Ketamine exerts its sustained antidepressant effects via cell-type-specific regulation of Kcnq2

A single sub-anesthetic dose of ketamine produces a rapid and sustained antidepressant response, yet the molecular mechanisms responsible for this remain unclear. Here, we identified cell-type-specific transcriptional signatures associated with a sustained ketamine response in mice. Most interestingly, we identified the Kcnq2 gene as an important downstream regulator of ketamine action in glutamatergic neurons of the ventral hippocampus. We validated these findings through a series of complementary molecular, electrophysiological, cellular, pharmacological, behavioral, and functional experiments. We demonstrated that adjunctive treatment with retigabine, a KCNQ activator, augments ketamine’s antidepressant-like effects in mice. Intriguingly, these effects are ketamine specific, as they do not modulate a response to classical antidepressants, such as escitalopram. These findings significantly advance our understanding of the mechanisms underlying the sustained antidepressant effects of ketamine, with important clinical implications.

We demonstrated that (1) systemic pharmacological manipulation of KCNQ channels modulates antidepressant behaviors in mice, (2) adjunctive treatment with retigabine, a KCNQ activator, augments ketamine’s effects, and (3) the effects of KCNQ are specific to ketamine, as they do not modulate the response to classical antidepressants. We provided new insights into the molecular mechanisms underlying the sustained antidepressant effects of ketamine and postulated the voltage-gated potassium channel KCNQ as a downstream regulator of ketamine action, as well as a promising target for the development of MDD treatments.

In this study, using single-cell RNA sequencing (scRNA-seq), we comprehensively cataloged the transcriptome of thousands of ventral hippocampus (vHipp) cells of mice treated with a single dose of (R,S)-ketamine or a saline vehicle control and found cell-type-specific transcriptional signatures associated with the sustained antidepressant effects of ketamine. Notably, we identified Kcnq2 as an important downstream regulator of ketamine action in glutamatergic neurons of the vHipp. These findings were validated using glutamatergic neurons sorted from a conditional reporter mouse line, in vitro treatment of primary hippocampal neurons, electrophysiological recordings in vitro and from acute vHipp slices, a validated mouse chronic stress model for depression, as well as a viral-mediated knockdown of Kcnq2 in the vHipp of mice. In addition, we identified a previously unknown mechanism of action for ketamine via Kcnq2 in glutamatergic neurons of the hippocampus.

Previous studies have investigated the molecular mechanisms underlying the antidepressant effects of ketamine, and several mechanisms of action have been proposed (). However, given that the antidepressant effects of ketamine remain long after its metabolism, its mechanism of action cannot be solely attributed to ketamine’s ability to inhibit NMDR or increase the function of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPAR). Instead, it may result from the activation of downstream signaling cascades causing long-lasting and sustained adaptations in key brain areas and neural circuits. Despite the progress (), the exact mechanism of ketamine action is still not fully understood. It is possible that some of the more elusive molecular components of this mechanism remain unclear due to important methodological limitations, specifically the absence of cell-type-specific information (). Previous gene expression (GE) studies provided data from brain homogenates; thus, any treatment response signature specific to a particular cell type is averaged out ().

The discovery that a single sub-anesthetic dose of ketamine, a glutamate N-methyl-D-aspartate (NMDA) receptor blocker, produces a rapid yet sustained antidepressant response, even in treatment-resistant patients, is one of the most significant improvements in the field of depression in over 60 years (). Ketamine is effective for the treatment of suicidal ideation in emergency room settings, and its antidepressant effects have been demonstrated in many human and animal studies (). Despite these encouraging clinical and preclinical findings, controversy remains regarding ketamine’s efficacy in treating major depressive disorder (MDD) (). This has encouraged research into ketamine’s mechanisms of action in an effort to understand its primary and downstream targets, so that better treatment interventions for depression can be developed.

