Sustained antidepressant effects of ketamine metabolite involve GABAergic inhibition-mediated molecular dynamics in aPVT glutamatergic neurons

Abstract

Despite the rapid and sustained antidepressant effects of ketamine and its metabolites, their underlying cellular and molecular mechanisms are not fully understood. Here, we demonstrate that the sustained antidepressant-like behavioral effects of (2S,6S)-hydroxynorketamine (HNK) in repeatedly stressed animal models involve neurobiological changes in the anterior paraventricular nucleus of the thalamus (aPVT). Mechanistically, (2S,6S)-HNK induces mRNA expression of extrasynaptic GABA_A receptors and subsequently enhances GABA_A-receptor-mediated tonic currents, leading to the nuclear export of histone demethylase KDM6 and its replacement by histone methyltransferase EZH2. This process increases H3K27me3 levels, which in turn suppresses the transcription of genes associated with G-protein-coupled receptor signaling. Thus, our findings shed light on the comprehensive cellular and molecular mechanisms in aPVT underlying the sustained antidepressant behavioral effects of ketamine metabolites. This study may support the development of potentially effective next-generation pharmacotherapies to promote sustained remission of stress-related psychiatric disorders.

Keywords: GABA(A) receptor; GPCR signal; antidepressant actions; depression; epigenetics; ketamine metabolite; paraventricular nucleus of the thalamus.

Figure 1.. (2S,6S)-HNK has rapid and sustained antidepressant effects without ketamine-related side effects

(A) Experimental design for rTST model. (B–D) Immobility time of the rTST-exposed B6 mice 30 min, 24 h, and 3 days after drug treatment. n = 12 per group. (E) Experimental design for CSDS model. (F–H) Representative heatmaps (F), time spent in the target zone of the SIT in nonstressed (NS) and CSDS-exposed resilient (RES) and susceptible (SUS) mice before (G) and after (H) drug treatment. n = 10 per group. (I) Experimental design for smSDS model. (J and K) Time spent in the target zone of the SIT (J) and sucrose preference in the SPT (K) in NS and smSDS-exposed mice receiving (2R,6R)-HNK, (2S,6S)-HNK, or saline. n = 14–15 per group. (L) Experimental design for sustained behavioral effects of smSDS-exposed BALB mice receiving (2S,6S)-HNK. (M) Time spent in the target zone of the SIT in NS and smSDS-exposed mice receiving (2S,6S)-HNK or saline. n = 15–16 per group. (N) Locomotor activity in the OFT was shown. (O) Locomotor activity during 30 min of OFT after drug treatment. n = 8–12 per group. (P) Latency to fall in the rotarod test. n = 11–12 per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Bar graphs show the mean ± SEM.

Figure 2.. Brain-wide Fos mapping identifies the aPVT as a specific brain area responsible for (2S,6S)-HNK but not (2R,6R)-HNK effects

(A) Experimental design of Fos quantification. (B) Quantification of Fos immunostaining in B6 mice receiving (2R,6R)-HNK, (2S,6S)-HNK, or saline. n = 8 per group. (C and D) Brain-wide Fos levels of mice receiving (2R,6R)-HNK (C) and (2S,6S)-HNK (D) compared with saline controls (z-score). (E and F) Inter-regional correlations for brain-wide Fos immunoreactivity in mice receiving (2R,6R)-HNK (E) and (2S,6S)-HNK (F). (G and H) Most robust (top 2%) inter-regional correlations. (I) Representative images showing aPVT projections to multiple brain regions. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Bar graphs show the mean ± SEM.

Figure 3.. aPVT glutamatergic neurons are necessary for the sustained, but not rapid, antidepressant effects of (2S,6S)-HNK

(A) Schematic of AAV microinjection. (B) Representative image showing tdTomato expression in the PVT. Scale bar, 200 μm. (C) TUNEL staining in coronal sections of the PVT. Scale bar, 100 μm. (D and E) Representative image (D) and quantification (E) showing Cre expression (indicator of ablation efficacy) in the PVT. Scale bar, 200 μm. n = 6 per group. (F) Experimental design for aPVT ablation. (G and H) Immobility time of the rTST 1 h (G) and 3 days (H) after (2S,6S)-HNK injection. n = 13–15 per group. (I and J) Correlation of aPVT neuronal ablation efficacy (Cre expression levels) with the immobility time of the TST at 1 h (I) and 3 days (J) after (2S,6S)-HNK injection. (K) Schematic of AAV microinjection. (L) Representative image showing hM3Dq-mCherry and Fos expression in the aPVT and pPVT. Scale bar, 200 μm. (M) Quantification of Fos expression in the aPVT and pPVT 90 min after injection of CNO or vehicle in hM3Dq mice. (N) Experimental design for hM3Dq-DREADD. (O) Immobility time of rTST hM3Dq mice 1 h, 24 h, 3 and 7 days after CNO or vehicle treatment. n = 10–12 per group. (P) Schematic of AAV microinjection. (Q) Representative image showing hM4Di-mCherry and Fos expression in the aPVT and pPVT of hM4Di mice 90 min after (2S,6S)-HNK treatment. Scale bar, 200 μm. (R) Quantification of Fos expression in the aPVT and pPVT 90 min after injection of CNO or vehicle in hM4Di mice. (S) Experimental design for hM4Di-DREADD. (T–W) Immobility time of rTST-exposed hM4Di mice 1 h (T), 24 h (U), 3 days (V), and 7 days (W) after (2S,6S)-HNK injection. n = 12–16 per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Bar graphs show the mean ± SEM.

