Fig. 3M, first orange bar). SGK1 knockdown prevented this effect of CSDS on social avoidance (Fig. 3M, second orange bar; Interaction, F1,43 = 4.19, *p = 0.046; control mCherry vs stress mCherry, *p = 0.036; stress mCherry vs stress CRE, **p = 0.004) and reduced the fraction of stress susceptible mice based on their SI ratio (Fig. 3N).

Together, these results indicate a mechanistic role for hippocampal SGK1 in mediating stress effects on behavior.

Pharmacological inhibition of SGK1 confers stress resilience in mice

Because we found that hippocampal SGK1 is mechanistically involved in stress vulnerability, we next wanted to test whether SGK1 could also be a target for pharmacological interventions aimed at preventing stress effects on behavior. We, therefore, injected mice with the small molecule inhibitor, GSK650394, which we previously found to rescue stress-hormone effects on hippocampal neurons in vitro [3]. To validate whether GSK650394 crosses the blood-brain barrier, we first injected mice with two different concentrations of the drug (1 mg/kg and 5 mg/kg, i.p.) and dissected hippocampus tissue 30 min after systemic injection (Fig. 4A). Liquid Chromatography-tandem Mass Spectrometry (LCMS) analyses showed that GSK650394 can be detected in the hippocampus following systemic injections, confirming that the drug indeed crosses the blood-brain barrier (Fig. 4B).

Fig. 4: SGK1 inhibition promotes stress resilience in mice.
figure 4

A Timeline and workflow for detecting GSK650394 in hippocampus tissue. B Systemic injections of GSK650394 result in detectable drug concentrations in the hippocampus. C Timeline of CSDS experiment with SGK1 inhibition using systemic GSK650394 injections. D GSK650394 does not affect interaction with a novel enclosure. E GSK650394 prevents a CSDS-induced reduction in social avoidance. F GSK650394 prevents a CSDS-induced increase in the fraction of susceptible mice based on their SI ratio. G GSK650394 increases the time mice spend in the center of the OF arena, and H rearing bouts in the OF. I GSK650394 prevents a CSDS-induced reduction in mobility during the last 4 min of the FST. CSDS – Chronic Social Defeat Stress; SI – Social Interaction; OF – Open Field; FST – Forced Swim Test. *p < 0.05; **p < 0.01.

Full size image

We then exposed male mice to CSDS and injected either GSK650394 (5 mg/kg, i.p.) or vehicle (0.9% NaCl) 30 min before each physical defeat session each day. Following the last social defeat encounter on day 10, we tested all mice in the social interaction test, an open field test (OFT), and a forced swim test (FST) (Fig. 4C). GSK650394 treatment did not affect interaction time with an empty enclosure (Fig. 4D) but prevented stress-induced avoidance of a novel mouse in the social interaction test (Fig. 4E; Interaction, F1,40 = 7.20, *p = 0.01; control saline vs stress saline, *p = 0.024; stress saline vs stress GSK, **p = 0.002) and reduced the fraction of stress susceptible mice based on their SI ratio (Fig. 4F).

In the OFT, GSK650394 increased the time that both control and stressed mice spent in the brightly lit, anxiogenic center of the arena (Fig. 4G; line indicates main effect of drug, F1,40 = 11.93, **p = 0.001), and the number of rearing bouts (Fig. 4H; line indicates main effect of drug, F1,40 = 7.34, **p = 0.009; main effect of stress, F1,40 = 5.74, *p = 0.021). GSK650394 also increased the distance traveled in the center of the OF arena (Fig S2a; main effect of drug, F1,40 = 5.29, *p = 0.027) without affecting overall locomotor activity (Fig S2b).

In the FST, CSDS reduced overall mobility in saline-injected mice during the last 4 min of the task, and this stress-induced reduction in mobility was prevented by GSK650394 (Fig. 4I; Interaction, F1,40 = 7.19, *p = 0.01; control saline vs stress saline, *p = 0.037; stress saline vs stress GSK, *p = 0.028).

Collectively, these data indicate that pharmacological inhibition of SGK1 reduces general avoidance / anxiety-like behavior in the OFT and prevents CSDS effects on social avoidance and on passive coping / despair-like behavior in the FST.

Pharmacological inhibition of SGK1 rescues stress effects on hippocampal neurogenesis and on DG hyperactivity in mice

Because systemic pharmacological inhibition of SGK1 promoted a stress resilient phenotype, we then asked whether SGK1 inhibition would cause neurobiological changes in the DG that we previously found to mediate individual differences in stress vulnerability: adult hippocampal neurogenesis and DG hyperactivity [8]. We, therefore, collected brains of vehicle and GSK650394 injected control and CSDS-exposed mice 1 h after behavior testing in the FST to quantify adult-born neurons and the immediate early gene product, cFos, in the DG (Fig. 5A).