Finally, we examined whether retigabine could increase the sustained antidepressant-like effects produced by a single injection of ketamine in mice. We treated a new cohort of mice with a single dose of saline, ketamine, or ketamine in combination with retigabine (1 mg/kg) and assessed antidepressant-like behaviors using the FST, 5 and 7 days post-injection ( Figure 8 A). Consistent with other studies, ketamine alone induced a significant decrease in immobility time during the FST that was sustained for several days post-injection (up to 5 days); however, these effects disappeared by day 7 ( Figures 8 B and 8C). Ketamine in combination with retigabine produced a significant decrease in immobility time during the FST at all time points (days 2, 5, and 7). Combined treatment had significantly stronger effects than ketamine alone at all time points ( Figures 8 B and 8C). Next, as a potential therapeutic strategy for MDD, we further investigated the synergistic effects of ketamine and retigabine by testing whether or not retigabine can increase the antidepressant-like effects produced by ketamine in mice. First, mice were treated with ketamine at sub-effective concentrations of 1 and 5 mg/kg/BW, an effective dose (10 mg/kg/BW), or a saline vehicle control and then assessed in the FST ( Figure 8 D). Consistent with our previous findings, we found a significant reduction in immobility time in the FST only in mice treated with 10 mg/kg of ketamine, since lower doses failed to produce antidepressant-like effects ( Figures 8 E (left) and S8 A). Remarkably, the combined treatment of ketamine with retigabine (1 mg/kg/BW) produced a significant reduction in immobility time during the FST in mice treated with both 5 and 10 mg/kg/BW of ketamine ( Figure 8 E (right) and S8 A). Next, we tested whether retigabine can produce similar synergistic effects with traditional antidepressants. We selected escitalopram, a selective serotonin reuptake inhibitor and commonly prescribed antidepressant in humans that has shown efficacy in preclinical studies (). Interestingly, we found that retigabine does not increase the antidepressant-like effects produced by escitalopram at any of the concentrations tested ( Figures S8 B and S8C). Lastly, using a new cohort of Nex-Cre-Ai9 mice, we tested whether the adjunctive treatment of ketamine with retigabine, or escitalopram alone, can modify the expression of Kcnq2 mRNA in FACS-sorted cells from the vHipp ( Figure 8 F). Consistent with our previous findings, we found a significant increase in Kcnq2 mRNA in GLUT neurons (tdTomato+) of the vHipp, 2 days after treatment ( Figure 8 G, left). The combined treatment with retigabine led to a stronger increase in Kcnq2 mRNA in GLUT neurons, as compared with ketamine alone ( Figure 8 G, left). We did not find any significant changes in Kcnq2 mRNA expression after the treatment with escitalopram ( Figure 8 G, left) or with any of the medications tested in other cell types of the vHipp (tdTomato−) ( Figure 8 G, right). Interestingly, we found that none of the drugs tested increased the mRNA levels of Kcnq2 in GLUT neurons (tdTomato+) or other cell types (tdTomato−) from the vHipp, 30 min after treatment ( Figures S8 D and S8E). These findings not only complement and verify our previous results but suggest that the adjunctive treatment with retigabine augments the antidepressant-like effects of ketamine at later time points and lower dosages. Moreover, our results imply that the synergistic effects of retigabine are specific to ketamine and not traditional antidepressants, suggesting that Kcnq2 may play an important role in the sustained but not the immediate antidepressant effects of ketamine. In summary, adjunctive treatment with retigabine increases the sustained antidepressant-like effects of ketamine in mice, indicating the KCNQ channel as a promising target for MDD treatment.

(G) Boxplots represent qPCR mRNA levels of Kcnq2 in tdTomato+ and tdTomato− cells. (n = 4, per condition). One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method. Data represented as mean ± SEM.p < 0.0001,p < 0.001,p < 0.05. See also Figure S8

(F) Overview of treatment. Mice were injected with saline, escitalopram (10 mg/kg/BW), or ketamine (10 mg/kg/BW), in the absence or in combination with retigabine (1 mg/kg/BW).

(D) Schematic overview of pharmacological manipulation. Mice were injected with saline, ketamine (1, 5, or 10 mg/kg/BW) in the absence or in combination with retigabine (1 mg/kg/BW).

(A) Overview of treatment. Each mouse was injected with saline or ketamine (10 mg/kg/BW) in the absence or in combination with retigabine (1 mg/kg/BW).

Furthermore, in order to determine whether pharmacological manipulation of KCNQ channels, following ketamine treatment, modulates the behavior of mice in an enriched group environment, a new cohort of 96 mice (24 groups) was introduced into the SBs. Baseline behaviors were assessed as above. On day 5, mice were administered 2 injections for a combination of either ketamine, retigabine (1 or 5 mg/kg), or a saline control ( Figure 7 H). An additional group of mice received a combination of ketamine and XE991 (1 mg/kg) ( Figure S7 E). The concentrations of retigabine and XE991 were based on our FST results ( Figures 7 C and S6 I). Using a prediction algorithm for ketamine response (P-KET) developed and trained on our original cohort of mice ( Figures 7 E–7G), we monitored the behavior of all mice after treatment with saline or ketamine alone, or ketamine in combination with KCNQ modulators, for 2 additional nights. Consistent with the FST, we found a significant difference between saline- and ketamine-treated mice ( Figure 7 I). The combination with retigabine (1 mg/kg) produced a stronger effect in the SB, as compared with saline-treated mice (FC: 0.72, p = 0.0005). Again, ketamine-treated mice from this cohort displayed more social and anxiolytic behaviors. We did not find any significant differences between mice treated with a combination of ketamine and retigabine (5 mg/kg) or XE991, as compared with saline-treated controls ( Figures 7 I and S7 E), suggesting that the combination of ketamine and retigabine at higher doses, or inhibition of KCNQ channels, eliminates the antidepressant-like effects of ketamine in mice. These results are consistent with our FST results and suggest that ketamine and pharmacological manipulation of KCNQ channels modulate antidepressant-like behaviors in a semi-naturalistic living environment.

Next, we used the “social box” (SB) to assess the sustained antidepressant effects of ketamine. The SB system is ideally suited for in-depth investigations of pharmacological manipulations, since it allows long-term, continuous tracking of the social behaviors of groups of mice, in an ethologically relevant environment, with minimal experimental intervention (). Adult male mice in groups of four were introduced into SBs ( Figure 7 D), under continuous video observation for 5 days and 6 nights ( Figure 7 E). The first 4 nights were used to establish individual and group baseline behaviors. Before the start of the dark phase on day 5, mice were administered either ketamine or saline. All mice were subsequently returned to a clean SB for response monitoring over the following 2 nights. This procedure was performed on an initial cohort of 64 mice (16 groups), allowing assessments of individual differences in ketamine response and establishment of an analysis pipeline for the SB data. A description of the pipeline is available in STAR Methods Figures 7 F, 7G, and S7 A–S7D). Based on our partial least squares discriminant analysis (PLSDA) classifier, mice treated with ketamine spent more time exploring in an open area of the SB arena, using the distal feeder (feeding and drinking away from the nest), exploring in the central labyrinth, approaching others in the group, and engaged in more social behaviors with other members of their group, such as nose to nose contacts. On the other hand, these mice also spent less time around the walls and inside the main nest. These measures are all associated with anxiolytic behaviors in mice ().