Figure 4.. RNA sequencing reveals the dynamic molecular processes underlying (2S,6S)-HNK’s effects

(A) Heatmap showing expression levels of all differentially expressed genes in NS-Saline, smSDS-Saline, and smSDS-(2S,6S)-HNK. n = 3 per group. R: replicate. (B) Venn diagram depicting the overlapping genes. (C–F) Significant (p-value or adj.p-value <0.05) enrichment of transcription factor binding (ENCODE) and Top 5 GO terms (Molecular Function) identified by Enrichr. (G–I) Protein-protein interaction network analysis for the three highlighted MCODE components identified in Metascape. (J) Proposed RNA-seq data-driven model of the molecular mechanisms underlying the sustained antidepressant effects produced by (2S,6S)-HNK.

Figure 5.. EZh2-mediated H3K27me3 upregulation in the aPVT drives antidepressant-like behaviors

(A) Representative image showing H3K27me3 and NeuN expression in smSDS-exposed BALB mice 7 days after (2S,6S)-HNK or saline treatment. Scale bar, 100 μm. (B) Cumulative distribution of immunofluorescent signals for H3K27me3 in aPVT cells. n = 27,702-29,440 cells from 6 mice per group. (C) Schematic of AAV microinjection. (D) Representative image showing saCas9-HA and Ezh2 expression. Scale bar, 200 and 100 μm for low and high magnification images. (E) Quantification of Ezh2 signals in HA-positive and HA-negative aPVT cells of mice expressing sgControl and sgEzh2. (F) Experimental design for Ezh2 knockdown. (G–J) Time spent in target zone of the SIT-1 (G) and sucrose preference of the SPT-1 (H) and in SIT-2 (I) and SPT-2 (J) after (2S,6S)-HNK treatment. n = 16–21 (G and H) and n = 10–11 (I and J) per group. (K) Schematic of AAV microinjection. (L) Experimental design for spatiotemporal overexpression of Ezh2-WT or its constitutive active mutant Ezh2-Y641F. (M) Representative image showing HA/tdTomato and H3K27me3 expression in mice expressing tdTomato, wildtype Ezh2, or Ezh2-Y641F. Scale bar, 100 μm. (N) Quantification of H3K27me3 immunostaining in mice expressing tdTomato, Ezh2-WT, or Ezh2-Y641F in the aPVT. n = 8 per group. (O and P) Time spent in target zone of the SIT (O) and sucrose preference of the SPT (P) in smSDS-exposed mice overexpressing tdTomato, wildtype Ezh2, or Ezh2-Y641F in the aPVT. n = 14–16 per group. *p < 0.05, ****p < 0.0001. Bar graphs show the mean ± SEM.

Figure 6.. KDM6B-mediated H3K27me3 regulation is critical for depression-like behaviors and the behavioral response to (2S,6S)-HNK.

(A) Schematic of AAV microinjection. (B) Experimental design for spatiotemporal KDM6 knockdown in BALB mice. (C) Representative image showing GFP and KDM6A or KDM6B expression in the aPVT of mice. Scale bar, 200 μm. (D) Quantification of KDM6A and KDM6B immunostaining in the aPVT of mice expressing miR-KDM6 or miR-Scramble. n = 5–6 per group. (E) Representative image showing GFP, H3K27me3, and Histone H3 expression in the aPVT of mice expressing miR-KDM6 or miR-Scramble. Scale bar, 200 μm. (F and G) Cumulative distribution (left) and mean value (right) of immunofluorescent signals for H3K27me3 (F) and Histone H3 (G) in GFP-positive neurons. n = 6 mice per group. (H) Time spent in target zone of the SIT. n = 12–13 per group. (I) Schematic of AAV microinjection into the aPVT of BALB mice. (J) Experimental design. BALB mice were tested by SIT (SIT-1) and SPT (SPT-1). Mice injected with AAV-KDM6B-CT-WT were treated with (2S, 6S)-HNK and tested gain with SIT (SIT-2) and SPT (SPT-2). (K) Representative image showing GFP/HA and H3K27me3 levels in the aPVT of mice expressing GFP, KDM6B-CT-WT, or KDM6B-CT-H1388A. Scale bar, 200 μm. High magnification images are shown in Figure S8G. (L) Quantification of H3K27me3 levels in the aPVT of mice expressing GFP, KDM6B-CT-WT, or KDM6B-CT-H1388A. (M–P) Time spent in target zone of the SIT-1 (M) and sucrose preference of the SPT-1 (N) and in SIT-2 (O) and SPT-2 (P) after (2S,6S)-HNK treatment. n = 14–19 (M and N) and n = 9–10 (O and P) per group. (Q) Experimental design for KDM6B nuclear localization in smSDS mice receiving (2S, 6S)-HNK or saline. (R) Representative image showing KDM6B and NeuN expression in the aPVT. Scale bar, 50 μm. (S) Quantification of KDM6B signals in the cytoplasm and nucleus of NeuN-positive neurons in the aPVT. n = 8 per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Bar graphs show the mean ± SEM.

