Abstract
GABAB receptors in habenula cholinergic neurons mediate strong presynaptic excitation and control aversive memory expression. K+ channel tetramerization domain (KCTD) proteins are key interacting partners of GABAB receptors; it remains unclear whether and how KCTDs contribute to GABAB excitatory signaling. Here, we show that KCTD8 and KCTD12 in these neurons facilitate the GABAB receptors expression in axonal terminals and contribute to presynaptic excitation by GABAB receptors. Genetically knocking out KCTD8/12/16 or KCTD8/12, but not other combinations of the three KCTD isoforms, substantially reduced GABAB receptors–mediated potentiation of glutamate release and presynaptic Ca2+ entry in response to axonal stimulation, whereas they had no effect on GABAB-mediated inhibition in the somata of cholinergic neurons within the habenulo–interpeduncular pathway in mice of either sex. The physiological phenotypes were associated with a significant decrease in the GABAB expression within the axonal terminals but not the somata. Overexpressing either KCTD8 or KCTD12 in the KCTD8/12/16 triple knock-out mice reversed the changes in axonal GABAB expression and presynaptic excitation. In mice lacking the KCTDs, aversion-predicting cues produced stronger neuronal activation in the interpeduncular nucleus, and the infusion of GABAB agonist in this nucleus produced a weaker effect on fear extinction. Collectively, our results reveal isoform-specific roles of KCTD proteins in enriching the axonal expression of GABAB receptors, facilitating their presynaptic signaling, and modulating aversion-related memory processes.
SIGNIFICANCE STATEMENT GABAB receptors represent the principal inhibitory neurotransmitter receptor, but they mediate strong presynaptic excitation in the habenulo–interpeduncular pathway and modulate aversion memory expression. KCTD proteins are integral constituents of GABAB receptors. By analyzing the physiological, neuroanatomical, and behavioral phenotypes of multiple KCTD knock-out mouse lines, we show that KCTD8 and KCTD12 facilitate the axonal expression and hence presynaptic excitation of GABAB receptors in habenula cholinergic neurons and control cued-aversion memory formation and expression in the habenulo–interpeduncular pathway. These results expand the physiological and behavioral functions of KCTDs in modulating the brain neural circuits.
- baclofen
- Ca2+ imaging
- GABAB auxiliary subunits
- habenulo–interpeduncular pathway
- K+ channel tetramerization domain
- optogenetics
Introduction
The cholinergic pathway projecting from the ventral part of the medial habenula (MHb) to the rostral and central core of the interpeduncular nucleus (IPN) controls aversive memories, anxiety, fear responses, and nicotine dependence in vertebrates (Kim, 2009; Qin and Luo, 2009; Quina et al., 2009; Agetsuma et al., 2010; Frahm et al., 2011; Kobayashi et al., 2013; Yamaguchi et al., 2013; Frahm et al., 2015; Soria-Gómez et al., 2015; Ables et al., 2017; Zhang et al., 2017). Habenula cholinergic neurons corelease glutamate, acetylcholine, and neurokinin B to activate postsynaptic neurons in the IPN (Ren et al., 2011; Zhang et al., 2016). GABAB receptors—heteromeric G-protein-coupled receptors composed of the GABAB1 and GABAB2 subunits—are highly expressed in these neurons (Gassmann and Bettler, 2012; Zhang et al., 2016; Fritzius and Bettler, 2020). Although GABAB receptors are predominantly inhibitory in the brain (Gassmann and Bettler, 2012), they mediate strong presynaptic excitation in the habenulo–interpeduncular pathway by amplifying presynaptic Ca2+ entry through Cav2.3 channels and potentiating neurotransmitter corelease to IPN neurons (Zhang et al., 2016). At the behavior level, inactivating GABAB receptors in habenula cholinergic neurons impairs the extinction of aversive memory in mice (Zhang et al., 2016).
Cytosolic K+ channel tetramerization domain (KCTD) proteins KCTD8, 12, 12b, and 16 assemble with GABAB receptors by interacting with the GABAB2 subunit (Schwenk et al., 2010, 2016). The different KCTDs are known to exert a variety of effects on GABAB-mediated responses. For example, KCTD12, 12b, and 16 shorten the rise time of GABAB-induced opening of potassium channels and facilitate the desensitization of GABAB receptors in cultured cells (Schwenk et al., 2010; Seddik et al., 2012; Turecek et al., 2014; Fritzius et al., 2017; Zheng et al., 2019; Zuo et al., 2019). The GABAB-associated KCTDs exhibit a distinct spatial and temporal distribution pattern in the mammalian brain and have been implicated in neuropsychiatric disorders in humans (Metz et al., 2011; Teng et al., 2019).
Several KCTDs are richly expressed in MHb neurons (Metz et al., 2011). Considering that GABAB-mediated presynaptic excitation in habenula neurons and that KCTD proteins modulate the kinetics of GABAB-mediated responses, we asked whether and how KCTDs affect GABAB-mediated presynaptic excitation in the MHb–IPN pathway. Recently, Bhandari et al. (2021) showed that KCTD8 and KCTD12b modulate Cav2.3-mediated release from MHb terminals but are not involved in GABAB receptors–mediated presynaptic excitation. This study did not examine the potential roles of KCTD12 and KCTD16. Here, we took advantage of the recently developed KCTD8, 12, and 16 triple knock-out (3KO) mouse model (Rajalu et al., 2015) and various combinations of isoform deficiency to investigate the role of KCTDs on GABAB signaling in habenula cholinergic neurons. Our results reveal that KCTDs have a strong contribution to GABAB axonal expression and presynaptic excitation, but not soma expression and postsynaptic inhibition, with KCTD8 and KCTD12 playing important and redundant roles. Our data demonstrate isoform- and site-specific association of KCTDs with presynaptic GABAB receptors expression and expand the physiological and behavioral functions of KCTDs in modulating the brain neural circuits.