Previous studies in rodents have shown that KCNQ function can be regulated with highly specific agonists and antagonists (). To further explore the KCNQ channel as a potential therapeutic target of ketamine, we examined whether pharmacological inhibition of KCNQ alone or in combination with ketamine had any effects on behavior. Mice received two injections with a combination of drugs. We first treated mice with XE991, a potent and selective KCNQ (Kcnq2/3) channel inhibitor (), using different concentrations (1 and 3 mg/kg/BW), in the absence or in combination with ketamine and compared them with saline-treated controls. To assess the behavioral effects elicited by ketamine, XE991, and its combination, we exposed these mice to a FST ( Figure 7 A, left). As expected, we found a significant decrease in immobility time during the FST in mice treated with ketamine, as compared with saline-treated mice ( Figure 7 B). We did not find any significant differences in mice that received XE991 in combination with saline ( Figure S6 I), suggesting that XE991 alone does not produce any acute behavioral effects in the FST. Consistent with our hypothesis, the antidepressant effects of ketamine were abolished in mice that were treated with both XE991 and ketamine ( Figures 7 B and S6 I). These results indicate that KCNQ activity might be necessary for ketamine to exert its antidepressant actions. We then sought to test whether we could mimic, increase, or amplify the effects of ketamine using retigabine, a selective KCNQ (Kcnq2/3) channel activator (). As in the previous experiment, mice were treated with saline, ketamine, or ketamine in combination with two different concentrations of retigabine (1 and 5 mg/kg) ( Figure 7 A, right). Again, we observed a significant effect of ketamine as compared with saline controls in the FST ( Figure 7 C). We did not find any effects of retigabine when it was administered in combination with saline alone ( Figure S6 J). However, in combination with ketamine, retigabine (1 mg/kg) produced a stronger effect in the FST that was significantly different versus ketamine alone (FC: 0.71, p = 0.0001) ( Figures 7 C and S6 J). Interestingly, the combined administration of ketamine and retigabine at a higher dose (5 mg/kg) did not produce any acute behavioral effects in the FST. This is in line with ketamine’s inverted U-shaped dose response (). These results suggest that pharmacological manipulation of KCNQ modulates antidepressant-like behavior.

(I) Boxplots represent response to ketamine (P-KET) in the SB. Conditions: saline (gray), saline and ketamine (dark blue), ketamine and retigabine (dark orange). One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method. Data represented as mean ± SEM.p < 0.001,p < 0.01,p < 0.05. See also Figures S6 and S7

(F and G) Behavioral outcomes from the SB were summarized as change from the mean over the baseline days and used as input for partial least squares discriminant analysis (PLS-DA).

(E) Experimental timeline. On day 5, all mice were removed from the SBs and injected with ketamine (10 mg/kg/BW, n = 3 per group) or a saline solution (n = 1 per group). Following a recovery period, all mice were returned to a clean SB at the start of the dark phase for response monitoring.

(A) Schematic overview of pharmacological manipulation of KCNQ using XE991 or retigabine. Mice were treated with saline, ketamine (10 mg/kg/BW) alone, or a combination of ketamine with XE991 (1 and 3 mg/kg/BW) or retigabine (1 and 5 mg/kg/BW).

Having identified Kcnq2 as a downstream regulator of the sustained antidepressant effects of ketamine, we wanted to further investigate how ketamine transcriptionally upregulates Kcnq2 mRNA to exert its sustained antidepressant-like effects. Previous studies have successfully shown that Kcnq2 mRNA can be regulated via calmodulin (CaM), calcineurin (CaN), and the A-kinase-anchoring protein 5 (AKAP5) (). This complex activates the transcription factor NFAT (nuclear factor of activated T cells), which acts on Kcnq2 gene regulatory elements, as opposed to Kcnq3 (). As a result, enhanced and sustained Kcnq2 transcription leads to increased M channel activity and regulation of neuronal excitability. Interestingly, the CaM genes, Calm1 and Calm2, as well as Akap5, were among the DEGs of our scRNA-seq analysis ( Table S2 ). Having already shown that ketamine increases the mRNA expression of Kcnq2 in primary hippocampal neurons ( Figure 3 A), we next wanted to investigate whether ketamine regulates Kcnq2 mRNA via Ca, CaM, CaN, or AKAP5 signaling. We used pharmacological inhibition of the key components of this pathway to see whether it interferes with the upregulation of Kcnq2 mRNA by a single ketamine treatment in vitro. Primary hippocampal neurons were stimulated with a single dose of either saline, HNK, or a combination of HNK plus nifedipine (L-type Cachannel blocker), W-7 hydrochloride (CaM inhibitor), or cyclosporine-A (CaN inhibitor). Neurons were collected at 4 time points: 30 min, 1, 2, or 6 h (post-treatment) and compared with a group of untreated controls ( Figures 6 A–6D ). We did not find any significant differences in the expression of Kcnq2 mRNA 30 min after treatment ( Figure 6 A). However, consistent with our results ( Figure 3 A), we found a significant upregulation of Kcnq2 mRNA 1, 2, and 6 h after a single treatment with HNK ( Figures 6 B–6D). Interestingly, blockade of L-type Cachannels and CaM by nifedipine and W-7, respectively, eliminated the upregulation of Kcnq2 by ketamine at all time points tested ( Figures 6 B–6D). In contrast, inhibition of CaN only abolished the effects of ketamine on Kcnq2 mRNA 6 h after treatment ( Figure 6 D). These results, summarized in Figure 6 E, suggest that Kcnq2 mRNA changes are not directly or immediately induced after ketamine treatment but the result of downstream events initiated by ketamine. Furthermore, our findings show that the ketamine-induced increase in the expression of Kcnq2 mRNA was blocked by nifedipine, W-7, or cyclosporine-A, suggesting a critical role for L-type Cachannels, CaM, and CaN in the transcriptional regulation of Kcnq2 by ketamine treatment at later time points.