Figure 7.. GNB3 regulates behavioral responses to (2S,6S)-HNK and stress

(A) Schematic of AAV microinjection. (B) Representative image showing GFP expression in the aPVT of mice injected with AAV-HA-Gnb3. Scale bar, 200 μm. (C) Western blotting showing successful HA-Gnb3 protein overexpression in mice injected with AAV-HA-GNB3. (D) Experimental design. (E–H) Time spent in target zone of the SIT-1 (E) and sucrose preference of the SPT-1 (F) and in SIT-2 (G) and SPT-2 (H) after (2S,6S)-HNK treatment. n = 15–19 (E and F) and n = 9–10 (G and H) per group. (I) Schematic of AAV microinjection. (J) Representative image showing saCas9-HA expression in the aPVT of mice injected with AAV-sgGnb3. Scale bar, 200 μm. (K) Western blotting showing successful GNB3 protein knockdown in mice injected with AAV-sgGnb3. (L) Experimental design for Gnb3 knockdown in BALB mice. (M) Time spent in target zone of the SIT in smSDS mice injected with either AAV-sgControl or AAV-sgGnb3. n = 15–16 per group. (N) Sucrose preference of the SPT in smSDS mice injected with either AAV-sgControl or AAV-sgGnb3. n = 15–16 per group. (O) Experimental design for ChIP assay in BALB mice. (P) Primer set against putative SUZ12/EZH2 binding sites (ENCODE) on the Gnb3 gene indicated by a black bar. (Q) Levels of H3K27me3 occupancy on the Gnb3 promoter in the aPVT of smSDS-exposed BALB mice or nonstressed (NS) conditions. n = 4 per group. (R) Levels of Gnb3 mRNA in the aPVT of BALB mice expressing EGFP, KDM6B-CT-WT, or KDM6B-CT-H1388A as shown in Figure 6I. n = 6 per group. (S) Levels of Gnb3 mRNA in the aPVT of BALB mice expressing Ezh2-WT, Ezh2-Y641F, or NLS-tdTomato as shown in Figure 5K. n = 6 per group. (T) qPCR quantification of Gnb3 expression in nonstressed and CSDS-exposed susceptible B6 mice seven days after the treatment with (2S,6S)-HNK or saline. n = 8 in each group. (U) Proposed model for how psychosocial stress and (2S, 6S)-HNK regulate Gnb3 gene expression via KDM6- and Ezh2-dependent epigenetic mechanisms, respectively. *p < 0.05, **p < 0.01, ****p < 0.0001. Bar graphs show the mean ± SEM.

Figure 8.. Extrasynaptic GABA_ARs are requited for (2S,6S)-HNK’s antidepressant-like effects

(A) Schematic of fosTRAP2 labeling. (B) Double immunofluorescence staining of GFP and Fos in the aPVT. (C–E) Quantification of TRAP-GFP labeled cells (C), Fos⁺ cells (D), and TRAP-labeled cells with Fos- and TRAP-double labeling (Fos⁺GFP⁺/GFP⁺) (E) in the aPVT. n = 8 per group. (F) Ribo-tag strategy. (G and H) qPCR quantification of Gabra1, Gabra4, Gabrb2, and Gabrd expression in the aPVT of GFP-positive TRAPed cells (relative to saline) (D) and GFP-negative non-TRAPed cells (E). n = 5 pooled samples; each sample was pooled from four mice (n = 20 per group). (I) Representative traces of the membrane current. (J) Holding current changes before and after TTX+PTX application. n = 7–8 neurons per group. (K) Schematic of AAV microinjection into the aPVT of BALB mice. (L) Representative image showing mCherry expression in the aPVT of BALB mice expressing miR-a4/b2/d. Scale bar, 200 μm. (M) Experimental design for the Gabra4/Gabrb2/Gabrd knockdown in BALB mice. (N) Quantification of Gabra1, Gabra4, Gabrb2, Gabrb3, Gabrd, and Gabrg mRNA in the aPVT. n = 6 per group. (O) Time spent in target zone of the SIT. n = 15–16 per group. (P) Sucrose preference of the SPT. n = 15–16 per group. (Q) Proposed model whereby (2S,6S)-HNK drives the sustained antidepressant effects (see Discussion). *p < 0.05, **p < 0.01, ***p < 0.001. Bar graphs show the mean ± SEM.