Enhanced ERK activity extends ketamine's antidepressant effects by augmenting synaptic plasticity

Abstract

Repeated ketamine treatment to maintain a rapid antidepressant effect can lead to side effects over time, highlighting an unmet clinical need for sustaining this drug's antidepressant action from a single administration. Ketamine-induced synaptic potentiation at CA3-CA1 synapses has been proposed to be a key synaptic substrate for antidepressant action. Here, we found that ketamine-induced CA3-CA1 synaptic potentiation could be augmented by transiently increasing extracellular signal-regulated kinase (ERK) activity through pharmacological inhibition of dual-specificity phosphatases 6 (DUSP6). The antidepressant-like behavioral effects of acute ketamine treatment were extended by DUSP6 inhibition for up to 2 months. The selective deletion of tropomyosin receptor kinase B (TrkB) in excitatory neurons abolished these DUSP6 inhibition-mediated synaptic and behavioral effects. These data suggest that ketamine's rapid antidepressant effects can be sustained by selectively targeting downstream intracellular signaling.

Fig. 1.. Enhanced ERK activity by BCI augments synaptic potentiation in the hippocampus CA1 region.

(A) Mice received two i.p. injections of Vehicle (Veh)/BCI (40 mg/kg) and Saline (Sal)/Ketamine (Ket, 5 mg/kg) with 1 hour apart. At 6 hours, CA1 tissues were dissected for immunoblotting. More immunoblotting and field-recording electrophysiology were performed at 24 hours. The lower diagram of TrkB intracellular signaling shows that DUSP6 represses ERK activity and could be inhibited by an allosteric compound, BCI. (B) Immunoblots of p-ERK in protein samples of mouse CA1 6 hours after treatments; p-ERK1/2 was significantly increased with BCI and ketamine co-treatment (n = 3/group, *p = 0.015 for Veh+Ket vs. Veh+Sal; *p = 0.031 for BCI+Ket vs. Veh+Ket, by one-way ANOVA and Tukey post hoc test). (C) At 24 hours, p-ERK1/2 is increased only by the co-treatment of ketamine and BCI (n = 3, Ket × BCI interaction ***p = 0.007; **p = 0.002 for BCI+Ket vs Veh+Ket). (D) CA1 field recording on hippocampal slices. With the CA3 region removed, Schaffer collaterals from CA3 neurons were stimulated, and field excitatory postsynaptic potential (fEPSP) was recorded in the stratum radiatum of CA1 neurons. Paired-pulse ratio (PPR), input-output (I/O), and fEPSPs for 20 minutes were recorded before (baseline phase 1) and after (potentiation phase 2) ketamine infusion and wash-out on slices without stimulation. (E) fEPSP were recorded before (black lines) and after (colored lines) ketamine infusion in each treatment condition. Synaptic potentiation (%) was assessed by measuring fEPSP slope change at baseline (1) and after 30 minutes of ketamine (Ket, 20 μM) infusion (2) in hippocampal slices. The mean values of potentiation for each group are listed (n = 10 slices from at least 5 mice, Ket × BCI interaction ***p < 0.0001; ***p < 0.0001 for BCI+Ket vs. Veh+Ket: ***p < 0.0001 for Veh+Ket vs. Veh+Sal). (F) The I/O curve of fEPSP in CA1 was measured at the baseline. The slope of I/O curve was increased only after ketamine, and BCI co-treatment (n = 10 slices from at least 5 mice; Ket × BCI interaction at 0.05–0.10 mV is **p = 0.0048, ***p = 0.0001 for BCI+Ket vs. Veh+Ket; Ket × BCI interaction at 0.10–0.15 mV is **p = 0.0051, BCI+Ket vs. Veh+Ket: ***p = 0.0001). Inset: field waveforms from slices recorded at 0.10–0.15 mV presynaptic volley. (G) The cell surface GluA1 and GluA2 protein levels in CA1 were measured by immunoblotting after the biotinylated surface proteins were pulled down from CA1 lysate with streptavidin beads. Only ketamine and BCI co-treatment increased both levels of GluA1 (n = 4, Ket × BCI interaction **p = 0.0025; ***p = 0.0008 for BCI+Ket vs Veh+Ket) and GluA2 (n = 4, Ket × BCI interaction *p = 0.0462; **p = 0.006 for BCI+Ket vs Veh+Ket) in the CA1 region. Data are presented as means ± SEM and analyzed by two-way ANOVA and Tukey post hoc test.

Fig. 2.. Ketamine and BCI co-treatment increases synaptic surface GluA1 and GluA2 protein levels and synapse number at CA1 stratum radiatum.

