Molecular adaptations of the blood-brain barrier promote stress resilience vs. depression

Abstract

Preclinical and clinical studies suggest that inflammation and vascular dysfunction contribute to the pathogenesis of major depressive disorder (MDD). Chronic social stress alters blood-brain barrier (BBB) integrity through loss of tight junction protein claudin-5 (cldn5) in male mice, promoting passage of circulating proinflammatory cytokines and depression-like behaviors. This effect is prominent within the nucleus accumbens, a brain region associated with mood regulation; however, the mechanisms involved are unclear. Moreover, compensatory responses leading to proper behavioral strategies and active resilience are unknown. Here we identify active molecular changes within the BBB associated with stress resilience that might serve a protective role for the neurovasculature. We also confirm the relevance of such changes to human depression and antidepressant treatment. We show that permissive epigenetic regulation of cldn5 expression and low endothelium expression of repressive cldn5-related transcription factor foxo1 are associated with stress resilience. Region- and endothelial cell-specific whole transcriptomic analyses revealed molecular signatures associated with stress vulnerability vs. resilience. We identified proinflammatory TNFα/NFκB signaling and hdac1 as mediators of stress susceptibility. Pharmacological inhibition of stress-induced increase in hdac1 activity rescued cldn5 expression in the NAc and promoted resilience. Importantly, we confirmed changes in HDAC1 expression in the NAc of depressed patients without antidepressant treatment in line with CLDN5 loss. Conversely, many of these deleterious CLDN5-related molecular changes were reduced in postmortem NAc from antidepressant-treated subjects. These findings reinforce the importance of considering stress-induced neurovascular pathology in depression and provide therapeutic targets to treat this mood disorder and promote resilience.

Keywords: antidepressant; epigenetic; inflammation; mood disorders; vascular.

Fig. 1.

Permissive epigenetic regulation at the cldn5 promoter is associated with stress resilience. (A) Experimental timeline and (B) behavioral profile of unstressed CTRL, SS, SI ratio <1, and RES, SI ratio >1 (one-way ANOVA: F_2,14 = 1,084; ***P < 0.0001, n = 10 to 12 mice/group). (C) Stress resilience is associated with higher permissive pan-acetylation on histone 3 (pan-acH3, two-way ANOVA: phenotype effect F_2,168 = 168; ***P < 0.0001) and lower repressive methylation on H3K27me3 (two-way ANOVA: phenotype effect F_2,168 = 32.05; ***P < 0.0001) on cldn5 gene promoter in the NAc of mice. Lower acetylation is also observed in SS mice when compared to CTRL (two-way ANOVA: phenotype effect F_1,108 = 13.68; ***P = 0.0003). (D) Repressive histone methylation at CLDN5 promoter is reduced in the NAc of major depressive disorder (MDD) subjects under antidepressant (AD⁺) treatment at time of death when compared to healthy CTRL or MDD subjects without treatment (MDD, AD⁻) (500 bp, one-way ANOVA: F_2,24 = 6.119; **P = 0.0071; 2,700 bp, one-way ANOVA: F_2,24 = 4.513; *P = 0.0217; n = 5 to 15 subjects/group). If one- or two-way ANOVA statistical test was significant, Bonferroni posttests were performed with *P < 0.05; **P < 0.01; ***P < 0.001. See SI Appendix, Tables S1 and S3 for primers and detailed demographic data.

Fig. 2.

Expression of repressive cldn5-related transcription factor FoxO1 is reduced in the endothelium of resilient mice. (A) Experimental timeline of a 10-d CSDS paradigm, SI test, and tissue collection for RNAscope in situ hybridization. (B) SI ratio of CTRL, SS, and RES mice (one-way ANOVA: F_2,15 = 12.46; ***P = 0.0006, n = 6 mice/group). (C) Representative images of FoxO1 expression in the NAc which is reduced in RES mice (two-way ANOVA: phenotype × n clusters interaction effect F_12,63 = 4.856; ***P < 0.0001, n = 2,306 to 3,657 cells from four mice/group) (Scale bar, 10 μm) (D). Nonendothelial cells are identified in light gray in the pie charts and endothelial cells, double labeled with CD31, split between shell (white for CTRL, dark gray for SS, pink for RES) and core (white for CTRL, black for SS, red for RES) subregions of the NAc. If one- or two-way ANOVA statistical test was significant, Bonferroni posttests were performed with *P < 0.05; **P < 0.01; ***P < 0.001.

Fig. 3.