Fig. 5: SGK1 inhibition rescues CSDS effects on hippocampal neurogenesis and DG hyperactivity.
figure 5

A Timeline for brain collections. B Representative images of Dcx-labeled young adult-born neurons in the ventral DG. C GSK650394 increases the number of Dcx+ young neurons in the dorsal DG. D GSK650394 prevents a CSDS-induced reduction in Dcx+ young neurons in the ventral DG. E Representative images of cFos-labeled granule cells in the ventral DG. F GSK650394 does not affect the number of cFos+ granule cells in the dorsal DG. G GSK650394 prevents a CSDS-induced increase in cFos+ granule cells in the ventral DG. CSDS – Chronic Social Defeat Stress; DG – Dentate Gyrus; Dcx – Doublecortin; DAPI is blue; Scale bars, 50μm. *p < 0.05; **p < 0.01.

Full size image

In the dorsal DG, CSDS did not reduce the number of doublecortin (Dcx)-positive adult-born neurons, similar to our previous findings in the same stress model (Fig. 5B, C) [8]. However, GSK650394 increased the number of Dcx+ cells in both control and defeated mice, indicating that SGK1 regulates adult hippocampal neurogenesis regardless of stress exposure (Fig. 5C; main effect of drug, F1,20 = 36.9, ***p < 0.001). In the ventral DG, CSDS reduced the number of Dcx+ cells in line with our previous studies showing that chronic stress predominantly affects the ventral DG [8, 10]. This stress-induced reduction in neurogenesis was prevented in mice that received daily injections of GSK650394 during CSDS (Fig. 5D; Interaction, F1,20 = 5.3, *p = 0.032; control veh vs CSDS veh, *p = 0.034; CSDS veh vs CSDS GSK650394, ***p = 0.0002).

We previously showed that chronic stress causes hyperactivity of ventral DG granule cells, which in turn increases stress susceptibility [8, 10]. Consistent with this previous finding, we find increased cFos expression in ventral DG granule cells of CSDS-exposed mice compared to controls 1 h after an acute swim stress (FST), indicating heightened neural activation in response to acute stress in mice that previously experienced chronic stress. Treatment with GSK650394 during CSDS prevented this increase in cFos+ cells, indicating that SGK1 mediates stress effects on ventral DG hyperactivity (Fig. 5E–G; Interaction, F1,20 = 5.03, *p = 0.036; control veh vs CSDS veh, *p = 0.013; CSDS veh vs CSDS GSK650394, **p = 0.004).

These data thus show that pharmacological inhibition of SGK1 prevents stress-induced neurobiological impairments in the DG that cause heightened susceptibility to stress, including adult neurogenesis and granule cell hyperactivity.

Discussion

Identifying targetable molecular mechanisms that determine individual differences in stress vulnerability is essential for developing novel strategies to prevent stress-induced psychiatric disorders. Here, we used a cross-species translational approach that leverages human postmortem brain tissue, population genetic data, and mechanistic experiments in rodents to show that hippocampal SGK1 promotes susceptibility for psychopathology. Our data demonstrate for the first time that SGK1 mediates stress effects on hippocampus function and behavior, and that genetic predisposition for heightened SGK1 expression moderates an association between ELA and psychopathology.

Our finding that SGK1 expression is elevated in the hippocampus of suicide completers is the first evidence for an association between psychopathology and SGK1 expression in the human hippocampus and extends our previous findings from rodents and cell culture models to the human brain [3]. While in our sample, ~60% of suicide completers had a diagnosis of MDD, it is possible that an underlying psychiatric condition remained undetected in subjects who died of suicide but who were not diagnosed with MDD before death. Interestingly, SGK1 levels were highest in suicide completers with a reported history of childhood adversity. These findings show that ELA has persistent effects on the brain’s transcriptome, including increased SGK1 expression in the human hippocampus many years after ELA exposure.

High SGK1 expression in human subjects with a reported history of ELA could be caused by the actual ELA exposure itself or by secondary effects of ELA, such as lifelong differences in socioeconomic status, differences in depression symptomatology, or responsiveness to psychotropic medication. Our experiments in mice help clarify these associations by removing potential confounding effects of late life environmental or socioeconomic group differences. These data show that ELA increases SGK1 in the adult hippocampus, indicating that ELA alone can indeed lead to increased SGK1 expression later in life. We observed similarly elevated hippocampal SGK1 expression following chronic CORT administration and adult chronic social stress. These findings suggest that adverse experiences across the lifespan contribute to hippocampal SGK1 expression, and extend our previous findings showing increased SGK1 mRNA in the adult hippocampus of rats exposed to prenatal stress [3]. While our previous work showed that ELA in the form of LBN affects adult stress responsivity specifically in female mice [14], we here find increased baseline SGK1 mRNA expression after LBN in both adult males and females. LBN thus likely exerts both sex-specific and sex-independent effects depending on the neurobiological pathway and on the level of analysis. While SGK1 is a known GR target gene, its expression is regulated by a variety of stimuli and other transcription factors, including the mineralocorticoid receptor (MR) [30, 31]. Thus, elevated SGK1 expression in suicide decedents and rodent stress models may also reflect contributions from non-GR pathways.