(E) Schematic model for transcriptional regulation of Kcnq2. NMDAR inhibition (via ketamine) or increased AMPAR function (via HNK) leads to an increase in L-type calcium channel (L-VDCC) activity, creating elevated levels of intracellular calcium (Ca 2+ ) leading to (1) the already known mechanisms of ketamine action or (2) a novel mechanism via transcriptional regulation of Kcnq2 mRNA by the Akap5-CaM-CaN complex.

(A–D) Hippocampal primary neurons (mouse) stimulated with either saline solution, HNK (10 μM), or a combination of HNK and nifedipine (calcium channel blocker), W-7 hydrochloride (calmodulin inhibitor), or cyclosporine-A (calcineurin inhibitor) for 30 min, 1, 2, or 6 h, and compared with an untreated control. n = 4 (biological replicates) per condition and treatment. Boxplots represent qPCR mRNA levels of Kcnq2. One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method. Data represented as mean ± SEM. ∗∗∗∗ p < 0.0001, ∗∗∗ p < 0.001, ∗ p < 0.05.

In order to test whether CSDS-induced Kcnq2 decrease can be reversed by ketamine treatment, we exposed a new group of Nex-Cre-Ai9 mice to the CSDS model. On day 11, mice were treated with either ketamine or a saline control ( Figure 5 D). The antidepressant effects of ketamine were assessed 2 days after treatment (day 13), using the FST. In addition, the vHipp of saline- and ketamine-treated CSDS mice were dissected, and individual cells were sorted using FACS. Interestingly, ketamine reversed the effects of chronic stress in the FST ( Figure 5 E) and Kcnq2 mRNA in GLUT neurons ( Figure 5 F). Kcnq2 mRNA expression was significantly increased to baseline levels (as seen in non-stressed controls, Figure 5 C) in ketamine-treated CSDS mice versus saline-treated controls ( Figure 5 F). In addition, these effects were specific to GLUT neurons (tdTomato+) of the vHipp ( Figure 5 F, left), since we did not find any significant differences in tdTomato− cells ( Figure 5 F, right). We found no significant increase in Kcnq2 mRNA in ketamine-treated CSDS mice using bulk mRNA ( Figure S6 H). These results demonstrate that Kcnq2 mRNA is altered after chronic stress exposure in GLUT neurons of the vHipp and that these effects can be reversed by ketamine treatment. Furthermore, these cell-type-specific effects could be diluted or distorted when using brain homogenates.

To further investigate the role of Kcnq2 in GLUT neurons of the vHipp, a new group of Nex-Cre-Ai9 mice was subjected to the chronic social defeat stress (CSDS) model, a validated paradigm to induce long-lasting depression- and anxiety-like endophenotypes in mice ( Figure 5 A) (). 10 days of CSDS exposure led to hallmark features of chronically stressed mice ( Figures S6 C–S6F). In the FST, chronically stressed mice showed a significant increase in immobility time, as compared with non-stressed controls ( Figure 5 B). Endpoint and tissue collection were performed 24 h after the last social defeat session (day 11), thus capturing the cumulative effects of chronic stress, rather than any acute effects of the last defeat. The vHipp was dissected, and individual cells were sorted (as described in Figure 2 A). We found that Kcnq2 mRNA expression was decreased only in GLUT neurons (tdTomato+) of the vHipp, as we did not find any significant differences in the remaining cell types (tdTomato−) ( Figure 5 C). We also found a significant difference in Kcnq2 mRNA expression between tdTomato+ and tdTomato− cells, confirming that Kcnq2 is enriched in GLUT neurons of the vHipp. Finally, to access bulk mRNA expression levels of Kcnq2 following chronic stress, we quantified its expression in vHipp brain punches from a new cohort of chronically stressed and control mice. Although there was a decrease in Kcnq2 mRNA levels in stressed mice versus controls, it did not reach statistical significance ( Figure S6 G; p = 0.08).

(F) Boxplots represent qPCR mRNA levels of Kcnq2 in tdTomato+ and tdTomato− cells between chronically stressed mice that received a saline (n = 4, purple) or ketamine (n = 4, blue) treatment. One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method. Data represented as mean ± SEM.p < 0.001,p < 0.01. See also Figure S6

(C) Boxplots represent qPCR mRNA levels of Kcnq2 in tdTomato+ and tdTomato− cells between non-stressed controls (n = 4, gray) and CSDS mice (n = 4, purple). One-way ANOVA. Multiple testing correction was performed using the Benjamini-Hochberg method.

