Antidepressant Effects of (S)-Ketamine through a Reduction of Hyperpolarization-Activated Current Ih.

Compelling evidence suggests that a single sub-anesthetic dose of (R,S)-ketamine exerts rapid and robust antidepressant effects. However, the cellular mechanisms underlying the antidepressant effects of (R,S)-ketamine remain unclear. Here, we show that (S)-ketamine reduced dendritic but not somatic hyperpolarization-activated current Ih of dorsal CA1 neurons in unstressed rats, whereas (S)-ketamine decreased both somatic and dendritic Ih in chronic unpredictable stress (CUS) rats. The reduction of Ih by (S)-ketamine was independent of NMDA receptors, barium-sensitive conductances, and cAMP-dependent signaling pathways in both unstressed and CUS groups. (S)-ketamine pretreatment before the onset of depression prevented CUS-induced behavioral phenotypes and neuropathological changes of dorsal CA1 neurons. Finally, in vivo infusion of thapsigargin-induced anxiogenic- and anhedonic-like behaviors and upregulation of functional Ih, but these were reversed by (S)-ketamine. Our results suggest that (S)-ketamine reduces or prevents Ih from being increased following CUS, which contributes to the rapid antidepressant effects and resiliency to CUS.

Introduction

Depression is a life-threatening mental illness with a prevalence rate of about 4.7% worldwide (Ferrari et al., 2013, Bostwick and Pankratz, 2000). Current monoaminergic antidepressants show limited effects such as delayed onset and partial efficacy, leading to an increase in risk of suicidal behavior. Unlike monoaminergic antidepressants, a single sub-anesthetic dose of (R,S)-ketamine, an NMDA receptor antagonist, has been shown to have rapid and sustained antidepressant effects in treatment-resistant depression (TRD) (Berman et al., 2000, Zarate et al., 2006a). However, the mechanism underlying the antidepressant effect of (R,S)-ketamine remains unclear. Although there has been controversy over (R,S)-ketamine's NMDA receptor-dependent or NMDA receptor-independent antidepressant effect (Preskorn et al., 2008, Lodge and Mercier, 2015, Zarate et al., 2006b, Zanos et al., 2016, Suzuki et al., 2017), the antidepressant effect is, in general, associated with (1) increases in brain-derived neurotrophic factor (BDNF)-mammalian target of rapamycin (mTOR) signaling pathway, (2) a decrease in eukaryotic elongation factor 2 (eEF2) kinase, and (3) increases in synaptogenesis, spine density, and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor expression (Li et al., 2010, Autry et al., 2011, Maeng et al., 2008).

Hyperpolarization-activated cyclic nucleotide-gated nonselective cation (HCN) channels are highly expressed in the hippocampus, cortex, and cerebellum (Monteggia et al., 2000). HCN1 is the main isoform of HCN channels (HCN1–HCN4) and is highly expressed in the hippocampal CA1 region with a gradient of increasing channel expression along the somatodendritic axis (Lorincz et al., 2002, Magee, 1998). HCN channels are active at the resting membrane potential, which contributes to the intrinsic membrane properties of neurons (e.g., resting membrane potential, input resistance [Rin], resonance frequency [fR], neuronal excitability, and synaptic integration; Narayanan and Johnston, 2007, Kim et al., 2012, Magee, 1999). We previously showed that knockdown of the HCN1 subunit in the dorsal CA1 region of the hippocampus, which corresponds to the posterior hippocampus in humans, produces anxiolytic- and antidepressant-like behaviors in unstressed rats (Kim et al., 2012). These changes in behaviors are associated with an increase in neuronal excitability, an enhancement of BDNF-mTOR signaling, an increase in synaptic excitation, and an increase in dorsal hippocampal activity (Kim et al., 2012). We also recently reported that HCN1 protein expression and Ih are increased in the perisomatic region of dorsal hippocampal CA1 following chronic, but not acute stress (Kim et al., 2018). A reduction of HCN1 protein expression in the dorsal CA1 region before the onset of CUS produces resiliency to CUS (Kim et al., 2018). Furthermore, in vivo block of the sarcoplasmic/endoplasmic reticulum Ca2+-ATPase (SERCA) pumps in dorsal CA1 region produces anxiogenic-like behaviors in the open field test and upregulation of functional Ih, similar to that observed in rats following CUS (Kim et al., 2018).

It has been reported that (S)-ketamine, the S enantiomer of (R,S)-ketamine, inhibits HCN1-mediated Ih via a hyperpolarization shift in voltage dependence of activation and a reduction of total Ih conductance in layer 5 cortical pyramidal neurons in unstressed conditions (Chen et al., 2009). Moreover, the US Food and Drug Administration (FDA)-approved intranasal (S)-ketamine shows a rapid antidepressant effect among patients with TRD and decreases suicide ideation in major depression (Canuso et al., 2018). Given that knockdown of HCN1 and treatment with (R,S)-ketamine activate similar downstream signaling pathways (e.g., increased BDNF-mTOR signaling) and produce similar behavioral outcomes (e.g., anxiolytic- and antidepressant-like behaviors) (Kim et al., 2012, Autry et al., 2011, Li et al., 2010), we sought to investigate whether (1) (S)-ketamine changes Ih in the dorsal CA1 neurons and (2) (S)-ketamine pretreatment before the onset of depression exerts resiliency to CUS.

In this study, we found that S-ketamine reduced dendritic but not somatic Ih in normal or unstressed conditions. Using the CUS model of depression, however, we found that (S)-ketamine reduced not only dendritic but also the CUS-induced upregulation of somatic Ih (Kim et al., 2018). The (S)-ketamine-induced reduction in Ih was due to a decrease in maximal h current and a hyperpolarizing shift to the h channel activation curve. The observed (S)-ketamine-induced reduction in Ih-sensitive electrophysiological measurements (i.e., higher input resistance, Rin, and lower resonance frequency, fR, see Methods) was independent of NMDA receptors, barium-sensitive conductances (e.g., inwardly rectifying K+ channels), and cAMP-dependent signaling pathways in both unstressed and CUS groups. We show that (S)-ketamine pretreatment before the onset of CUS prevented the CUS-induced abnormal behaviors (i.e., anxiogenic- and depressive-like behaviors) and neuropathological changes (i.e., upregulation of somatic Ih and a decrease in neuronal excitability). Finally, in vivo infusion of thapsigargin (TG) followed by infusion of (S)-ketamine showed that TG-induced changes in behaviors (e.g., anxiogenic- and anhedonic-like behaviors) and upregulation of functional Ih (e.g., decreased Rin and increased fR) were reversed by (S)-ketamine.

Results

Changes in Dendritic but Not Somatic Ih following (S)-Ketamine Application in Unstressed Conditions

A single sub-anesthetic dose of (R,S)-ketamine (0.5 mg/kg intravenous [i.v.] infusion for 40 min) exerts rapid and sustained antidepressant effects in major depressive disorder (Berman et al., 2000). Given the similar downstream signaling pathways (e.g., increased BDNF-mTOR signaling) and behavioral outputs (e.g., anxiolytic- and antidepressant-like behaviors) between a reduction of HCN1 protein expression (Kim et al., 2012) and (R,S)-ketamine treatment (Autry et al., 2011, Li et al., 2010), our goal was to investigate whether (S)-ketamine had effects on Ih in dorsal CA1 neurons. We focused on dorsal neurons because (1) knockdown of HCN1 in the ventral CA1 region has no anxiolytic- or antidepressant-like effects and (2) there are no changes in Ih-sensitive electrophysiological measurements (e.g., Rin and fR) of ventral CA1 neurons following CUS (Kim et al., 2018). It is known that (S)-ketamine inhibits HCN1 channels, resulting in a hyperpolarization shift in voltage dependence of activation and a reduction of total Ih conductance in layer 5 cortical pyramidal neurons in unstressed conditions and human embryonic kidney (HEK) 293 cells expressing homomeric mouse HCN1 channels (Chen et al., 2009). A key feature of HCN1 channels is a distance-dependent increase in protein expression (Lorincz et al., 2002) (Figure S1A) and Ih-sensitive measurements (i.e., lower Rin and higher fR, Figure S1B) along the somatodendritic axis of CA1 pyramidal neurons. ZD7288, an HCN channel blocker, reduced Ih-sensitive measurements (i.e., it increased Rin and removed fR, Figure S1C). We performed whole-cell current-clamp recordings at the soma (Figures 1A–1E) and dendrite (Figures 1F–1J), where HCN channels are heavily expressed, and all experiments were done in the presence of glutamatergic synaptic blockers (D-AP5 and DNQX). Somatic Rin (Figures 1B–1D) and fR (Figures 1B, 1C, and 1E) were not significantly altered following bath application of (S)-ketamine (Figure S2). On the other hand, (S)-ketamine significantly increased dendritic Rin and decreased dendritic fR consistent with a reduction of Ih (Figures 1G–1J). In keeping with the known somatodendritic gradient of Ih (Magee, 1998), there was also a correlation between a change (%) in Ih-sensitive measurements (i.e., Rin and fR) and the distance from the soma to dendrite (Figures 1K and 1L). We also examined whether the effects of (S)-ketamine on dendritic Ih-sensitive measurements were dose dependent in dorsal CA1 neurons. We used four different concentrations of (S)-ketamine (1, 10, 50, and 100 μM) and measured changes in dendritic Rin and fR at −60 mV. Changes in Ih-sensitive measurements at −60 mV were significantly affected by bath application of either 50 or 100 μM of (S)-ketamine compared with baseline (Figure S3). There was no significant difference between 50 and 100 μM (S)-ketamine (Figure S3). We, therefore, used 50 μM for all our experiments.