Stress-induced transcriptome-wide changes in NAc endothelial cell gene expression. (A) Experimental timeline and (B) behavior profiles of CTRL, SS, and RES mice used to compare NAc endothelial cell gene expression (one-way ANOVA: F_2,12 = 31.44; ***P < 0.0001, n = 4 to 6 mice/group). (C) Enrichment of NAc endothelial cells following MACS was confirmed by flow cytometry (C, 81.7% vs. 4.2% for heterogenous preparation including all cell types) and (D) quantitative PCR (500- to 12-fold enrichment for endothelial cell-related genes vs. 4- to 2-fold for astrocytes, 0.6-fold for oligodendrocytes, and 0.06-fold for neurons after MACS purification vs. heterogenous preparation). (E) Venn diagrams revealed poor overlap of gene expression changes when group comparisons were performed. Gene expression was mostly down-regulated (blue) in NAc endothelial cells of stressed mice when compared to unstressed controls with a similar number of genes being up-regulated (yellow) vs. down-regulated (blue) in RES vs. SS animals. Significance was set at ±2-fold change and P < 0.05. See Dataset S1 for detailed gene lists. (F) Hierarchical clustering heat map of SS vs. CTRL mice and biological pathways including the higher number of genes up-regulated (yellow) or down-regulated (blue) in the NAc endothelial cells of these groups of mice. One-way ANOVA statistical test was significant thus Bonferroni posttests were performed with ***P < 0.001.

Fig. 4.

Inhibition of stress-induced increased hdac1 expression promotes resilience and rescues cldn5 expression. (A) Following 10 d of CSDS, hdac1 expression is down-regulated in the NAc endothelial cells of RES mice when compared to SS mice (one-way ANOVA, *P = 0.0301, n = 4 to 5 mice/group). (B and C) Higher stress-induced hdac1 level in the NAc of SS mice was confirmed by quantitative PCR and negatively correlated with SI (SI ratio: one-way ANOVA: F_2,34 = 44.27; ***P < 0.0001; hdac1 fold change: one-way ANOVA: F_2,29 = 6.122; **P = 0.0061, Pearson’s correlation between SI ratio and hdac1 fold change: **P = 0.0068, n = 10 to 16 mice/group). (D) Mice were screened with the SI test following 10 d of CSDS and split equally in two groups then administered vehicle or MS-275. (E) A second SI test was performed after 10 d of treatment revealing increased social interactions in the MS-275–treated group (unpaired t test: ***P < 0.0001) in line with higher cldn5 expression (Pearson’s correlation: **P = 0.0019, n = 11 mice/group). One-way ANOVAs statistical test were significant, thus Bonferroni posttests were performed with **P < 0.01; ***P < 0.001.

Fig. 5.

Increased HDAC1 expression is associated with CLDN5 loss in human depression. (A) FOXO1 expression is increased in the NAc of MDD patients with or without antidepressant treatment when compared to healthy controls (one-way ANOVA: F_2,38 = 7.616; **P = 0.0017, 13 to 14 subjects/group). No change was observed for CTNNB1 (one-way ANOVA: F_2,39 = 0.08791; P = 0.9160, n = 13 to 14 subjects/group). (B) HDAC1 expression is increased in the NAc of MDD patients without antidepressant treatment (one-way ANOVA: F_2,36 = 3.865; *P = 0.0329) and correlated with CLDN5 mRNA level (Pearson’s correlation: *P = 0.0337, n = 12 to 14 subjects/group). (C) CLDN5 protein expression is reduced in postmortem NAc samples of depressed subjects when compared to healthy controls. (Scale bar, 100 μm.) See SI Appendix, Tables S1 and S3 for primers and detailed demographic data.

Fig. 6.

Claudin5 (cldn5) permissive epigenetic regulation is associated with stress resilience while lack of endothelial molecular adaptations and inflammation leads to cldn5 loss and depression. Low hdac1 expression and high acetylation/low methylation on the cldn5 gene promoter allows maintenance of the BBB integrity under chronic stress leading to normal social and stress coping behaviors. These proresilient adaptive changes also include reduced expression of the repressive transcription factor Foxo1, possibly to prevent its binding to the cldn5 promoter now accessible through permissive epigenetic changes, and subsequent suppression of cldn5 production (Left). Conversely, endothelial inflammation mediated by high levels of circulating proinflammatory cytokines (7, 33), activation of TNFα/NFκB signaling, and elevated hdac1 activity prevents proper molecular adaptations, including access to the cldn5 gene promoter, leading to cldn5 loss. Stress-induced loss of this tight junction protein results in BBB hyperpermeability, passage of circulating proinflammatory mediators, and the establishment of depression-like behaviors (Right). Adapted with permission from ref. .