(A) dSTORM images of synaptic surface GluA1 or GluA2 proteins colocalized with the presynaptic marker, Bassoon, or the postsynaptic marker, Homer1. (B) Images of GluA1 (left panel) and GluA2 (right panel) particle clusters in the s. radiatum CA1 24 hours after treatment, show the largest average size of GluA1 and GluA2 clusters after co-treatment of BCI and ketamine. A typical particle cluster circled in each image is displayed in an inset (scale bar, 150 nm). (C) Images of synapses co-labeled with Bassoon and Homer1 at low (left panel) and high (right panel) magnification. The small square denotes the area for the high-powered image in each treatment. (D) Quantification by field of views (FOVs, 50 × 50 μm) for cluster volume of GluA1 (Ket × BCI interaction ***p < 0.0001; ***p < 0.0001 for BCI+Ket vs Veh+Ket) and GluA2 (Ket × BCI interaction ***p < 0.0001; ***p < 0.0001 for BCI+Ket vs Veh+Ket). (E) Quantification by hippocampal slices for cluster volume of GluA1 (Ket × BCI interaction ***p = 0.0005; ***p < 0.0001 for BCI+Ket vs Veh+Ket) and GluA2 (Ket × BCI interaction ***p = 0.0002; ***p < 0.0001 for BCI+Ket vs Veh+Ket). (F) Quantification by FOVs for the cluster number of Bassoon (Ket × BCI interaction **p = 0.0071; ***p < 0.0001 for BCI+Ket vs. Veh+Ket) and Homer1 (Ket × BCI interaction ***p = 0.0009; ***p = 0.0004 for BCI+Ket vs. Veh+Ket). (G) Quantification by hippocampal slices for the cluster number of Bassoon (Ket × BCI interaction *p = 0.0136; ***p = 0.0001 for BCI+Ket vs. Veh+Ket) and Homer1 (Ket × BCI interaction **p = 0.0023; **p = 0.0022 for BCI+Ket vs. Veh+Ket). For FOVs, n = 69, 66, 60, 66 (GluA1); 68, 64, 60, 65 (GluA2); 66, 69, 39, 71 (Bassoon and Homer1); for slices, n = 21, 23, 17, 19 (GluA1); 22, 23, 18, 22 (GluA2); 22, 22, 14, 19 (Bassoon and Homer1) from 4, 4, 3, 4 mice, for Veh+Sal, Veh+Ket, BCI+Sal, BCI+Ket group, respectively. Data are means ± SEM and analyzed by two-way ANOVA and Tukey post hoc test.

Fig. 3.. BCI prolonged the antidepressant effects of ketamine for up to eight weeks.

(A) Experimental timeline (B) BCI and ketamine co-treatment decreased the feeding latency in NSFT (n = 10–11/group, Ket × BCI interaction **p = 0.001; ***p = 0.0008 for BCI+Ket vs. Veh+Ket). (C) Cumulative feeding latency is measured after treatment (***p = 0.0003 for BCI+Ket vs. Veh+Ket, by Mantel-Cox log-rank test). The black vertical dashed line highlights the latency time when 100% BCI-ket treated mice had eaten food, compared with a significantly lower percentage in the other three groups. (D) Appetite, measured as food consumption/mouse within 5 minutes after NSFT, did change among the groups. (E) BCI and ketamine co-treatment decreases immobility time in FST (n = 10/group, Ket × BCI interaction *p = 0.013; ***p = 0.0003 for BCI+Ket vs. Veh+Ket). (F) Both male (n = 12/group, Ket × BCI interaction *p = 0.042; ***p = 0.0002 for BCI+Ket vs. Veh+Ket) and female (n = 8–13/group, Ket × BCI interaction ***p < 0.0001; ***p = 0.0001 for BCI+Ket vs. Veh+Ket) mice display reduced immobility time in FST after BCI and ketamine co-treatment. (G) The FST was performed on mice that had been treated with drugs eight weeks before. (H) The co-treatment of BCI and ketamine for 8 weeks decreased immobility time in FST (n = 9–11 mice/group; Ketamine × BCI interaction, *p = 0.011; **p = 0.007 for BCI+Ket vs. Veh+Ket). Data are means ± SEM, analyzed by two-way ANOVA and Tukey post hoc test (except C).

Fig. 4.. BCI extended ketamine’s antidepressant effects in the chronic stress model.