Figure 1

Changes in Dendritic but Not Somatic Ih-Sensitive Electrophysiological Measurements within 10–20 Min Following (S)-Ketamine Application

(A and F) Schematic of the somato-apical trunk depicting the somatic (A) and the dendritic (F) recordings.

(B and G) Representative voltage traces from -65 mV in response to a sinusoidal current injection of constant amplitude and linearly spanning 0-15 Hz in 15 sec.

(C and H) The profile of impedance amplitude for voltage traces in (B and G). Vertical lines indicate the resonance frequencies. Somatic fR was not significanlty altered following (S)-ketamine application (C), whereas (S)-ketamine significantly reduced dendritic fR (H).

(D and E) Somatic (D) Rin and (E) fR at −65 mV were not altered following (S)-ketamine application.

(I and J) Dendritic (I) Rin and (J) fR at −65 mV were significantly changed following (S)-ketamine application. ∗∗∗p < 0.001 by Wilcoxon matched-pairs signed rank test.

(K and L) Changes in (K) Rin (%) and (L) fR (%) at −60 mV were distance dependent along the somatodendritic axis of dorsal CA1 neurons following (S)-ketamine application. p < 0.0001 by linear regression analysis. Data are expressed as mean ± SEM.

The (S)-Ketamine Effect on Dendritic Ih Was Independent of Barium-Sensitive Conductances and cAMP-Dependent Signaling

Given the voltage dependence of HCN channels, we measured dendritic Rin and fR at different membrane potentials (ranging from −60 mV to −75 mV, −5 mV interval) before and after bath application of (S)-ketamine. Vm, Rin, and fR were monitored during (S)-ketamine wash-in in dorsal CA1 neurons (Figure 2A). We consistently observed reduced dendritic Ih-sensitive measurements (i.e., increased Rin and decreased fR) within 10–20 min (Figure 2A). Dorsal CA1 neurons showed increased dendritic Rin and decreased dendritic fR at depolarized membrane potentials (Figures 2B–2D) following bath application of (S)-ketamine. We have previously found that a G protein-coupled inwardly rectifying potassium (GIRK) conductance, in part, contributes to intrinsic membrane properties (e.g., Rin and resting membrane potential) of dorsal CA1 neurons (Kim and Johnston, 2015). We, therefore, examined whether the increased Rin following (S)-ketamine application was due to a change in GIRK conductance. Low concentrations of Ba2+ (25–50 μM) block inwardly rectifying K+ channels such as GIRK and IRK (Kim and Johnston, 2015, Malik and Johnston, 2017). Dendritic Ih-sensitive measurements were significantly reduced (i.e., increased Rin and decreased fR) at depolarized membrane potentials in the presence of Ba2+ (25 μM) following bath application of (S)-ketamine (Figure 2E), suggesting that its effects were independent of a barium-sensitive conductance (e.g., IRK and GIRK conductances). It has been reported that acute (R,S)-ketamine treatment (10 μM for 15 min) on C6 glioma cells or primary astrocytes increased translocation of Gαs from lipid rafts to nonrafts after 24 h, suggesting cAMP-dependent antidepressant action of (R,S)-ketamine (Wray et al., 2018). Because HCN channels have a cyclic nucleotide-binding domain (Wang et al., 2001), we further explored whether the (S)-ketamine-induced reduction of dendritic Ih-sensitive measurements was dependent on cAMP-dependent signaling. 8-bromo-cyclic AMP (cAMP) is a membrane-permeable activator of cAMP-dependent protein kinase. Because we did not observe any changes in intrinsic membrane properties (e.g., Vm, Rin, and fR) of dorsal CA1 neurons following bath application of 8-bromo-cAMP (100 μM) (Figure S4), as a control for 8-bromo-cAMP, we examined whether the applied 8-bromo-cAMP was having the known physiological effects on active properties of dorsal CA1 neurons. It has been reported that back-propagating action potential (bAP) amplitude, which decreases with distance from the soma due to an increased density of A-type K+ channels, is increased by protein kinase A activation (Hoffman and Johnston, 1998). We, therefore, measured bAPs elicited by antidromic extracellular stimulation in the stratum oriens at the soma, proximal dendrites (∼220 μm from the soma), and distal dendrites (∼320 μm from the soma) (Figures S5A and S5B). Consistent with the report by Hoffman and Johnston (1998) we observed increased bAP amplitude at −65 mV in the distal dendrites of dorsal CA1 neurons following 8-bromo-cAMP, indicating 8-bromo-cAMP-induced physiological effects on dorsal CA1 neurons (Figures S5A and S5B). Bath application of (S)-ketamine led to increased dendritic Rin and decreased dendritic fR at depolarized membrane potentials in the presence of 8-bromo-cAMP (Figures 2F, S5C, and S5D), suggesting that the (S)-ketamine-induced decrease in Ih-sensitive measurements was independent of the cAMP-dependent signaling. Last, we asked whether block of Ih occluded the (S)-ketamine-dependent reduction of dendritic Ih-sensitive measurements in the dorsal CA1 neurons. We performed successive ZD7288 and (S)-ketamine wash-in experiments and measured dendritic Rin and fR at resting membrane potential (RMP) and at different membrane potentials. Vm, Rin, and fR at RMP were monitored during bath application of ZD7288 (10 μM, 4–5 min) and (S)-ketamine (50 μM) (Figure S6A). Bath application of ZD7288 significantly increased Rin and abolished fR at RMP (Figures S6B, S6C, S6E, and S6F) and at different membrane potentials (Figure 2G). Subsequent addition of (S)-ketamine had no further effect on Vm, Rin, and fR at RMP (Figures S6B–S6F) and at different membrane potentials (Figure 2G) in the dorsal CA1 neurons, suggesting that ZD7288 occluded the effect of (S)-ketamine. These results suggested that (S)-ketamine reduced dendritic but not somatic Ih-sensitive measurements independent of NMDA receptors, barium-sensitive conductances, and cAMP-dependent signaling.

Figure 2

Changes in Dendritic Ih-Sensitive Measures by (S)-Ketamine Were Independent of a Barium-Sensitive Conductance and cAMP-Dependent Signaling

(A) Time courses of changes in Vm, Rin, and fR during (S)-ketamine wash-in experiment in the dorsal CA1 neurons (green: Vm, orange: fR, and black: Rin).

(B) Representative voltage traces and current injections at different membrane potentials (ranging from −60 mV to −75 mV; −5 mV interval).

(C) The profile of impedance amplitude for voltage traces in (B).

(D) Dendritic Rin and fR were significantly changed at depolarized membrane potentials following (S)-ketamine application. ∗p < 0.05, ∗∗p < 0.01 by two-way ANOVA with Sidark's multiple comparisons test.

(E) Dendritic Rin and fR were significantly changed at depolarized membrane potentials in the presence of barium chloride (25 μM) following (S)-ketamine application. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 by two-way ANOVA with Sidark's multiple comparisons test.

(F) Bath application of (S)-ketamine increased dendritic Rin and decreased dendritic fR at depolarized membrane potentials in the presence of 8-bromo-cAMP (100 μM). ∗∗p < 0.01, ∗∗∗p < 0.001 by two-way ANOVA with Sidark's multiple comparisons test.

(G) The effects of (S)-ketamine were occluded by ZD7288 (10 μM) in the dorsal CA1 neurons. Subsequent bath application of (S)-ketamine (50 μM) had no further effects on dendritic Rin and fR. ∗∗∗p < 0.001 (baseline versus ZD7288) and ###p < 0.001 (baseline versus ZD7288+(S)-ket) by two-way ANOVA with Sidark's multiple comparisons test. Data are expressed as mean ± SEM.