The M channel (KCNQ) is formed by the proteins encoded by the Kcnq2 and Kcnq3 genes, both integral membrane proteins (). We did not find any significant differences in the mRNA expression of Kcnq3 after ketamine treatment in our original cohort, our FACS-sorted sample, or in primary hippocampal neurons ( Figures S4 A–S4C; Table S2 ), suggesting that ketamine produces an effect specific to Kcnq2, but not to Kcnq3. To explore the expression of Kcnq2 and Kcnq3 in the brain, we examined publicly available in situ hybridization (ISH) data from 12 different regions of the mouse brain (). Kcnq2 shows its highest expression in the hippocampus formation, whereas Kcnq3 is highly expressed throughout multiple brain regions ( Figures S4 D–S4F). In addition, our single-cell data, as well as other publicly available single-cell datasets, show that Kcnq2 is exclusively expressed in neurons, whereas Kncq3 is expressed in neurons, astrocytes, oligodendrocytes, and OPCs () ( Figures S5 A–S5F). These findings reinforce the idea that Kcnq2 plays an important and more centralized role in neurons of the hippocampus, making this region a good candidate for in vivo viral manipulations. To functionally explore the role of Kcnq2 in mediating the antidepressant effects of ketamine, we designed adeno-associated virus (AAV) constructs to knock down Kcnq2 in vivo ( Figure 4 A). Transfection of Neuro2a (N2a) cultured cells and viral injection into the vHipp of adult mice resulted in a significant decrease in Kcnq2 mRNA levels, both in vitro and in vivo ( Figures 4 B and 4C). We injected shRNA-Kcnq2 or shRNA-scramble control AAVs bilaterally into the vHipp of a new cohort. Four weeks after viral injection, half of the mice were randomly selected to receive a ketamine or saline injection, and the antidepressant effects of ketamine were assessed using the forced swim test (FST), a validated test for evaluating antidepressant efficacy in rodents () ( Figures 4 D and 4E). In the group of mice treated with an shRNA-scramble control, we found a significant decrease in immobility time during the FST in mice treated with ketamine versus controls ( Figure 4 E, left). Notably, the antidepressant effects of ketamine were no longer detected in mice expressing the shRNA-Kcnq2 virus ( Figure 4 E, right). We also assessed locomotor activity to rule out any confounding effects of hyperlocomotion in the FST after ketamine treatment and found no differences in total activity between groups ( Figures S6 A and S6B). These results indicate that the vHipp is an important site for Kcnq2 function and the sustained antidepressant effects of ketamine.

(E) Boxplots represent total immobile time (S) during the forced swim test (FST) in mice that received shRNA-Ctrl (gray) or shRNA-Kcnq2 (green) after ketamine (dark green) or saline injection (dark gray) (n = 10, per condition). The FST was performed 2 days after treatment. Data represented as mean ± SEM. Two-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg methodp < 0.001,p < 0.01,p < 0.05. See also Figures S4–S6

(D) Representative images of mouse brains injected with shRNA-Kcnq2 and shRNA control AAV. Fluorescent eGFP signal (green); DAPI signal (blue). Mice were injected in the vHipp (bilateral) with either shRNA-Kcnq2 (n = 22) or shRNA control (n = 19) AAV. 4 weeks after viral injection, mice were randomly selected to receive a ketamine (10 mg/kg/BW) or saline injection.

(C) Coronal map with the region and bregma coordinates (mm) targeted for in vivo viral manipulation. AP anteroposterior, ML mediolateral, DV dorsoventral. Bar plots represent qPCR mRNA expression levels (n = 5, per condition). qPCR data were normalized to the geometric mean of Gapdh and Rpl13. One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method.

(B) Representative images of N2a cells transfected with an shRNA control or two different shRNA-Kcnq2 vectors (n = 5, per condition). Fluorescent eGFP signal (green); DAPI signal (blue). Boxplots represent qPCR mRNA expression levels of Kcnq2. One-way ANOVA, corrected for multiple comparisons.

The Kcnq2 gene encodes for the Kv7.2 protein, a slow-acting, voltage-gated potassium channel that plays a critical role in the regulation of neuronal excitability (). The Kv7.2 and Kv7.3 proteins (Kcnq3 gene) can form KCNQ (Kv7) homo- or heterotetramers that generate a signature M-current, modulating the overall excitability of neurons (). To investigate the KCNQ channel as a mediator of the sustained antidepressant effects of ketamine, we treated mouse primary hippocampal neurons with HNK (10 μM) or a saline control for 24 h and quantified M-current density (I) using whole-cell voltage-clamp recordings ( Figure 3 B). Neurons treated with HNK displayed a significant increase in Icurrent density versus saline-treated controls ( Figures 3 C and 3D). Next, we performed ex vivo patch-clamp recordings from acute hippocampal slices to test the effects of ketamine treatment on Icurrent density ( Figure 3 E). Briefly, mice were injected with ketamine or a saline control and sacrificed 2 days later. Electrophysiological recordings were taken specifically from glutamatergic (CA1 pyramidal) neurons of the vHipp. Similarly to our in vitro findings, ketamine-treated mice showed a significant increase in Icurrent density versus saline-treated controls ( Figures 3 F, 3G, S3 H, and S3I), suggesting that ketamine increases the expression of KCNQ channels in vivo. These findings indicate that the increased Kcnq2 mRNA expression observed in glutamatergic neurons 2 days after ketamine treatment is accompanied by a significant gain in the number of functional KCNQ channels expressed in GLUT neurons of the hippocampus both in vitro and in vivo. Together, our findings further support the idea that KCNQ channels are a downstream regulator of sustained ketamine action and suggest that modulation of these channels in GLUT neurons of the vHipp could be a potential target for MDD treatment.

