Main text

Depression is one of the most common mental disorders affecting people of all ages, and can arise from a variety of causes, including genetic susceptibility, endocrine dysregulation, and stresses in life [1]. When exposed to acute and temporary stress, while the body can protect itself from stress, chronic stress can disturb the function of the brain system. Chronic accumulation of stress leads to abnormal and excessive cortisol secretion in the hypothalamic-pituitary-adrenal axis, which affects a variety of physical activities, including brain function, leading to mental disorders such as depression or post-traumatic stress disorder (PTSD) [2]. Proper regulation through negative feedback of glucocorticoid receptors (GR) is important for the stress response, and long-term or excessive activation of this system is associated with the development of depression or anxiety disorders [3]. The FK506-binding protein 51 (FKBP5) is a co-chaperone of Hsp90 in the GR molecular complex and is a key modulator of GR sensitivity [4, 5]. Although FKBP5 is an important factor that is responsible for coping behavior as well as neuroendocrine responses [6], results from previous studies investigating the association between Fkbp5 gene variants and stress remain controversial. In order to investigate whether genetic FKBP5 variants affect behavior in response to chronic restraint stress exposure, we examined the behavior of mice following the induction of chronic restraint stress in homozygous wild-type (WT) and knock-out (KO) mice. After 21 days of exposure to restraint stress, while WT mice showed anhedonia in the sucrose preference test, Fkbp5-deficient mice did not exhibit significant depressive-like behavior compared to the WT (Fig. 1)a and b.

Fig. 1
figure 1

Transcriptome analyses of the mPFC in homozygous wild-type (WT) and Fkbp5 knock-out (KO) mice following chronic stress. a Schematic timeline of the induction of chronic restraint stress. b Effect of CRS on sucrose preference. Data were combined with the 1st and 2nd sucrose preference test (SPT) results. Control mice (CT) n = 9; CRS-exposed WT mice (WT_ST) n = 6; CRS-exposed Fkbp5 KO mice (KO_ST) n = 6. One-way ANOVA (F[2,39] = 11.67, p = 0.0001); Fisher’s LSD: ***p < 0.001. c Multidimensional scaling (MDS) plot for transcriptomes of individual samples of CT (yellow), WT_ST (blue) or KO_ST (red). d The interleaved scatter plots of modules which have a significant negative correlation with stress. Data represent mean ± SEM. One-way ANOVA; Fisher’s LSD: *p ≤ 0.05, **p < 0.01. e A heatmap showing the expression of M55 genes in the mPFC of CT (left), WT_ST (middle) and KO_ST (right). f A network plot of M55 genes and their intramodular connections. The ten hub genes (the top ten genes with highest intramodular connectivity; Cul9, Polm, Ttll8, Vmn1r90, Tacc3, Mir877, Mmp25, Bhlhe23, Wtip and Ube2d-ps) are shown in red. Note their central position in the network, suggesting high intramodular connectivity. g Enrichment dot plot for Gene Ontology (GO) analysis of M55 genes. The 13 GO terms with the lowest p-value each annotation cluster are plotted in order of gene ratio. The dot size represents the number of genes associated with a specific term. The dot color represents the adjusted p-value

Full size image

Recent studies performed in humans and rodents have suggested that long-term stress and pathological anxiety leads to structural degeneration and functional alteration of the frontal cortex, and increases the risk of mental disorders [7, 8]. In addition, it has been suggested that the medial prefrontal cortex (mPFC), a region controlling higher brain function including cognition and emotion, is a primary target of stress hormones [9, 10]. However, little is known about the molecular mechanisms in the mPFC involved in stress-associated psychiatric disorders. Transcriptome profiling has helped provide an unbiased insight into the pathophysiological mechanism underlying complicated brain disorders [11]. Therefore, using RNA sequencing (RNAseq) analysis, we investigated the dynamic transcriptomic changes that occur after stress in the mPFC of Fkbp5-deficient mice.

Multidimensional scaling analysis showed a clear separation between the stressed and the non-stressed control (CT) mice. There was also a slight overlap between the CRS-exposed WT mice (WT_ST) and CRS-exposed Fkbp5 KO mice (KO_ST) groups, which were clustered according to their genotype (Fig. 1)c. To identify the genotypes and genes affected by stress, we analyzed differentially expressed genes (DEG) between each group (Additional file 1). Of the 24,532 mRNA genes profiled, 224 (0.91%; CT vs WT_ST), 258 (1.05%; CT vs KO_ST), and 135 (0.55%; WT_ST vs KO_ST) genes were dysregulated in the mPFC following chronic restraint stress, and the percentage of DEG induced by stress was higher than the percentage of DEG by genotype (Additional file 2: Figure S1, Additional file 3: Table S1-S3). Gene ontology (GO) enrichment analysis of these expression profiles showed functional categories that were potentially dysregulated, including lipid metabolic process (Benjamini adjusted p = 9.63 × 10− 4), regulation of immune response (Benjamini adjusted p = 7.24 × 10− 4), cell adhesion (Benjamini adjusted p = 1.52 × 10− 3), regulation of cell differentiation (Benjamini adjusted p = 1.17 × 10− 3), and neurogenesis (Benjamini adjusted p = 7.66 × 10− 2) (Additional file 4: Table S4-S6).

To compare the differences in expression observed in the multidimensional data set with the pattern of stress-response behavior, we performed a weighted gene co-expression network analysis (WGCNA) [12]. Through WGCNA, we identified 60 modules of co-expressed genes following chronic restraint stress in both WT and KO homozygous genotypes (Additional file 5: Figure S2, Additional file 6: Table S7). Among the 60 modules, we identified characteristic modules which showed a significant correlation to the genotype (M3, M6, M25, M33, M39, M44 and M57) and to the stress exposure (M3, M10, M18, M21, M31, M34, M50 and M55) (Fig. 1, Additional file 7: Figure S3)d, a-c. Interestingly, one of the modules that negatively correlated with stress, M55, had a pattern similar to the stress resilience behavior of Fkbp5-deficient mice. This was down-regulated, consistent with depression-like behavior in the WT_ST group, and was restored in the KO_ST group (Fig. 1)d and e. GO enrichment analysis of M55 revealed the biological functions potentially involved, including gland morphogenesis (p = 1.25 × 10− 2), activation of immune response (p = 2.25 × 10− 2), and nervous system development (p = 7.65 × 10− 3) (Fig. 1, Additional file 8: Table S8)f and g. Hub genes, with the highest degree of connectivity within a module of the M55 include Mmp25. This gene has been functionally implicated in the regulation of immune response through NF-B signaling [13] and has been linked to neuropsychiatric disorders including PTSD [14].

In this study, we compared brain transcriptome altered by chronic stress in the mPFC between Fkbp5-deficient and wild-type mice by RNAseq analysis. In addition to the DEG analysis, by employing WGCNA, we identified a distinct co-expression network module associated with stress resilience caused by Fkbp5 knock-out, and characterized the biological processes affected by this module, leading to this unique behavior. Our systematic transcriptome analysis demonstrated that aberration in the development of the neuroendocrine system, and regulation of the immune response may underlie the stress resilient behavior observed in the Fkbp5 deficient mice. This is the first study, to the best of our knowledge, to identify the stress resilience associated genes through gene co-expression network analysis in Fkbp5 deficient mice. Altogether, we confirmed that FKBP5 may be an important component of the stress response, suggesting that identification of the module associated with the stress response can provide a treatment strategy and therapeutic target to attenuate the depressive symptoms caused by stress.