The Effect of (S)-Ketamine on Voltage-Dependent Gating of h Channel and the Amplitude of h Current

We have thus far demonstrated that (S)-ketamine decreased dendritic but not somatic Ih-sensitive electrophysiological measurements at depolarized membrane potentials (Figure 2). To measure directly the effect of (S)-ketamine on Ih, we performed cell-attached patch-clamp recordings from the soma and dendrites of dorsal CA1 neurons in the presence of D-AP5 and DNQX with or without the prior treatment for 30 min of (S)-ketamine (50 μM). We determined voltage-dependent gating of h channel and the amplitude of h current with 500-ms hyperpolarizing voltage steps ranging from −40 to −170 mV in −10 mV increments from a holding potential of −30 mV (Figures 3A and 3H). Given that somatic Ih was too small to determine the conductance-voltage (G-V) relationship from the peak of the tail current, we constructed a steady-state G-V relationship from the steady-state h current in response to each voltage step (Figure 3B). With the solutions used (see Methods) the reversal potential of Ih was experimentally determined from dendritic recordings (Eh = −10 mV; data not shown). At the soma, we found that the V1/2 of the h channel activation curve (Figures 3B and 3C), the slope factor (Figures 3B and 3D), and the maximal Ih at −170 mV (Figures 3E and 3F) were not significantly different between artificial cerebrospinal fluid (ACSF)-treated and (S)-ketamine-treated groups. Steady-state Ih amplitude at different membrane potentials (i.e., I-V curve) was also not different between ACSF-treated and (S)-ketamine-treated groups (Figure 3G), consistent with no measurable effects of (S)-ketamine on current-clamp somatic Ih-sensitive measurements (Figure 1). Dendrites had much larger Ih currents, allowing us to determine the voltage dependence of activation from the normalized peak tail current (Itail/Itail max). At ∼270 μm from the soma, we observed that the V1/2 of the h channel activation curve with (S)-ketamine treatment was significantly shifted to the left by around −12 mV (Figures 3I and 3J) (V1/2 of ACSF: −108.2 mV versus V1/2 of (S)-ket: −120.5 mV), whereas the slope factor was not different (Figures 3I and 3K) compared with those from ACSF-treated group. Furthermore, maximal dendritic Ih at −170 mV was significantly decreased in the (S)-ketamine-treated group compared with those from the ACSF-treated group (Figures 3L and 3M) (Max Ih of ACSF: −7.36 pA versus Max Ih of (S)-ket: −5.09 pA). In addition, steady-state Ih amplitude at different membrane potentials was significantly decreased in the (S)-ketamine-treated group compared with those from the ACSF-treated group (Figure 3N). Thus, both current-clamp measurements of Ih-sensitive properties (Figures 1 and 2) and direct recordings of Ih (Figure 3) indicate that the effects of (S)-ketamine occur in the dendrite but not in the soma in unstressed conditions. We then asked whether (S)-ketamine reduced Ih of dorsal CA1 neurons from the CUS model of depression.

Figure 3

(S)-Ketamine Changed Voltage-Dependent Gating of h Channel and the Amplitude of h Current in the Dendrites of Dorsal CA1 Neurons

(A and H) Somatic (A) or dendritic (H; ∼270 μm from the soma) h current was measured using cell-attached patches from dorsal CA1 neurons after ACSF-treated or (S)-ketamine-treated (50 μM, 30 min) slices with 500-ms hyperpolarizing voltage steps ranging from −40 mV to −170 mV in −10 mV increments from a holding potential of −30 mV. Insets in (H): tail currents are enlarged from the dashed box. Vertical dashed lines indicate the location for determining the peak tail current.

(B) The voltage dependence of activation for h channel was determined from the somatic steady-state G-V relationship. The activation curve was fitted a Boltzmann function with the following values: ACSF treatment V1/2 = −116.8 mV, k = −10.07 mV; (S)-ketamine treatment V1/2 = −119.2 mV, k = −9.6 mV.

(C and D) The half-activation voltage (V1/2; C) and slope factor (D) were not significantly different between ACSF-treated and (S)-ketamine-treated groups.

(E) Representative maximal h current traces in response to a 500-ms hyperpolarizing voltage step (−170 mV).

(F) Maximal somatic h current at −170 mV was not different between ACSF-treated and (S)-ketamine-treated groups.

(G) There was no difference in the I-V curve between ACSF-treated and (S)-ketamine-treated groups.

(I) The voltage dependence of activation for h channel was determined from tail currents. The activation curve was fitted with a Boltzmann function with the following values: ACSF treatment V1/2 = −108.2 mV, k = −15 mV; (S)-ketamine pretreatment V1/2 = −120.5 mV, k = −13.4 mV.

(J and K) (J) The half-activation voltage of h channel (V1/2) with (S)-ketamine treatment was significantly shifted to the left by around −12 mV, whereas the slope factor (K) was not different compared with those from ACSF-treated group. ∗p < 0.05 by Mann-Whitney test.

(L) Representative maximal h current traces in response to a 500-ms hyperpolarizing voltage step (−170 mV).

(M) Maximal dendritic h current at −170 mV was significantly decreased in (S)-ketamine-treated group compared with those from ACSF-treated group. ∗∗p < 0.01 by Mann-Whitney test.

(N) Steady-state Ih amplitude at different membrane potentials was significantly decreased in (S)-ketamine-treated group compared with those from ACSF-treated group. ∗p < 0.05 by two-way ANOVA with Sidark's multiple comparisons test. Traces in (A) and (E) were digitally filtered at 1 kHz for clarity. Data are expressed as mean ± SEM.

(S)-Ketamine Reduced Somatic and Dendritic Ih-Related Properties in the CUS Model of Depression

Rats exposed to CUS for 2–3 weeks (Figure 4A) showed decreased locomotor activity (Figures 4B and 4C) and center square entries (Figures 4B and 4D) in the open field test and decreased sucrose preference (Figure 4E) in a two-bottle choice test, consistent with previous work (Kim et al., 2018). We then examined whether (S)-ketamine exerted rapid and sustained antidepressant effects in the CUS model of depression. Unstressed and CUS rats were split into four groups: (1) unstressed rats injected with saline 1 h before forced swim test (FST), (2) unstressed rats injected with (S)-ketamine (15 mg/kg intraperitoneal [i.p.] injection) 1 h before FST, (3) CUS rats injected with saline 24 h before FST, and (4) CUS rats injected with (S)-ketamine (15 mg/kg i.p. injection) 24 h before FST. CUS-exposed rats displayed an increase in passive activity (i.e., behavioral despair) that was reversed 1 h (Figures 4F) and 24 h (Figure 4G) after (S)-ketamine treatment in the FST, indicating the rapid and sustained antidepressant effects of (S)-ketamine. After the behavioral tests, acute dorsal hippocampal slices from saline-treated unstressed or CUS rats were prepared. Consistent with previous work (Kim et al., 2018), dorsal CA1 neurons from CUS-treated rats showed decreased somatic Rin and increased somatic fR (Figure S7A) at different membrane potentials compared with those from the unstressed rats. Given that chronic stress has been shown to produce hypoactive neuronal excitability of dorsal CA1 neurons (Kim et al., 2018), we tested whether there was a change in neuronal excitability in the CUS model of depression. The number of action potential (APs) at the RMP in response to depolarizing current steps (30–400 pA for 750 ms) was significantly decreased in dorsal CA1 neurons from CUS-treated group compared with those from the unstressed group (Figure S7B).

Figure 4

Rapid and Sustained Antidepressant-like Effects of (S)-Ketamine in the CUS Model of Depression

(A) Timeline of CUS, behavioral tests, and electrophysiology.

(B–D) (B) Representative video tracking images during the last 5 min of open field test of age-matched individual rats—unstressed versus CUS groups. CUS-treated rats showed decreases in total distance (C) and center square entries (D) compared with those from unstressed group. ∗∗p < 0.01 and ∗∗∗p < 0.001 by Mann-Whitney test.

(E) CUS-treated rats showed decreased sucrose preference. ∗∗∗p < 0.001 by Mann-Whitney test.

(F) Passive activity was significantly decreased 1 h after (S)-ketamine treatment (15 mg/kg, i.p. injection) in either unstressed or CUS groups (i.e., acute effects). ∗p < 0.05 and ∗∗∗p < 0.001 by one-way ANOVA with Tukey multiple comparisons test.

(G) Passive activity was significantly decreased 24 h after (S)-ketamine treatment in either unstressed or CUS groups (i.e., sustained effects). ∗∗∗p < 0.001 by one-way ANOVA with Tukey multiple comparisons test. Data are expressed as mean ± SEM.