Next, we examined whether a single ketamine treatment of primary hippocampal neurons could modify the mRNA expression of the 8 genes tested earlier. Mouse primary hippocampal neurons are mostly made up of GLUT neurons and therefore make a very good model system to further validate our previous in vivo findings. Primary neurons were cultured for 21 days and then stimulated with a single dose of ketamine (10 μM) or its active metabolite, (2R, 6R)-hydroxynorketamine (HNK) (10 μM). These concentrations were selected after careful experimental consideration and review of the available literature ( Figures S3 A–S3G). Following the treatment, neurons were collected 2, 12, 24, or 48 h later and compared with untreated and saline controls ( Figure 3 A). We found no significant differences in the expression of any of the 8 genes tested after treatment with a saline control, as compared with the untreated primary neurons ( Figure 3 A). However, we found significant changes in the mRNA expression of 7 out of 8 genes after ketamine treatment, which were directionally consistent with our scRNA-seq findings ( Figure 3 A). Among the DEGs, Kcnq2 displayed the largest changes in GE after treatment, showing a significant upregulation at all time points. Our findings suggest that Kcnq2 is transcriptionally regulated by ketamine and could be an important downstream regulator of ketamine action in glutamatergic neurons of the vHipp.

(B–G) Electrophysiological analysis of Itail current density in primary hippocampal neurons and CA1 pyramidal cells from the vHipp in acute brain slices. (B) Primary cultures were treated with saline (gray) or HNK (10 μM, orange red) and assessed 24 h after treatment. (E) vHipp slices were obtained from CD1 mice that received an injection of saline (gray) or ketamine (10 mg/kg/BW) (blue), 2 days before slice preparation. Total Itail current (C and F): representative current traces from whole-cell voltage-clamp recordings describe the amplitude of the total Itail current as obtained during recordings. KCNQ2/3 tail current (D and G): boxplots represent the isolated KCNQ2/3 current density (pA/pF), which is the current amplitude of the Icurrent carried specifically by KCNQ2/3 and divided by the cell capacitance. In vitro (n = 8 cells for each group from 2 independent cultures, gray = saline; orange = HNK). Ex vivo (gray: n = 5, 10 cells; blue: n = 6, 12 cells). n = number of mice (biological replicates). Data represented as mean ± SEM. Unpaired t tests, two tailed.p < 0.001,p < 0.01,p < 0.05. See also Figure S3

(A) Primary hippocampal neurons treated with a single dose of either ketamine (10 μM), HNK (10 μM), or a saline control. Neurons were collected after 2, 12, 24, or 48 h post-injection (n = 4, per condition). Bar plots represent qPCR mRNA levels. All qPCR data were log 2 -transformed and normalized to the geometric mean of Gapdh and Rpl13. One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method.

To compare whole tissue “bulk” versus cell-type-specific methods, we quantified the mRNA expression of these 8 genes using brain punches from the vHipp of a new cohort of mice. Our results showed small changes in a direction that was consistent with our scRNA-seq results; however, these effects did not reach statistical significance ( Figures S2 I and S2J). Overall, these results show that ketamine elicits sustained cell-type-specific GE changes in the vHipp of mice and that these changes are masked in bulk analyses.

The GLUT neurons were the most interesting cell type based on their multigenic response (165 DEGs) ( Figure 1 D) and their known role in modulating the antidepressant effects of ketamine (). Next, we generated a conditional reporter mouse line (Nex-Cre-Ai9) where most GLUT neurons of the forebrain, including the hippocampus, are fluorescently labeled by tdTomato ( Figure S2 A) (). Mice were injected with ketamine or a saline control. The vHipp was dissected 2 days post-injection ( Figure 2 A). Individual whole cells were sorted using fluorescence-activated cell sorting (FACS) into two separate pools of cells from each mouse. One pool contained GLUT neurons (tdTomato+), and the other contained all other vHipp cell types (tdTomato−) ( Figures 2 A and S2 B–S2D). To confirm the presence of GLUT neurons in the tdTomato+ pool, we quantified mRNA and found higher levels of the genes coding for Neurod6, tdTomato, as well as Slc17a7, a known marker of GLUT neurons, as compared with the tdTomato− pool ( Figure S2 E). We quantified mRNA of the established cell-type-specific markers Slc32a1 (GABAergic neurons), Slc1a3 (astrocytes), Mog (oligodendrocytes), C1qc (microglia), and Cldn5 (endothelial cells) and found that tdTomato+ cells expressed lower levels of these genes, as compared with tdTomato− cells ( Figure S2 E), confirming the presence of GLUT neurons in the tdTomato+ cells. We used these two pools to validate our scRNA-seq findings at the population level, using quantitative real-time polymerase chain reaction (qPCR). We found no significant differences in the total number of tdTomato+ cells from the ketamine or saline groups, confirming our scRNA-seq results ( Figures S2 F–S2H). As a proof of principle, we selected 8 of the top DEGs in GLUT neurons (4 upregulated and 4 downregulated) from our scRNA-seq analysis ( Table S2 ). We found that most of the genes quantified in the tdTomato+ cells were significantly dysregulated and directionally consistent with our scRNA-seq findings ( Figure 2 B). The voltage-gated potassium channel subfamily Q member 2 (Kcnq2) showed the strongest effect after ketamine treatment (p < 0.001, FC = 2.2). We found no significant differences in mRNA levels for any of the 8 DEGs in the tdTomato− cells ( Figure 2 C).