Genetic factors, including polymorphisms in stress-related genes, shape vulnerability for psychiatric disorders [32, 33]. Genetic polymorphisms in SGK1 have previously been associated with blood pressure [34] and coronary heart disease [35], but their role in psychiatric disorders is unknown. We found 23 SNPs in SGK1 that are associated with high hippocampal SGK1 mRNA and that confer depression risk in a large population sample of children from the ABCD study. This finding suggests that genetic predisposition for heightened hippocampal SGK1 expression can predict early-onset depression and severity of depressive symptoms. Although some SNPs included in our ePRS analysis may lie within or near GR binding sites based on ChIP-seq data from the ReMap Atlas of Regulatory Regions across various cell types and tissues, (https://remap2022.univ-amu.fr/) the precise functional role of these SNPs remains unknown and may involve alternative, GR-independent transcriptional mechanisms.

Risk for psychopathology is determined not only by genetic polymorphisms but by their interactions with environmental factors [16, 36,37,38,39,40,41,42]. In addition to genetic risk, our data show that the 23 SGK1 SNPs moderate associations between ELA and both depressive symptoms and suicide attempts. These findings expand our understanding of G x E interactions in psychiatry by providing insight into effects of these interactions on tissue-specific mRNA expression and by highlighting a role for SGK1 in promoting vulnerability to ELA effects on depression and suicide risk. Understanding genetic risk factors has potential to not only inform our understanding of the biology underlying depression, but also to improve diagnosis for psychiatric disorders. Indeed, ePRS predicting hippocampal SGK1 levels could be used in the future to inform an individual’s risk for developing depression and help estimate vulnerability to ELA based on genetic variances in the SGK1 gene.

In addition to identifying an association between hippocampal SGK1 and depression in humans, our mouse data establish for the first time a mechanistic role for SGK1 in mediating stress effects on behavior. We find that SGK1 knockdown in the mouse DG prevents CSDS effects on social avoidance, indicating that SGK1 indeed functionally mediates stress effects on behavior. These data support the notion that hippocampal SGK1 causes disease-relevant symptoms and behavior phenotypes in conditions in which expression is increased, as we find to be the case in suicide, depression, and following ELA and adult chronic stress [3]. While our study focused on identifying a role for hippocampal SGK1 in stress resilience and psychopathology, our ePRS analysis suggests that SGK1 in other brain regions, such as the amygdala and ACC, may also contribute to the association between ELA and depressive symptoms. Future studies should examine whether stress increases SGK1 expression across the brain, and whether SGK1 in other regions may also play a causal role in mediating stress effects on behavior.

In addition to being a biological mediator for stress effects, our data demonstrate that SGK1 can be pharmacologically targeted to promote resilience. SGK1 is broadly expressed throughout the brain and periphery, and systemic inhibition likely impacts SGK1 function throughout the body. Nevertheless, systemic inhibition effectively prevented stress-induced alterations in the hippocampus, including reduced adult neurogenesis and DG hyperactivity - two mechanisms we previously found to determine individual differences in stress vulnerability [8]. Considering that SGK1 inhibition increased the number of adult-born neurons in both stressed and unstressed mice, SGK1 likely regulates neurogenesis regardless of stress exposure. SGK1 inhibition may thus not only rescue stress effects but also improve neurogenesis-dependent behaviors in general, as we indeed find to be the case in the OFT, in which SGK1 inhibition increased exploration of the center and rearing behavior in both control and stressed animals. Previous work has shown that SGK1 is expressed in various cell types of the hippocampus [43], and it is unclear which cell type is responsible for the effects of SGK1 on neurogenesis. SGK1 is expressed in adult-born hippocampal neurons [44] and increased neurogenesis following SGK1 inhibition may thus be the result of SGK1 inhibition in these cells. While increased neurogenesis contributes to reducing DG activity [8], changes in stress-induced hyperactivity could also result from direct inhibition of DG granule cells as glucocorticoids and SGK1 both affect expression of ion channels involved in neural excitability [4, 10, 45]. Previous work has reported that stress increases SGK1 expression in oligodendrocytes [43, 46, 47], which may contribute to neural activity by affecting myelination. It is thus possible that SGK1 inhibition simultaneously affects several cell types relevant for neurogenesis and neural activity in the hippocampus. While further dissection of the precise cellular targets of SGK1 will be important to fully understand the molecular signaling mechanisms underlying stress vulnerability, new drugs to treat or prevent stress effects will likely need to be administered systemically and thus act throughout the body.

It is important to note that while our genetic study included both male and female participants, our human postmortem and rodent studies were conducted only in males. It will thus be important to determine whether our characterization of SGK1 expression and its functional role in stress vulnerability extend to females. Moreover, the pharmacological inhibitor GSK650394 has some off-target effects on SGK2. Testing more specific SGK1 inhibitors may therefore help further validate SGK1 as a pharmacological target for enhancing stress resilience.

In conclusion, our findings identify hippocampal SGK1 as a key mechanism regulating stress vulnerability and as a potential target for novel resilience-promoting interventions.