We previously reported that the expression of the HCN1 subunit of HCN channels and Ih were upregulated in the soma, but not in the dendrite of dorsal CA1 neurons following CUS (Kim et al., 2018). Given that (S)-ketamine reduced dendritic but not somatic recordings consistent with the influence of Ih on dorsal CA1 neurons in unstressed conditions (Figure 1), we next examined whether (S)-ketamine reduced somatic and/or dendritic Ih-sensitive properties in the CUS model of depression. Rin and fR at different membrane potentials were determined using whole-cell current-clamp recordings at the soma (Figures 5A–5C) and dendrite (Figures 5D–5F). Consistent with the results shown in Figure 1, (S)-ketamine did not affect somatic Ih-sensitive measurements in unstressed rats (Figure 5B). However, (S)-ketamine reduced the CUS-induced upregulation of somatic Ih-sensitive measurements (i.e., increased Rin and decreased fR) at depolarized membrane potentials (Figure 5C). Dendritic Ih-sensitive measurements were significantly reduced at depolarized membrane potentials in both unstressed (Figure 5E) and CUS (Figure 5F) groups. Consistent with the effect of (S)-ketamine on somatic Rin, the number of APs elicited by depolarizing current steps was increased by (S)-ketamine in post-CUS but not unstressed CA1 neurons (Figures 5G–5J).

Figure 5

(S)-Ketamine Altered Somatic and Dendritic Ih-Sensitive Measures in the Dorsal CA1 Neurons of CUS Group

(A and D) Schematic of the somato-apical trunk depicting the somatic (A) and the dendritic (D) recordings.

(B) Somatic Rin and fR at different membrane potentials were not changed following (S)-ketamine application in unstressed group.

(C) (S)-ketamine increased somatic Rin and decreased fR at depolarized membrane potentials in CUS group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 by two-way ANOVA with Sidark's multiple comparisons test.

(E and F) Dendritic Rin and fR at depolarized membrane potentials were significantly changed following (S)-ketamine application in unstressed (E) and CUS (F) groups. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 by two-way ANOVA with Sidark's multiple comparisons test.

(G and I) Representative voltage responses at RMP with somatically injecting (270 pA; 750 ms) depolarizing current before (black) and after (red) (S)-ketamine treatment.

(H) The number of action potentials was not changed following (S)-ketamine application in dorsal CA1 neurons of unstressed group.

(J) A CUS-induced reduction of neuronal excitability was significantly increased following (S)-ketamine application in the dorsal CA1 neurons. ∗∗p < 0.01 by two-way ANOVA with Sidark's multiple comparisons test.

(K and M) In unstressed group, somatic Rin and fR were not significantly altered at different membrane potentials in the presence of barium chloride (K, 25 μM) or 8-bromo-cAMP (M, 100 μM) following (S)-ketamine application.

(L and N) In CUS group, CUS-induced upregulation of somatic Ih-sensitive measurements were significantly changed at depolarized membrane potentials in the presence of barium chloride (L, 25 μM) or 8-bromo-cAMP (N, 100 μM) following (S)-ketamine application. ∗p < 0.05, ∗∗p < 0.01 by two-way ANOVA with Sidark's multiple comparisons test. Data are expressed as mean ± SEM.

Because the (S)-ketamine-induced reduction in dendritic Ih-sensitive properties was independent of barium-sensitive conductances and cAMP-dependent signaling in unstressed conditions (Figure 2), we next examined whether the same was true for the somatic effects observed following CUS. Vm, Rin, and fR were monitored during baseline, application of 25 μM Ba2+, and addition of 50 μM (S)-ketamine application (with 25 μM Ba2+) in dorsal CA1 neurons (Figure S8). 25 μM Ba2+ significantly depolarized the membrane potential (Figure S9A) and increased Rin (Figure S9B), with no change in fR (Figure S9C) in dorsal CA1 neurons of both unstressed and CUS groups. Somatic Ih-sensitive measurements (i.e., Rin and fR) were not altered at different membrane potentials in the presence of Ba2+ following (S)-ketamine application in the dorsal CA1 neurons of unstressed group (Figures 5K, S9D, and S9E). On the other hand, the CUS-induced upregulation of somatic Ih-sensitive measurements was significantly changed at depolarized membrane potentials in the presence of Ba2+ following (S)-ketamine application in the dorsal CA1 neurons of CUS group (Figures 5L, S9F, and S9G). We next examined whether the effect of (S)-ketamine on CUS-induced somatic Ih was affected by 8-bromo-cAMP treatment (100 μM) in the dorsal CA1 neurons of CUS group. Vm, Rin, and fR were monitored during baseline (100 μM 8-bromo-cAMP) and addition of 50 μM (S)-ketamine application (with 100 μM 8-bromo-cAMP) in the dorsal CA1 neurons (Figure S10). In the unstressed group, somatic Ih-sensitive measurements were not changed at different membrane potentials in the presence of 8-bromo-cAMP following (S)-ketamine application (Figures 5M, S11A, and S11B). However, CUS-induced somatic Ih-sensitive measurements were significantly changed at depolarized membrane potentials in the presence of 8-bromo-cAMP following (S)-ketamine application (Figures 5N, S11C, and S11D). These results suggest that the effects of (S)-ketamine on the CUS-induced upregulation of somatic Ih were independent of barium-sensitive conductances and cAMP-dependent signaling.

Direct Measurements Showed that (S)-Ketamine Reduced Somatic and Dendritic Ih in the CUS Model of Depression

We next examined the somatic effect of (S)-ketamine on voltage-dependent gating of h channel and the amplitude of h current in the CUS model of depression. We performed cell-attached patch-clamp recordings from the CUS-soma of dorsal CA1 neurons in the presence of D-AP5 and DNQX with or without (S)-ketamine treatment (50 μM, 30 min). Consistent with the results showing decreased somatic Rin, increased somatic fR, and decreased in the number of APs in the dorsal CA1 neurons of the CUS group, we found that the V1/2 of the h channel activation curve for CUS group was significantly shifted to the right by around +29 mV (V1/2 of unstress-ACSF: −116.8 mV versus V1/2 of CUS-ACSF: −88 mV), whereas the slope factor was not different compared with those from the unstressed group (Figures 6A–6C). Maximal somatic Ih at −170 mV was significantly increased in the dorsal CA1 neurons of the CUS group compared with those from unstressed group (Figures 6D and 6E) (Max Ih of unstress-ACSF: −0.59 pA versus Max Ih of CUS-ACSF: −3.89 pA). Steady-state Ih amplitude at different membrane potentials was significantly higher for CUS group than those from unstressed group (Figure 6F). Thus, the CUS-induced decrease in Rin and increase in fR was mediated by an increased h current and a depolarizing shift to the h channel activation curve. When CUS-dorsal CA1 neurons were treated with (S)-ketamine (50 μM, 30 min), the V1/2 of the h channel activation curve was significantly shifted to the left by around −23 mV (Figures 7A–7C) (V1/2 of ACSF: −88 mV versus V1/2 of (S)-ket: −110.9 mV), whereas the slope factor was not different (Figures 7A, 7B, and 7D) compared with those from the ACSF-treated CUS group. The CUS-induced increase in maximal somatic Ih was significantly decreased in the dorsal CA1 neurons of the (S)-ketamine-treated CUS group compared with those from the ACSF-treated CUS group (Figures 7E and 7F) (Max Ih of CUS-ACSF: −3.89 pA versus Max Ih of CUS-(S)-ketamine: −1.5 pA). Furthermore, steady-state Ih amplitude at different membrane potentials was significantly decreased in (S)-ketamine-treated group compared with those from ACSF-treated group (Figure 7G). Therefore, (S)-ketamine-induced reduction of somatic Ih was mediated by changes in the activation properties of the channels in the CUS model of depression.

Figure 6

Changes in Voltage-Dependent Gating of h Channel and the Amplitude of h Current of Dorsal CA1 Neurons in the CUS Model of Depression

(A) Somatic h current was measured from unstressed group and CUS group of dorsal CA1 neurons with 500-ms hyperpolarizing voltage steps ranging from −40 mV to −170 mV in −10-mV increments from a holding potential of −30 mV.

(B) The voltage dependence of activation for h channel was determined from the somatic steady-state gV (ss gV) relationship. The activation curve was fitted a Boltzmann function with the following values: unstressed group V1/2 = −116.8 mV, k = −10.07 mV; CUS group V1/2 = −88.07 mV, k = −10.2 mV.

(C) The half-activation voltage (V1/2) was significantly shifted to the right by +29 mV, whereas the slope factor was not different compared with those from unstressed group. ∗p < 0.05 by Mann-Whitney test.

(D) Representative maximum h current traces in response to a 500-ms hyperpolarizing voltage step (−170 mV).

(E) Maximal somatic h current at −170 mV was significantly higher for CUS group than those of unstressed group. ∗p < 0.05 by Mann-Whitney test.

(F) Steady-state Ih amplitude at different membrane potentials was significantly higher for CUS group than those for unstressed group. ∗p < 0.05 by two-way ANOVA with Sidark's multiple comparisons test. Traces in (A) and (D) were digitally filtered at 1 kHz for clarity. Data are expressed as mean ± SEM.