(A) After drinking water or corticosterone (CORT) for 21 days to induce stress, acute stress in mice was assessed by SPT (n = 13 for the water group and 49 for the CORT group; ***p < 0.0001 by Mann-Whitney test). Stressed mice were selected to receive one of the drug treatments (vehicle or BCI, plus saline or ketamine). Four weeks post-treatment, mice were assessed with SPT, NSFT, and FST. (B) BCI and ketamine co-treatment shows higher sucrose preference (%) than other treatments in the chronically stressed mice, comparable to unstressed mice drinking water ad libitum (n = 8–10/ test group, n = 7 for unstressed mice; Ket × BCI interaction **p = 0.006; **p = 0.003 for BCI+Ket vs. Veh+Ket). (C) BCI and ketamine co-treatment of stressed mice decreased the feeding latency in NSFT (n = 11–12/group, Ket × BCI interaction *p = 0.015; ***p < 0.0001 for BCI+Ket vs. Veh+Ket). (D) Cumulative feeding latency is shown (***p < 0.0001 for BCI+Ket vs. Veh+Ket, by Mantel-Cox log-rank test). The black dashed line highlights the latency time when 100% BCI-Ket treated mice had eaten food, compared with a significantly lower percentage of mice in the other three groups. (E) The appetite of stressed mice, measured by food consumption within 5 minutes after NSFT, did not change among all groups. (F) The immobility time in FST decreased after four weeks of BCI and ketamine co-treatment in stressed mice (n = 11–12/group, Ket × BCI interaction ***p = 0.0001; ***p < 0.0001 for BCI+Ket vs. Veh+Ket). Data are means ± SEM; all data, except (D), are analyzed by two-way ANOVA and Tukey post hoc test.

Fig. 5.. TrkB in CA1 neurons is required for BCI-mediated enhancement of ketamine’s synaptic potentiation and antidepressant effects.

(A) CamKII-cre, TrkB flox/flox (TrkB-cKO) or control mice were i.p. injected with Veh/BCI and Sal/Ket, 1 hour apart. 24 hours after treatments, immunoblotting (WB) and field recording (Ephys.) were performed. Mice were assessed by FST at 4 weeks post-treatment. BCI and ketamine co-treatment increased p-ERK1/2 levels in CA1 of control mice, but not in TrkB-cKO mice (n = 5/group; treatment × genotype interaction is **p = 0.002; ***p = 0.0003 for BCI+Ket in cKO vs. control and **p = 0.0015 for BCI+Ket vs. BCI+Sal in control mice). (B) fEPSP were recorded at baseline (black lines) and after ketamine infusion in hippocampal slices of mice of four groups. The synaptic potentiation (%) was assessed by measuring the fEPSP slope at baseline (1) and after 30 minutes of ketamine (Ket, 20 μM) infusion (2) in hippocampal slices. Synaptic potentiation is blocked in slices of TrkB-cKO (n = 7 slices from at least 4 mice; Ket × BCI interaction ***p < 0.0001; ***p < 0.0001 for BCI+Ket vs. BCI+Sal in control mice and ***p < 0.0001 for cKO vs. control). (C) The I/O curve of fEPSP in CA1 was measured at baseline; the slope of I/O curve increased only after ketamine and BCI co-treatment in control animals, not TrkB-cKO (n = 7 slices from at least 4 mice; treatment × genotype interaction at 0.10–0.15 mV is *p = 0.0340, *p = 0.0144 for BCI+Ket in cKO vs. control and BCI+Ket vs BCI+Sal in control mice: **p = 0.0031; treatment × genotype interaction at 0.15–0.20 mV is ***p = 0.0008, *p = 0.0145 for BCI+Ket in cKO vs control, and BCI+Ket vs. BCI+Sal in control mice: **p = 0.0071). Field waveforms from slices recorded at 0.10–0.15 mV presynaptic volley are shown. (D) The cell surface GluA1 and GluA2 were assessed by biotinylation and immunoblotting. Co-treatment of BCI and ketamine elevated surface GluA1 (F) and GluA2 (G) protein levels in CA1 of control mice, but not in TrkB-cKO mice (n = 4–6 mice/group; GluA1, treatment × genotype interaction ***p = 0.0009, ***p = 0.0001 for BCI+Ket vs. BCI+Sal in control mice and ***p < 0.0001 for BCI+Ket in cKO vs. control mice; GluA2, treatment × genotype interaction **p = 0.0055, **p = 0.0033 for BCI+Ket vs. BCI+Sal in control mice and ***p = 0.0002 for BCI+Ket in cKO vs control mice). (E) Control but not TrkB-cKO mice displayed antidepressant response with decreased immobility time in FST only after BCI and ketamine co-treatment for 4 weeks (n = 11 mice/group, treatment × genotype interaction **p = 0.0021; ***p = 0.0004 for BCI+Ket vs. Veh+Ket in control mice and ***p = 0.0002 for BCI+Ket in cKO vs. control mice). (F) Stereotaxic injection of AAV expressing GFP-Cre or GFP alone as a control targets the CA1 region of TrkB flox/flox mice. After three weeks, mice were i.p. treated with Veh/BCI combined with Sal/Ket, followed by behavioral tests four weeks post-treatment. (G) Mice with TrkB deletion in CA1blocked the prolonged antidepressant responses in FST after BCI and ketamine co-treatment for 4 weeks (n = 8–9 mice/group, treatment × genotype interaction **p = 0.0032; **p = 0.0010 for BCI+Ket vs. BCI+Sal in AAV-GFP control mice and **p = 0.0011 for BCI+Ket in AAV-GFP-Cre vs. AAV-GFP mice). Data are means ± SEM and are analyzed by two-way ANOVA and Tukey post hoc test.