(B and C) Boxplots represent qPCR mRNA levels of 8 (4 upregulated and 4 downregulated) DEGs in tdTomato+ (red) and tdTomato− (gray) cells between ketamine- and saline-treated mice. (n = 4, per condition). All qPCR data were normalized to the combined mRNA expression of the endogenous controls, Gapdh and Rpl13. Data represented as mean ± SEM. One-way ANOVA. Multiple testing was corrected using the Benjamini-Hochberg method.p < 0.001,p < 0.05, ‡p < 0.1. See also Figure S2

(A) Nex-Cre-Ai9 mutant mice were injected with ketamine (10 mg/kg/BW) (n = 4) or a saline solution (n = 4). After 2 days, individual cell suspensions were prepared from the vHipp of all mice. Glutamatergic neurons (tdTomato+) and all remaining cell types (tdTomato) from the vHipp were isolated using FACS.

To characterize cell-type-specific molecular changes associated with the sustained antidepressant effects of ketamine, we treated mice with either ketamine or a saline control. Brains were collected 2 days post-treatment, and single-cell suspensions were prepared for scRNA-seq ( Figure 1 A). The transcriptome of thousands of single cells from the vHipp, a well-known region for ketamine antidepressant action in rodents and humans, was sequenced (). After quality control and preprocessing ( Figure S1 A), single cells were grouped according to their GE profiles using graph-based clustering. Uniform manifold approximation and projection (UMAP) plots were used to visualize clusters. Our unsupervised cluster analysis revealed 13 cell clusters with distinct GE signatures ( Figures 1 B and S1 B). We found no significant differences in the relative cell-type composition for each cluster between the ketamine- and saline-treated groups ( Figure 1 C; Table S1 ). Subsequently, we performed differential expression analyses to evaluate cell-type-specific molecular signatures of ketamine action (see STAR Methods ). We identified a total of 263 differentially expressed genes (DEGs) in 7 of the 13 clusters, ranging from 1 to 165 DEGs per cell type ( Figure 1 D; Table S2 ). We found that 31 of the 263 DEGs were significantly dysregulated in more than 1 cluster; however, 135 DEGs were exclusively in glutamatergic (GLUT) neurons, 27 in astrocytes, 16 in oligodendrocytes, 3 in oligodendrocyte progenitor cells (OPCs), 1 in endothelial cells, and 1 in vascular cells ( Figure S1 C; Table S3 ). Additionally, we performed a pathway enrichment analysis for the three cell types with the largest DEGs, using Enrichr (). In GLUT neurons, our analysis revealed many significant pathways involved in calcium signaling, synaptic function and plasticity, and neurodevelopmental disorders ( Figure S1 D). Astrocytes showed an enrichment for fatty acid elongation, gap junction, phagosome activity, and Alzheimer’s disease ( Figure S1 E), whereas oligodendrocytes showed an enrichment for vitamin digestion, absorption, and fatty acid elongation ( Figure S1 F).

(A) Mice were injected with ketamine (10 mg/kg/BW) or saline and sacrificed 2 days post-injection. Cell suspensions were prepared from the ventral hippocampus (vHipp) of ketamine (n = 4) and saline-treated controls (n = 4).

Discussion

Duman, 2018 Duman R.S. Ketamine and rapid-acting antidepressants: a new era in the battle against depression and suicide. Yang et al., 2018 Yang Y.

Cui Y.

Sang K.

Dong Y.

Ni Z.

Ma S.

Hu H. Ketamine blocks bursting in the lateral habenula to rapidly relieve depression. The discovery that a single sub-anesthetic dose of ketamine produces a rapid and sustained antidepressant response is one of the most important breakthroughs in psychiatry (), yet the molecular mechanisms responsible for this remain unclear. Our study describes a previously unknown cell-type-specific molecular mechanism for the sustained effects of ketamine and provides new strategies for MDD treatment.

Aguilar-Valles et al., 2020 Aguilar-Valles A.

De Gregorio D.

Matta-Camacho E.

Eslamizade M.J.

Khlaifia A.

Skaleka A.

Lopez-Canul M.

Torres-Berrio A.

Bermudez S.

Rurak G.M.

et al. Antidepressant actions of ketamine engage cell-specific translation via eIF4E. Bagot et al., 2017 Bagot R.C.

Cates H.M.

Purushothaman I.

Vialou V.

Heller E.A.

Yieh L.

LaBonté B.

Peña C.J.

Shen L.

Wittenberg G.M.

Nestler E.J. Ketamine and imipramine reverse transcriptional signatures of susceptibility and induce resilience-specific gene expression profiles. Carreno et al., 2016 Carreno F.R.

Donegan J.J.

Boley A.M.

Shah A.

DeGuzman M.

Frazer A.

Lodge D.J. Activation of a ventral hippocampus-medial prefrontal cortex pathway is both necessary and sufficient for an antidepressant response to ketamine. Kim et al., 2021 Kim J.W.

Autry A.E.

Na E.S.

Adachi M.

Björkholm C.

Kavalali E.T.

Monteggia L.M. Sustained effects of rapidly acting antidepressants require BDNF-dependent MeCP2 phosphorylation. Moda-Sava et al., 2019 Moda-Sava R.N.

Murdock M.H.

Parekh P.K.

Fetcho R.N.

Huang B.S.