Figure 7

(S)-Ketamine Changed Voltage-Dependent Gating of h Channel and the Amplitude of h Current in the CUS Model of Depression

(A) Somatic h current was measured with cell-attached patches from ACSF-treated or (S)-ketamine-treated (30 min) dorsal CA1 neurons with 500-ms hyperpolarizing voltage steps ranging from −40 mV to −170 mV in −10-mV increments from a holding potential of −30 mV.

(B) The voltage dependence of activation for h channel was determined from the somatic steady-state GV relationship. The activation curve was fitted with a Boltzmann function with the following values: ACSF treatment V1/2 = −88.07 mV, k = −10.2 mV, (S)-ketamine treatment V1/2 = −110.9 mV, k = −9.36 mV.

(C and D) (C) The half-activation voltage of h channel (V1/2) with (S)-ketamine treatment was significantly shifted to the left by around −23 mV, whereas the slope factor (D) was not different compared with those from ACSF-treated group. ∗p < 0.05 by Mann-Whitney test.

(E) Representative maximal h current traces in response to a 500-ms hyperpolarizing voltage step (−170 mV).

(F) Maximal somatic h current at −170 mV was significantly different between ACSF-treated and (S)-ketamine-treated CUS groups.

(G) I-V curve was significantly altered in (S)-ketamine-treated group compared with those from ACSF-treated group. Traces in (A) and (E) were digitally filtered at 1 kHz for clarity. ∗p < 0.05 by two-way ANOVA with Sidark's multiple comparisons test. Data are expressed as mean ± SEM.

(S)-Ketamine Pretreatment Provided Resiliency to CUS

We have previously shown that lentiviral-mediated reduction of HCN1 subunit of HCN channels in the dorsal CA1 region before exposure to CUS provides resiliency to CUS (Kim et al., 2018). Therefore, we next investigated whether (S)-ketamine pretreatment before the onset of the CUS-induced depression prevented the CUS-dependent behavioral phenotypes and neuropathological changes. Unstressed and CUS rats were split into two groups, respectively, (1) saline-pretreated unstressed group, (2) (S)-ketamine-pretreated unstressed group, (3) saline-pretreated CUS group, and (4) (S)-ketamine-pretreated CUS group. Rats were administered a single injection of saline or (S)-ketamine (15 mg/kg, i.p. injection) 7 days before the onset of CUS (Figure 8A). After 2–3 weeks of CUS, we performed a series of behavioral tests such as sucrose preference test, open field test, and FST. In the unstressed group, there were no significant differences in locomotor activity and center square entries between saline- and (S)-ketamine-pretreated groups in the open field test (Figures 8B–8D). Saline-pretreated CUS rats showed decreased locomotor activity and center square entries compared with those from unstressed group (either saline- or (S)-ketamine-pretreated) (Figures 8B–8D), indicating anxiogenic-like behaviors. Interestingly, (S)-ketamine pretreatment in the CUS group prevented CUS-induced decreases in locomotor activity and center square entries compared with those from the saline-pretreated CUS group (Figures 8B–8D), suggesting the prevention of the CUS-induced anxiogenic-like behaviors by (S)-ketamine. Saline-pretreated CUS rats showed decreased sucrose preference (Figure 8E) and increased passive activity (Figure 8F) compared with those from unstressed group (saline- or (S)-ketamine-pretreated) or (S)-ketamine-pretreated CUS group, indicating the CUS-induced anhedonia and behavioral despair. On the other hand, (S)-ketamine pretreatment in the CUS group prevented decreased sucrose preference (Figure 7E) and increased passive activity time (Figure 8F) compared with those from the saline-pretreated CUS group, suggesting the prevention of the CUS-induced anhedonia and behavioral despair by (S)-ketamine.

Figure 8

(S)-Ketamine Pretreatment Provided Resiliency to CUS

(A) Timeline of CUS, behavioral tests, and electrophysiology.

(B) Representative video tracking images during the last 5 min of open field test of age-matched individual rats—unstressed (sal- or (S)-ket-pretreated) versus CUS (sal- or (S)-ket-pretreated) rats.

(C and D) (S)-ketamine-pretreated CUS rats showed increases in total distance (C) and center square entries (D) compared with those from unstressed (sal- or (S)-ket-pretreated) or saline-pretreated CUS group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 by one-way ANOVA with Tukey multiple comparisons test.

(E and F) (S)-ketamine-pretreated CUS rats showed increased sucrose preference (E) and decreased passive activity time (F) compared with those from unstressed (sal- or (S)-ket-pretreated) or saline-pretreated CUS group. ∗p < 0.05, ∗∗∗p < 0.001 by one-way ANOVA with Tukey multiple comparisons test. Data are expressed as mean ± SEM.

Following the behavioral tests, whole-cell current-clamp recordings were made from the soma and dendrite. Somatic Rin and fR at RMP (Figures 9A–9C) and at different membrane potentials (Figure 9D) were not significantly different between saline- and (S)-ketamine-pretreated unstressed groups. Dorsal CA1 neurons from the saline-pretreated CUS group had a lower Rin and a higher fR at RMP (Figures 9B and 9C) and at different membrane potentials (Figures S12A and S12B) compared with those from either the unstressed group (saline- or (S)-ketamine-pretreated) or (S)-ketamine-pretreated CUS group. Interestingly, (S)-ketamine pretreatment in CUS group prevented the CUS-induced upregulation of somatic Ih-sensitive measurements at RMP (Figures 9B and 9C) and at different membrane potentials (Figure 9E). Consistent with a previous work (Kim et al., 2018), there were no significant differences in dendritic Ih-sensitive measurements between groups (Figure S13). These results suggest that (S)-ketamine pretreatment before the onset of depression prevented the CUS-induced upregulation of somatic Ih.

Figure 9

(S)-Ketamine Pretreatment Altered CUS-Induced Upregulation of Somatic Ih-Sensitive Measures

(A) Schematic of the somato-apical trunk depicting the somatic recordings.

(B and C) Dorsal CA1 neurons of (S)-ketamine-pretreated CUS group showed increased Rin (B) and decreased fR (C) at RMP compared with those from saline-pretreated CUS group. ∗p < 0.05, ∗∗∗p < 0.001 by one-way ANOVA with Tukey multiple comparisons test.

(D) There were no different somatic Rin and fR between saline- and (S)-ketamine-pretreated unstressed groups at different membrane potentials (ranging from −60 to −75 mV, interval: −5 mV).

(E) Dorsal CA1 neurons of (S)-ketamine-pretreated CUS group showed increased Rin and decreased fR at different membrane potentials compared with those from saline-pretreated CUS group. ∗p < 0.05 by two-way ANOVA with Sidark's multiple comparisons test. Data are expressed as mean ± SEM.

(S)-Ketamine Pretreatment Prevented the CUS-Induced Neuropathological Changes

Functional neuroimaging of patients with major depression shows bimodal, abnormal regional metabolic activities in limbic-cortical areas, including hippocampus, which can be reversed by clinical antidepressant treatment or chronic electrical stimulation (Mayberg et al., 2000, Mayberg et al., 2005). Similar observations have been reported in chronic stress-induced animal models of depression, which show increased neuronal excitability of lateral amygdala neurons (Rosenkranz et al., 2010) and decreased neuronal excitability of dorsal CA1 neurons (Kim et al., 2018). We, therefore, examined whether (S)-ketamine pretreatment before the onset of the CUS-induced depression prevented the CUS-induced neuropathological changes (i.e., upregulation of somatic Ih and reduced neuronal excitability). There were no significant differences in the number of APs measured in the dorsal CA1 neurons between the saline-pretreated and (S)-ketamine-pretreated unstressed groups (Figures 10A and 10B). However, we consistently observed decreased neuronal excitability in the saline-pretreated CUS group compared with those from the saline-pretreated unstressed group (Figure S14). A CUS-induced decrease in neuronal excitability was normalized in the dorsal CA1 neurons of (S)-ketamine-pretreated CUS group (Figures 10A and 10C). There were no significant differences in AP properties at RMP such as amplitude, peak, half-width, max dv/dt, and threshold (Figure S15). These results suggest that (S)-ketamine pretreatment before the onset of depression is sufficient to prevent the CUS-induced neuropathological changes such as the upregulation of Ih-sensitive measures and decrease in neuronal excitability.

Figure 10

(S)-Ketamine Pretreatment Prevented the CUS-Induced Decrease in Neuronal Excitability

(A) Representative voltage responses with depolarizing current step (270 pA; 750 ms) at RMP.

(B) The number of action potentials was not different between saline- and (S)-ketamine-pretreated unstressed groups.

(C) Dorsal CA1 neurons of (S)-ketamine-pretreated CUS group showed increased action potential firing at RMP. ∗∗p < 0.01, ∗∗∗p < 0.001 by two-way ANOVA with Sidark's multiple comparisons test. Data are expressed as mean ± SEM.

