Volume 96, Issue 2, Pages 402-413.e5 (October 2017) Role of the Astroglial Glutamate Exchanger xCT in Ventral Hippocampus in Resilience to Stress  Carla Nasca, Benedetta Bigio, Danielle Zelli, Paolo de Angelis, Timothy Lau, Masahiro Okamoto, Hideyo Soya, Jason Ni, Lars Brichta, Paul Greengard, Rachael L. Neve, Francis S. Lee, Bruce S. McEwen  Neuron  Volume 96, Issue 2, Pages 402-413.e5 (October 2017) DOI: 10.1016/j.neuron.2017.09.020 Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 1 Stress Differentially Regulates Glutamatergic Gene Expression in Whole, Dorsal, and Ventral Hippocampus: Role of xCT in vDG in Loss of Resilience to Stress (A–C) Glutamatergic gene expression profiling within the whole hippocampus showing adaptive plasticity of the glutamatergic system (A and C) in response to acute (1 day: ARS), sub-chronic (7 days and 14 days: subCRS), and chronic (21 days: CRS) stress. (B) Prolonged 21-day stress, and not ARS and subCRS, led to increased immobility time at the forced-swim test as compared to age-matched not-stressed mice (one-way ANOVA repeated-measures F4,52 = 0.74, p = 0.57 [time]; F1,13 = 2.43, p = 0.14 [stress]; F4,52 = 2.89, p < 0.05 [interaction]). CRS also led to impaired SIs (Figure S2). (D and E) CRS led to a substantial habituation within the dorsal hippocampus (dHipp, D) with no changes in all glutamate genes, except for upregulation of NR1/NMDA receptors and Glt-1 transporters, which exert opposite functions on glutamate homeostasis (E). (D and F) In contrast, CRS led to dysregulated glutamatergic activity in the ventral hippocampus (vHipp, D) with downregulation of mGlu2, mGlu3 and xCT transcripts (F). (G) Representative 40× confocal images of hippocampal vDG sections showing xCT, DAPI, and merged immunofluorescence from not-stressed mice (top) and CRS mice (bottom). (H–K) Immunofluorescence (H and J) and immunohistochemistry (I and K) approaches show that CRS reductions of xCT and mGlu2 expression were narrowed to the ventral dentate gyrus (vDG). Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 at Student’s two-tailed t tests. Representative brain sections in (D) are available from 2015 Allen Institute for Brain Science. Allen Mouse Brain Atlas (Lein et al., 2007). Green and red signs above bars indicate putative glutamate homeostasis (up, down, and equal); green and red colors indicate adaptive and maladaptive behavioral outcomes in terms of depressive-like behaviors (see also B). Scale bar, 1 μm. See also Figures S1, S2, and S3. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 2 RNA-Seq Reveals Transcriptome-wide Alterations in the Hippocampal vDG in Response to Chronic Stress (A) Representative coronal brain images with vDG reference for microdissection (vDG, highlighted in red). (B) Heatmap of CRS-regulated expression changes as compared to not-stressed control mice. RNA-sequencing to capture transcriptome-wide alterations in the vDG identified 499 transcripts (adjusted-p < 0.15, fold change >1.3) differentially expressed in three biological replicate experiments with 254 downregulated and 245 upregulated genes. Z scores of all altered RNA between 21-day CRS group and not-stressed age-matched group (Ctrl) are shown. (C) qPCR validation on larger biological replicate cohorts confirmed changes in the genes xCT and mGlu2 showing agreement between xCT and mGlu2 protein and mRNA expression. (D) GO analysis showing meaningful gene categories within the cellular component (green), biological process (blue), and molecular function (yellow) altered by CRS in the hippocampal vDG sub-region. (E) Enrichment pathway analysis shows that CRS affected several relevant signaling pathways within the vDG, including genes converging into the glutamatergic signaling (e.g., xCT: adj p value < 0.02, FC = 1.4 [p value <0.0005]), energy metabolism, inflammation, insulin pathway, calcium signaling, endocannabinoid and addiction pathways, and TrkB signaling cascade). Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗p < 0.05, ∗∗p < 0.01 at Student’s two-tailed t tests. See also Tables S1 and S2. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 3 Double-ChIP Shows an Epigenetic Mechanism of Co-occupancy of H3K27ac and REST in the Regulation of xCT (A) Immunohistochemical analysis shows that CRS resulted in a specific upregulation of REST protein expression in the vDG with no significant change in the vCA3 and vCA1. (B) Immunohistochemical analysis shows that CRS decreased acetylation of the histone marker H3K27 in the vDG. (C) ChIP analyses revealed a CRS-induced increased and decreased enrichment of REST and H3K27ac on the xCT promoter, respectively. (D) Representative 4× (left column) and 20× (three right columns) images of hippocampal sections of acetylation of H3K27 within the ventral hippocampal subregions vDG, vCA3, and vCA1, highlighting a specific decrease in H3K27ac in the vDG. Top, not-stressed control mice; bottom, CRS mice. (E) Schematic of double chromatin immunoprecipitation procedure (double-ChIP or Co-ChIP) for REST and H3K27ac to study functional consequences of their co-enrichment for target genes. (F) As shown by double-ChIP for the xCT promoter gene in naive animals, REST and H3K27ac were co-enriched on the xCT promoter, indicating that a combinatorial mechanism with co-occupancy of REST and H3K27ac regulates xCT expression. Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗∗p < 0.01, ∗∗∗p < 0.001 at Student’s two-tailed t tests. Scale bars, 100 μm and 20 μm for 4× and 20×, respectively. See also Figures S4 and S5. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 4 xCT, Acting in a Network with mGlu2 Receptors, as Novel Target for Pharmacological Preventive Approaches to Promote Resilience to Stress (A) Chemical structure of agents used to implement gain-of-function pharmacological approaches to increase resilience to stress: N-acetyl-cysteine (NAC) in comparison with the novel antidepressant candidate and acetylating agent acetyl-L-carnitine (LAC) and the reference SSRI antidepressant fluoxetine (Fluox). Dashed squares highlight acetyl groups in NAC and LAC. (B) Schematic of pharmacological preventive approach showing the time course of drug administration. (C) Rapid pro-resilient effects of NAC, and the acetylating agent LAC, at the FST. A same time course of administration with fluoxetine did not result in pro-resilient effects at the same behavioral test. (F1,51 = 5.64, p < 0.05 [stress]; F3,51 = 7.22, p < 0.001 [treatment]; F3,51 = 6.89, p < 0.001 [interaction]). (D and E) xCT mRNA (D) and protein expression (E) in CRS mice treated with NAC. (mRNA: F1,22 = 5.8, p < 0.05 [stress]. Intensity: F1,44 = 11.82, p < 0.01 [treatment]). (F) ChIP showing CRS-induced increase in REST enrichment to the xCT promoter and the CRS-induced decrease in H3K27ac bound to the same promoter target. (REST: F2,17 = 24.5, p < 0.0001. H3K27ac: F2,33 = 7.06, p < 0.001)). (G and H) A single injection of SAS to CRS mice treated with NAC reduced the pro-resilient effects of NAC at the FST (G) and SI test (H). (FST: F2,17 = 9.76, p < 0.01. SI:). (I) SAS also blocked NAC effects in elevating xCT and mGlu2 transcript levels in the vDG of CRS mice (mRNA, xCT: F2,16 = 6.28, p < 0.01; mRNA, mGlu2: F2,11 = 7.92, p < 0.01). (J) Schematics featuring a mechanistic model in that stress, glucocorticoids, and glutamate interact each other in a gene network system to regulate adaptive brain plasticity. In pathological condition, stress activates a glutamate cascade with decrease in extracellular release of glutamate own to reduced expression of xCT in vDG and failure to provide glutamate to prime the nearby mGlu2 receptors with consequent suppression of mGlu2-mediated inhibition of neuronal glutamate release. This gene network system accounts for the overflow of glutamate in the synaptic space and underlying brain plasticity and maladaptive responses to stress with occurrence of depressive-like behaviors. Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗p < 0.05, ∗∗p < 0.001, ∗∗∗p < 0.0001. See also Figure S6. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 5 CRS Effects on Behavior and xCT Reduction in vDG Endures after Discontinuing the Stress and Are Improved by Antidepressant-like Pharmacological Treatment Approaches (A) Schematic of pharmacological treatment approach showing the time course of drug administration. (B–D) 3 days after the end of CRS, mice receiving water only showed an enduring stress-induced susceptibility to changes in copying style at the FST (B) and in SI (C) that were paralleled by an enduring xCT reduction in the vDG (D). Pharmacological treatment with NAC, orally administered for 3 days at the end of CRS, improved copying style at the FST (B) and SI (C), and elevated xCT expression in the vDG of CRS mice (D). Similar rapid antidepressant-like effects were observed using a pharmacological treatment with LAC, orally administered for 3 days at the end of CRS (Figure S7). Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗p < 0.05, ∗∗p < 0.001, ∗∗∗p < 0.0001. See also Figure S7. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 6 xCT Overexpression with Anatomical Specificity to the vDG Controls Pro-resilient Responses to Stress (A) Schematic of viral approach with anatomical specificity to the vDG. (B and C) Effects of xCT overexpression in vDG of CRS mice at the FST (B) and SI test (C) as compared to CRS mice receiving injection of the control viral-vector HSV-GFP. (D) mGlu2 mRNA levels after HSV-xCT overexpression in vDG of CRS mice as compared to CRS mice receiving injection of the control viral-vector HSV-GFP. Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗p < 0.05. Scale bar, 100 μm. See also Figure S8. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 7 Cell-Type-Specific xCT Overexpression in GFAP+ Cells in vDG Controls Pro-resilient Responses to Stress (A and B) Confocal images of hippocampal vDG sections of xCT, GFAP, DAPI, and merged immunofluorescence from not-stressed mice (top) and CRS mice (bottom) (A) showing CRS induced downregulation of xCT in GFAP+ cells in vDG (B). (C) Schematic of viral approach with cellular and anatomical specificity to GFAP+ cells in vDG of CRS GFAP+ reporter mice. (D and E) Effects of xCT overexpression in GFAP+ cells in vDG of CRS mice at the FST (D) and SI test (E) as compared to CRS mice receiving injection of the control viral-vector HSV-Cre-GFP. Bars represent mean ± SEM, and asterisk indicates significant comparisons with corresponding controls, ∗p < 0.05, ∗∗p < 0.001. Scale bar, 1 μm. See also Figure S8. Neuron 2017 96, 402-413.e5DOI: (10.1016/j.neuron.2017.09.020) Copyright © 2017 Elsevier Inc. Terms and Conditions