Huynh T.N.

Witztum J.

Shaver D.C.

Rosenthal D.L.

Alway E.J.

et al. Sustained rescue of prefrontal circuit dysfunction by antidepressant-induced spine formation. The vHipp is an important site for the sustained antidepressant effects of ketamine at the electrochemical, molecular, cellular, and circuit levels (). However, due to methodological limitations, such as the absence of cell-type-specific information, some of the more elusive components of these mechanisms are still unidentified. Here, we used scRNA-seq to comprehensively catalog the transcriptome of thousands of single cells from the vHipp of male mice treated with ketamine and found cell-type-specific transcriptional signatures associated with its sustained antidepressant effects. Specifically, we identified a total of 263 DEGs in 7 of the 13 clusters. We identified the gene coding for the voltage-gated potassium channel, Kcnq2, as an important regulator of ketamine action in GLUT neurons of the vHipp. We validated these findings through a series of molecular, cellular, behavioral, pharmacological, and functional experiments using conditional reporter lines, FACS-sorted populations of GLUT neurons, in vitro assays in primary hippocampal neurons, electrophysiological recordings from in vitro cells and ex vivo acute slices, a validated mouse model to induce long-lasting depression- and anxiety-like endophenotypes, as well as a knockdown of Kcnq2 in the vHipp.

Friedman et al., 2016 Friedman A.K.

Juarez B.

Ku S.M.

Zhang H.

Calizo R.C.

Walsh J.J.

Chaudhury D.

Zhang S.

Hawkins A.

Dietz D.M.

et al. KCNQ channel openers reverse depressive symptoms via an active resilience mechanism. Feng et al., 2019 Feng M.

Crowley N.A.

Patel A.

Guo Y.

Bugni S.E.

Luscher B. Reversal of a treatment-resistant, depression-related brain state with the Kv7 channel opener retigabine. Tan et al., 2020 Tan A.

Costi S.

Morris L.S.

Van Dam N.T.

Kautz M.

Whitton A.E.

Friedman A.K.

Collins K.A.

Ahle G.

Chadha N.

et al. Effects of the KCNQ channel opener ezogabine on functional connectivity of the ventral striatum and clinical symptoms in patients with major depressive disorder. Costi et al., 2021 Costi S.

Morris L.S.

Kirkwood K.A.

Hoch M.

Corniquel M.

Vo-Le B.

Iqbal T.

Chadha N.

Pizzagalli D.A.

Whitton A.

et al. Impact of the KCNQ2/3 channel opener ezogabine on reward circuit activity and clinical symptoms in depression: results from a randomized controlled trial. Kim et al., 2021 Kim J.W.

Autry A.E.

Na E.S.

Adachi M.

Björkholm C.

Kavalali E.T.

Monteggia L.M. Sustained effects of rapidly acting antidepressants require BDNF-dependent MeCP2 phosphorylation. From a translational and clinical perspective, our results provide additional evidence that pharmacological modulators of KCNQ channels can regulate antidepressant-like behaviors in mice. Previous studies have shown that chronic or repeated retigabine treatment (8-day intraperitoneal [i.p.] injections) normalizes neuronal hyperactivity and depressive-like behaviors in mice (). Others have shown that neuroinflammation produced by stress exposure leads to overproduction and release of inflammatory cytokines, which ultimately increases neuronal excitability. Notably, these effects can also be reversed by retigabine (). In 2020, a small open-label clinical trial assessed the antidepressant effects of retigabine in MDD patients and showed that chronic treatment (10 weeks) was associated with an improvement in depressive symptoms (). These findings were recently replicated in a small randomized placebo-controlled trial testing the effect of retigabine on clinical outcomes in depressed patients (). In both studies, retigabine was well tolerated, and no serious adverse events were reported. However, none of these studies used ketamine or tested any potential interactions between retigabine and ketamine. Finally, in a recent study (), the authors show that similar to ketamine, scopolamine can exert fast and sustained antidepressant-like effects in mice via BDNF-dependent MeCP2 phosphorylation. Interestingly, scopolamine is a muscarinic acetylcholine receptor (mAChR) antagonist, and the KCNQ channel, also known as “M” channel, is deactivated upon mAChR activation, hinting at the possibility that KCNQ channels might be playing a complementary role exerting the sustained antidepressant-like effects of other rapidly acting antidepressants, such scopolamine. Collectively, our findings not only demonstrate that a single dose of retigabine is sufficient to enhance the antidepressant-like effects of ketamine in mice, but further suggest that (1) it augments the sustained antidepressant-like effects of ketamine (up to 7 days), (2) it can induce similar antidepressant-like effects with lower (sub-effective) ketamine concentrations, which could potentially reduce some of the undesired side effects, and (3) the effects of KCNQ are ketamine specific, as they do not modulate a response to classical antidepressants. These findings have important clinical implications considering that both medications are currently FDA-approved agents and are already in widespread clinical use.

Our high-throughput, cell-type-specific findings for the role of Kcnq2 in GLUT neurons of the vHipp provide a significant advancement to our understanding of the mechanisms underlying a sustained antidepressant effect of ketamine. Future studies should test the effects that the enantiomers and metabolites of ketamine have on KCNQ channels, as well as test whether inhibitors of NMDARs, AMPARs, the opiate system, and other potential signaling molecules affect Kncq2 mRNA and protein expression. Our findings suggest that modulating KCNQ function, in combination with ketamine therapy, may be important in MDD treatment.