Thapsigargin-Induced Anxiogenic- and Anhedonic-like Behaviors Were Reversed by (S)-Ketamine in the Dorsal CA1 Region

We previously reported that in vivo block of the SERCA pumps in CA1 region leads to anxiogenic-like behaviors in the open field test and upregulation of functional Ih, similar to that observed in rats following CUS (Kim et al., 2018). To test whether (S)-ketamine's antidepressant effects are via the Ih/HCN1 channels, we performed in vivo infusion of TG, an irreversible inhibitor of the SERCA pumps, followed by infusion of (S)-ketamine. The specific doses of D-AP5 (25 μM) and (S)-ketamine (50 μM) used in in vivo study were chosen based on the results from in vitro experiments. To reduce the contribution of NMDA receptor in behavioral tests, TG (1 μM) or (S)-ketamine (50 μM) was dissolved with D-AP5 (25 μM). Rats were bilaterally cannulated into the dorsal CA1 region (Figures 11C) and were housed individually. After a minimum 5-day recovery period, rats were divided into five groups: (1) saline, (2) D-AP5, (3) (S)-ketamine (with D-AP5), (4) TG (with D-AP5), and (5) TG (with D-AP5) followed by (S)-ketamine (with D-AP5) (Figures 11A). It has been reported that 10 μM D-AP5, a competitive NMDA receptor antagonist, is sufficient to block NMDA receptor-mediated currents (Benveniste and Mayer, 1991). Behaviorally, NMDA receptor dysfunction by intrahippocampal infusion of cerebrospinal fluid from anti-NMDA receptor encephalitis shows no changes in locomotor activity or anxiety-related behavior (Kersten et al., 2019). Consistently, rats bilaterally infused with 25 μM D-AP5 showed no changes in locomotor activity or anxiety level (Figures 11B–11E) compared with those from the saline-infused group. Consistent with a previous result (Kim et al., 2018), rats bilaterally infused with 1 μM TG (with D-AP5) showed decreased locomotor activity (Figures 11B and 11D) and center square entries (Figures 11B and 11E) indicating anxiogenic-like behavior compared with those from saline-infused group. Rats bilaterally infused with 50 μM (S)-ketamine (with D-AP5) showed increased center square entries (Figures 11B and 11E) indicating anxiolytic-like behavior compared with those from D-AP5 (25 μM)-infused group. Interestingly, 1 μM TG (with D-AP5) followed by 50 μM (S)-ketamine (with D-AP5) in the dorsal CA1 region increased center square entries compared with those from the TG-infused group, suggesting that TG-induced anxiogenic-like behavior was reversed by (S)-ketamine. Because cannulated rats were not suitable for an FST, we performed a sucrose preference test following the open field test (Figures 11F and 11G). TG-infused rats showed decreased sucrose preference during 13-h dark cycle compared with those from saline- or D-AP5- or (S)-ketamine-infused group (Figures 11G) without a change in total fluid intake (Figures 11F), indicating anhedonic-like behavior. Surprisingly, rats infused with TG followed by (S)-ketamine showed increased sucrose preference compared with those from TG-infused group (Figure 11G), indicating antidepressant-like effect of (S)-ketamine.

Figure 11

Thapsigargin-Induced Anxiogenic- and Anhedonic-like Behaviors Were Reversed by (S)-Ketamine

(A) Timeline of thapsigargin and (S)-ketamine experiment.

(B) Representative video tracking during the 10 min of open field test of age-matched individual rats with saline, D-AP5 (25 μM), (S)-ketamine (50 μM) with D-AP5 (25 μM), thapsigargin (1 μM) with D-AP5 (25 μM), or infusion of thapsigargin (with D-AP5) followed by infusion of (S)-ketamine (with D-AP5).

(C) Representative coronal sections of the brains display the location of the infusion needle track within the dorsal hippocampus of CA1.

(D) Total distance during a 10-min open field test. In vivo infusion of TG decreased locomotor activity. ∗p < 0.05, ∗∗p < 0.01 by one-way ANOVA with Tukey multiple comparisons test.

(E) (S)-ketamine increased center square entries, whereas TG decreased center square entries compared with those from either saline- or D-AP5-infused group. TG-induced anxiogenic-like behavior was reversed by (S)-ketamine. ∗p < 0.05, ∗∗∗p < 0.001 by one-way ANOVA with Tukey multiple comparisons test.

(F) There were no significant differences in total fluid intake between groups.

(G) TG-injected rats showed decreased sucrose preference compared with those from saline- or D-AP5- or (S)-ketamine-infused group. TG-induced anhedonic-like behavior was reversed by (S)-ketamine.

∗p < 0.05, ∗∗p < 0.01 by one-way ANOVA with Tukey multiple comparisons test. Data are expressed as mean ± SEM.

Thapsigargin-Induced Upregulation of Functional Ih Was Reversed by (S)-Ketamine in the Dorsal CA1 Region

Given that in vivo block of the SERCA pumps in dorsal CA1 region/neurons produces an anxiogenic-like behavior and upregulation of functional Ih (Kim et al., 2018), we hypothesized that in vivo infusion of TG followed by infusion of (S)-ketamine might alter functional Ih. Dorsal hippocampal slices were prepared immediately after a 10-min open field test (Figure 12A). Functional Ih (e.g., RMP, Rin, fR, and # of APs) was determined using whole-cell current-clamp recordings at the soma in the presence of glutamatergic synaptic blockers (i.e., D-AP5 and DNQX). There were no differences in somatic RMP between groups (Figures S16A and S16C). Functional Ih at RMP was decreased (i.e., increased Rin and decreased fR) in the dorsal CA1 neurons of the (S)-ketamine-infused group, whereas increased (i.e., decreased Rin and increased fR) in the dorsal CA1 neurons of TG-infused group compared with those from the saline- or D-AP5-infused group (Figures S16A–S16E). TG-induced upregulation of functional Ih at RMP was decreased in the dorsal CA1 neurons of TG+(S)-ketamine group (Figures S16A–S16E). Dorsal CA1 neurons of (S)-ketamine-infused group showed significantly increased Rin (Figure 12B) and decreased fR (Figure 12C) at different membrane potentials (ranging from −65 mV to −75 mV, interval −5 mV) compared with those from either saline- or D-AP5-infused group, indicating a decrease in functional Ih. Consistent with a previous result (Kim et al., 2018), dorsal CA1 neurons of TG-infused group showed decreased Rin (Figure 12B) and increased fR (Figure 12C) at different membrane potentials compared with those from either saline- or D-AP5-infused group, indicating upregulation of functional Ih. Interestingly, dorsal CA1 neurons of TG+(S)-ketamine group showed increased Rin (Figure 12B) and decreased fR (Figure 12C) at different membrane potentials compared with those from TG-infused group indicating the reversal of the effects of TG by (S)-ketamine. The number of AP firings elicited by depolarizing current steps was significantly increased in the dorsal CA1 neurons of (S)-ketamine-infused group compared with those from either saline- or D-AP5-infused group (Figures 12D and 12E) indicating an increase in neuronal excitability. Consistent with TG-induced decrease in Rin, the number of APs was significantly decreased in the dorsal CA1 neurons of TG-infused group compared with those from either saline- or D-AP5-infused group (Figures 12D and 12E) indicating a decrease in neuronal excitability. Dorsal CA1 neurons of TG+(S)-ketamine group showed increased neuronal excitability compared with those from the TG-infused group (Figures 12D and 12E), suggesting that TG-induced hypoexcitability was reversed by (S)-ketamine.

Figure 12

TG-Induced Upregulation of Functional Ih Was Reversed by (S)-Ketamine

(A) Timeline of thapsigargin and (S)-ketamine experiment.

(B and C) Dorsal CA1 neurons of (S)-ketamine-infused group showed increased Rin (B) and decreased fR (C) at different membrane potentials (ranging from −65 mV to −75 mV; −5 mV interval) compared with those from saline- or D-AP5-infused group. Dorsal CA1 neurons of TG-infused group showed decreased Rin (B) and increased fR (C) at different membrane potentials compared with those from saline- or D-AP5-infused group. Dorsal CA1 neurons of TG+(S)-ketamine-infused group showed that TG-induced upregulation of functional Ih was reversed by (S)-ketamine. ∗p < 0.05 (sal or D-AP5 versus (S)-ket) and #p < 0.05 (sal or D-AP5 versus TG) by two-way ANOVA with Sidark's multiple comparisons test.

(D) Representative voltage responses with depolarizing current step (240 pA; 750 ms) at RMP.

(E) Dorsal CA1 neurons of (S)-ketamine-infused group showed increased action potential firing at RMP, whereas decreased at RMP in the dorsal CA1 neurons of TG-infused group compared with those from saline- or D-AP5-infused group. Dorsal CA1 neurons of TG+(S)-ketamine-infused group showed increased neuronal excitability compared with those from TG-infused group. ∗p < 0.05, ∗∗p < 0.01 (sal or D-AP5 versus (S)-ket) and #p < 0.05 (sal or D-AP5 versus TG) by two-way ANOVA with Sidark's multiple comparisons test. Data are expressed as mean ± SEM.

Discussion

We found that (S)-ketamine reduced dendritic Ih in unstressed conditions, whereas it decreased both somatic and dendritic Ih in the CUS model of depression. The (S)-ketamine-induced change in Ih-sensitive parameters was consistent with the reduction in h current and a hyperpolarizing shift to the h channel activation curve. (S)-ketamine also reversed the CUS-induced decreases in neuronal excitability. (S)-ketamine pretreatment before the onset of depression prevented the CUS-induced abnormal behaviors (i.e., anxiogenic- and depressive-like behaviors) and neuropathological changes (i.e., upregulation of somatic Ih and a decrease in neuronal excitability). Finally, in vivo infusion of TG-induced anxiogenic- and anhedonic-like behaviors and upregulation of functional Ih were reversed by (S)-ketamine.

In humans a single sub-anesthetic dose of (R,S)-ketamine (0.5 mg/kg, i.v. injection for 40 min) has been shown to have rapid (hours) and sustained (days to a week) antidepressant effects in TRD (Berman et al., 2000, Zarate et al., 2006a). Recently, Canuso et al., reported efficacy and safety of FDA-approved intranasal (S)-ketamine in TRD and antisuicidal effects in major depression (Canuso et al., 2018). In addition, continued treatment with intranasal (S)-ketamine plus an oral antidepressant had a significantly delayed time to relapse (i.e., sustained effect) among patients with TRD compared with placebo nasal spray plus an oral antidepressant treatment for 16 weeks (Daly et al., 2019). In preclinical studies, a single sub-anesthetic dose of (R,S)-ketamine (10–15 mg/kg i.p.) produces rapid and sustained antidepressant-like effects in the FST (Autry et al., 2011, Li et al., 2010). Consistent with (R,S)-ketamine's antidepressant effects, we also observed that a single sub-anesthetic dose of (S)-ketamine (15 mg/kg i.p.) produced rapid (1 h after (S)-ket treatment) and sustained (24 h after (S)-ket treatment) antidepressant effects in the FST. In spite of the compelling clinical and preclinical evidence supporting (R,S)-ketamine's rapid and robust antidepressant effects on depression, the cellular mechanisms underlying its antidepressant actions are still under debate. Because (R,S)-ketamine is a noncompetitive NMDA receptor antagonist (Anis et al., 1983), the prevailing hypothesis for (R,S)-ketamine's antidepressant effects is through blockade of NMDA receptors. However, it has been reported that antidepressant-relevant concentrations of (2R,6R)-hydroxynorketamine (HNK; less than 10 μM), one of ketamine's metabolites, exert acute antidepressant-like effects in the social defeat model of depression in an NMDA receptor-independent manner (Zanos et al., 2016, Lumsden et al., 2019). In contrast, Suzuki et al., reported that 50 μM (R,S)-ketamine or (2R,6R)-HNK blocks synaptic NMDA receptors (Suzuki et al., 2017). Furthermore, rapid antidepressant effects of (R,S)-ketamine are mediated by the opioid receptor activation, although the sample size (n = 7) of (R,S)-ketamine-responsive patients with TRD is small (Williams et al., 2018). Despite the controversy over the role of NMDA receptors in the antidepressant action of (R,S)-ketamine (Preskorn et al., 2008, Lodge and Mercier, 2015, Zarate et al., 2006b, Zanos et al., 2016, Suzuki et al., 2017, Williams et al., 2018), the antidepressant effect is generally associated with increased BDNF/mTOR signaling, a decrease in eEF2 kinase, and synaptic AMPA receptors (Li et al., 2010, Autry et al., 2011, Maeng et al., 2008).

We have previously reported that a lentiviral small hairpin RNA-mediated reduction of the HCN1 subunit of h-channels (and Ih) in the dorsal CA1 region/neurons leads to an increase in neuronal excitability, upregulation of BDNF/phosphorylation of mTOR protein expression, and an increase in excitatory synaptic transmission (i.e., field excitatory postsynaptic potential slope), which contributes to antidepressant- and anxiolytic-like behaviors in unstressed conditions (Kim et al., 2012). Given that a single sub-anesthetic dose of (R,S)-ketamine in the rodent produces similar downstream signaling pathways (e.g., increased BDNF-mTOR signaling) and behavioral outcomes (antidepressant-like behaviors) (Li et al., 2010, Autry et al., 2011, Zanos et al., 2016), it is possible that (R,S)-ketamine-mediated antidepressant effects are associated with a reduction of Ih. Rats exposed to CUS for 2–3 weeks show anxiogenic-like (e.g., decreased locomotor activity and center square entries in the open field test) and depressive-like (e.g., decreased sucrose preference in a two-bottle choice test and increased passive activity in the FST) behaviors (Kim et al., 2018). Because CUS-treated rats showed decreased locomotor activity in the open field test, a correlation between locomotor activity and anxiety level is suggested. For example, a low dose of diazepam (1 mg/kg i.p.), an anxiolytic drug, increases locomotor activity and center square entries in the open field test, which can be interpreted as increased locomotion caused by decreased anxiety (Kim et al., 2012). Although CUS-treated rats showed increased anxiety, which might be caused by decreased exploration or increased innate fear response to a novel environment, there might be a limitation to this interpretation. In our previous report, somatic Ih and the expression of the HCN1 protein are upregulated in the dorsal CA1 neurons/region following CUS but not acute stress (Kim et al., 2018). This upregulation of somatic Ih is associated with chronic stress-induced elevated intracellular calcium levels (Narayanan et al., 2010, Kim et al., 2018). There is also a strong correlation between the time course of the increase in somatic Ih and the development of the depression-like symptoms (Kim et al., 2018). These results suggest a possible link between HCN channels and depression (Kim et al., 2018, Kim and Johnston, 2018). The important question was whether (S)-ketamine reduced the CUS-induced upregulation of Ih of dorsal CA1 neurons in the CUS model of depression.

Chen et al. reported that (S)-ketamine reduces HCN1 but not HCN2 subunit-mediated Ih via a hyperpolarizing shift in voltage dependence of activation in layer 5 cortical pyramidal neurons (V1/2: ∼ −10 mV) and in HEK 293 cells expressing homomeric mouse HCN1 channels (V1/2: ∼ −15 mV) with somatic recordings in unstressed conditions (Chen et al., 2009). We initially attempted to reproduce this (S)-ketamine reduction in somatic Ih-sensitive electrophysiological measurements (i.e., Rin and fR) of dorsal CA1 neurons. However, we did not observe any changes in somatic Ih-sensitive measurements despite using four different concentrations of (S)-ketamine (20, 40, 80, 160 μM) in dorsal or ventral CA1 neurons with or without NMDA receptor blockers (Figure S2). Consistent with these negative results, we did not see any somatic changes in Ih-sensitive measurements following 50 μM (S)-ketamine in the presence of glutamatergic synaptic blockers (D-AP5 and DNQX). Given that HCN channels are heavily expressed in dendrites of CA1 neurons (Magee, 1998), we, therefore, measured Ih-sensitive measurements in dendrites in the presence of D-AP5 (25 or 50 μM) and DNQX. Indeed, (S)-ketamine reduced dendritic Ih-sensitive measurements (i.e., increased Rin and decreased fR) at depolarized membrane potentials within 10–20 min. Consistent with this result, a reduction of Ih-sensitive measurements by (S)-ketamine was distance-dependent along the somatodendritic axis of dorsal CA1 neurons. Given a very low expression of HCN1 channels (therefore Ih) in the perisomatic region of dorsal CA1 neurons in unstressed conditions, it is possible that the lack of effects of (S)-ketamine on somatic Ih might be due to a very low expression of HCN1 channels at the soma and the difficulty of measuring Ih in cell-attached patches. It is also possible that heteromeric somatic h channel in unstressed conditions are composed predominantly of HCN2 subunits, which are insensitive to (S)-ketamine (Wang et al., 2001), but shift to predominantly HCN1-containing h channel after CUS. Further experiments will be needed to address these and other possibilities.

GIRK conductance in the dorsal CA1 neurons contributes to intrinsic membrane properties including Rin (Kim and Johnston, 2015). We found that the increased Rin by (S)-ketamine was independent of barium-sensitive conductances (GIRK and IRK) in both unstressed and CUS groups. HCN channels are also regulated by binding of cAMP to the cyclic nucleotide-binding domain, which results in a subunit specific generation of a depolarizing shift of the Ih voltage activation curve (Wang et al., 2001). Recently, Wray et al. showed that treatment of C6 glioma cells or primary astrocytes with (R,S)-ketamine (10 μM for 15 min) increases cAMP-dependent BDNF protein expression after 24 h (Wray et al., 2018). We, therefore, examined whether cAMP-dependent signaling was involved in the regulation of Ih by (S)-ketamine. 8-Bromo-cAMP is an activator of cAMP-dependent protein kinase. The (S)-ketamine-induced reduction in dendritic Ih in unstressed group or CUS-induced upregulation of somatic Ih was independent of cAMP-dependent signaling. We also examined the possibility that (S)-ketamine could further change Ih -sensitive measurements (e.g., Rin) after blockade of Ih by ZD7288. However, successive ZD7288 and (S)-ketamine wash-in experiment with dendrite recordings revealed that the ZD7288-induced change in Rin was not further altered by addition of (S)-ketamine treatment, suggesting an occlusion effect with (S)-ketamine.

Cell-attached patch-clamp recordings showed that (S)-ketamine treatment led to changes in voltage-dependence of h channel (i.e., a hyperpolarizing shift to the h channel activation curve) and the amplitude of h current (i.e., decreased maximal h current) in dendrites but not soma in the unstressed group. Consistent with previous results (Kim et al., 2018), somatic but not dendritic Ih-sensitive measurements of dorsal CA1 neurons were upregulated in the CUS model of depression. Cell-attached patch-clamp recordings at the soma revealed that the V1/2 of the h channel activation curve for CUS group was significantly shifted to the right around +29 mV compared with those from unstressed group. Maximal Ih at −170 mV for CUS group was around – 3.89 pA, whereas maximal Ih for unstressed group was around −0.59 pA. Whole-cell current-clamp recordings revealed that (S)-ketamine significantly reduced the CUS-induced upregulation of somatic Ih-sensitive measurements of dorsal CA1 neurons at depolarized membrane potentials, whereas no changes in the unstressed group were observed. We also confirmed that (S)-ketamine treatment for CUS group led to a hyperpolarizing shift to the h channel activation curve around - 23 mV and a decrease in h current around −2.4 pA compared with those from ACSF-treated CUS group. (S)-ketamine also reduced dendritic Ih-sensitive measurements of dorsal CA1 neurons in both unstressed and CUS groups.

It has been reported that a single sub-anesthetic dose of (R,S)-ketamine pretreatment before the onset of stress prevents either chronic stress-induced depression-like behaviors (Brachman et al., 2016) or acute stress (e.g., inescapable tail shock)-induced fear responses or anxiety (Amat et al., 2016, Mcgowan et al., 2017). Consistent with these reports, we also observed that a single sub-anesthetic dose of (S)-ketamine pretreatment 1 week before the onset of CUS prevented the CUS-induced anxiogenic- and depressive-like behaviors. We have reported that a reduction of HCN1 protein expression in the dorsal CA1 region before the onset of CUS provides resiliency to the depression-like symptoms (Kim et al., 2018). In addition, knockdown of HCN1 protein expression or neuropeptide Y-induced reduction of Ih into basolateral amygdala is sufficient to exert resiliency to stress in rats (Silveira Villarroel et al., 2018). Interestingly, the saline-pretreated CUS group showed upregulation of somatic Ih, whereas no changes in the (S)-ketamine-pretreated CUS group were observed.

Functional neuroimaging of patients with major depression has shown bimodal, abnormal regional metabolic activities in different regions of limbic-cortical areas (Mayberg et al., 2000, Mayberg et al., 2005). Given a positive correlation between brain energy metabolism and neuronal excitability, abnormal change in neuronal excitability (e.g., a number of APs at RMP) has been reported following chronic stress (Kim et al., 2018, Rosenkranz et al., 2010). Dorsal CA1 neurons from CUS group had a lower Rin and a high fR compared with those from unstressed group. As expected, neuronal excitability was significantly reduced in the dorsal CA1 neurons following CUS. This decrease in neuronal excitability was reversed by bath application of (S)-ketamine within 10–20 min. Furthermore, (S)-ketamine pretreatment before the onset of depression prevented CUS-induced decrease in neuronal excitability. Therefore, (S)-ketamine-induced changes in intrinsic membrane properties may contribute to the changes in behaviors (i.e., anxiolytic- and antidepressant-like behaviors).

Recently, Lumsden et al. reported that (2R,6R)-HNK, at the dose of 10 mg/kg (i.p), exerts antidepressant-like effects associated with increased mature BDNF and phosphorylation of mTOR protein expression (Lumsden et al., 2019). This antidepressant-relevant concentration of (2R,6R)-HNK (at dose of 10 mg/kg, i.p.) generates around 8 μM ketamine metabolite in the serum, whole brain, and ventral hippocampus (Lumsden et al., 2019). Given that the antidepressant-relevant concentrations of (R,S)-ketamine are ranging from 3 to 30 mg/kg (i.p.) in rodent (Kim et al., 2012, Lumsden et al., 2019, Autry et al., 2011, Iijima et al., 2012, Koike et al., 2013), their concentrations in the brain might be higher than estimated (e.g., more than 10 μM of (R,S)-ketamine and its metabolite). Similarly, Fava et al., reported that even higher concentration of (R,S)-ketamine (i.e., 1 mg/kg i.v. for 40 min) exerts antidepressant effects in TRD, whereas lower doses of (R,S)-ketamine (0.1 and 0.2 mg/kg i.v. for 40 min) do not exert clinically meaningful antidepressant effects (Fava et al., 2018). Consistent with dose-dependent effects of (R,S)-ketamine, we found that a reduction of dendritic Ih-sensitive measurements by (S)-ketamine was dose dependent in the dorsal CA1 neurons.

It has been reported that patients with depression showed an elevated basal intracellular Ca2+ concentration in platelets and lymphocytes (Kerr et al., 1992, Karst et al., 2000). In our previous reports (Narayanan et al., 2010, Kim et al., 2018), in vitro or in vivo block of the SERCA pumps in CA1 neurons increases perisomatic Ih. Furthermore, in vivo infusion of TG produces anxiogenic-like behavior, similar to that observed in CUS model of depression (Kim et al., 2018). In this study, TG and (S)-ketamine were dissolved with D-AP5 to reduce the contribution of NMDA receptors in our in vivo study. We observed not only decreased center square entries in the open field test (i.e., anxiogenic-like behavior) but also decreased sucrose preference in the two-bottle choice test (i.e., anhedonic-like behavior) from in vivo TG-infused group. TG-induced changes in behaviors were reversed by (S)-ketamine. Whole-cell current-clamp recordings after a 10-min open field test revealed that dorsal CA1 neurons of (1) TG-infused group showed increased functional Ih (i.e., decreased Rin and increased fR), (2) (S)-ketamine-infused group showed decreased functional Ih (i.e., increased Rin and decreased fR), and (3) TG+(S)-ketamine-infused group showed decreased functional Ih. Consistent with changes in Rin from TG- or (S)-ketamine- or TG+(S)-ketamine-infused group, we observed similar changes in neuronal excitability from these groups. These behavioral and electrophysiological results suggest that (S)-ketamine's antidepressant effects are through a reduction of HCN1 channels, and therefore Ih, in the dorsal CA1 region/neurons.

In summary, we found that (S)-ketamine reduced dendritic but not somatic Ih in an NMDA receptor-independent manner in unstressed conditions. (S)-ketamine normalized the CUS-induced neuropathological changes, which resulted in reduced somatic Ih and increased neuronal excitability. A single dose of (S)-ketamine pretreatment before the onset of depression prevented the CUS-induced neuropathological changes. Finally, in vivo infusion of TG-induced anxiogenic- and anhedonic-like behaviors and upregulation of functional Ih were reversed by (S)-ketamine. To our knowledge, this is the first report that (1) (S)-ketamine reduced the CUS-induced upregulation of somatic Ih, (2) (S)-ketamine pretreatment before the onset of depression prevented the CUS-induced neuropathological changes such as upregulation of somatic Ih, and (3) antidepressant effects of (S)-ketamine are through a reduction of HCN1/Ih. Our findings suggest that the NMDA receptor-independent cellular mechanisms of (S)-ketamine reported here may underlie some of (S)-ketamine's antidepressant actions and resiliency to chronic stress.

Limitations of the Study

In this study, we systematically investigated the effects of (S)-ketamine on Ih. Further investigation is required, however, into whether a change in Ih by pretreatment of (S)-ketamine is due to direct or indirect effects on Ih. We only used male rats in this study.

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Chung Sub Kim (ckim5@augusta.edu).

Materials Availability

This study did not generate any new reagents.

Data and Code Availability

This published article includes all datasets/code generated or analyzed during this study.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.