Materials and Methods
Animals
All procedures were conducted following the approval of the Animal Care and Use Committee of the National Institute of Biological Sciences, Beijing, in accordance with the governmental regulations of China. We used adult mice (8–12 weeks, 18–25 g) of either sex for the patch-clamp recording, two-photon Ca2+ imaging, and immunostainings. We used adult male mice for the aversive conditioning, fiber photometry, and intra-IPN drug infusion. The ChAT-ChR2-EYFP BAC-transgenic mice were a gift from G. Feng (Massachusetts Institute of Technology) and had been crossed to the C57BL/6N background for more than 10 generations. Homozygous KCTD8/12/16 KO mice were a gift from Bernhard Bettler (University of Basel). We crossed ChAT-ChR2-EYFP BAC-transgenic mice with KCTD8/12/16 KO mice to obtain ChAT-ChR2-EYFP mice with heterozygous KCTD8/12/16 as F1. By self-fertilizing the F1 generation, we obtained ChAT-ChR2-EYFP mice with different types of KCTD deficiency: KCTD8 KO::ChAT-ChR2, KCTD12 KO::ChAT-ChR2, KCTD16 KO::ChAT-ChR2, KCTD8/12 KO::ChAT-ChR2, and KCTD8/12/16 KO::ChAT-ChR2 mice. All mice were bred and maintained at a specific-pathogen-free mouse facility. Wild-type (WT) C57BL/6N mice were purchased from VitalRiver. All mice were maintained on controlled temperature (22–25°C) and a 12/12 light/dark cycle with access to food and water ad libitum.
AAV preparation and injection
Adeno-associated virus (AAV) vectors carrying GCaMP6m, mCherry, KCTD8-P2A-mCherry, and KCTD12-P2A-mCherry were packaged into 2/9 serotypes with titers of 1–5 × 1012 viral particles/ml. We constructed the plasmid pAAV-EF1a-GCaMP6m by replacing the coding region of DIO-ChR2-mCherry in pAAV-EF1a-DIO-hChR2 (H134R)-mCherry plasmid (catalog #20297, Addgene; a gift from Dr. Karl Deisseroth) with that of GCaMP6m plasmid (catalog #40754, Addgene; a gift from Dr. Douglas Kim). The KCTD8 and KCTD12 sequences were cloned from mouse brain cDNA. We built a fusion construct that was composed of the sequences for KCTD, the P2A peptide, and an mCherry reporter (KCTD-P2A-mCherry). These fusion constructs were inserted into a pAAV-EF1a backbone.
For AAV injections, adult mice were anesthetized with pentobarbital (80 mg/kg, i.p.) and then mounted on a stereotaxic apparatus. The skin was cut, and a small craniotomy was performed above the MHb or the IPN. Injections were performed using a microsyringe pump (Nanoliter 2010 injector, World Precision Instruments). A micro controller (World Precision Instruments) was used to deliver the virus solution to the target areas at a rate of 46 nl/min. The following coordinates were used to target specific brain areas: 1.3–1.7 mm posterior to the bregma, 0.1 mm lateral to midline, and 2.6 mm ventral to the skull surface for the MHb; 3.4 mm posterior to the bregma, 1.25 mm lateral to midline, and 4.8 mm ventral to the skull surface with a 15° angle (lateral to midline) for the IPN.
Brain slice preparation, patch-clamp recording, and two-photon Ca2+ imaging
Adult mice were anesthetized with pentobarbital (80 mg/kg, i.p.) and then transcardially perfused with 5 ml ice-cold oxygenated slice solution (saturated with 95% O2 and 5% CO2). The slice solution contains the following as reagent (in mm): 110 choline chloride, 2.5 KCl, 0.5 CaCl2, 7 MgCl2, 1.3 NaH2PO4, 25 NaHCO3, 10 glucose, 1.3 Na-ascorbate, and 0.6 Na-pyruvate. The slice solution was adjusted to 305–315 mOsm/L using sucrose. Next, the mouse brains were dissected and transferred into ice-cold oxygenated slice solution. Brains were first blocked at a 54° deviate angle from the horizontal plane, and sections containing the MHb–IPN pathway (200 μm, two per mouse) were cut with a vibratome (VT1200s, Leica). Slices were incubated for at least 1 h at 33°C within oxygenated artificial cerebrospinal fluid (ACSF) containing the following (in mm): 125 NaCl, 2.5 KCl, 2 CaCl2, 1.3 MgCl2, 1.3 NaH2PO4, 1.3 Na-ascorbate, 0.6 Na-pyruvate, 10 glucose, and 25 NaHCO3 (305-315 mOsm/L). The brain slices were transferred to a recording chamber at room temperature for recordings and imaging. All chemicals for slicing preparation were purchased from Sigma.
For recordings, slices were submerged and perfused with ACSF at a rate of 3 ml/min at room temperature. Neurons were identified with differential interference contrast optics (Zeiss Examiner.Z1). The recording pipettes (3–4 MΩ) for whole-cell recording and cell-attached recording were pulled by a P-1000 glass pipette puller (Sutter Instrument). For whole-cell recordings, the pipettes were filled with internal solution that contained the following as reagent (in mm): 130 K-gluconate, 10 HEPES, 0.6 EGTA, 5 KCl, 3 Na2ATP, 0.3 Na3GTP, 4 MgCl2 and 10 Na2-phosphocreatine, (pH 7.2–7.4, 295–305 mOsm/L). For cell-attached recordings, ACSF was used as the intrapipette solution. Slice recordings were performed with MultiClamp 700B (Molecular Devices) and pClamp software. We conducted whole-cell recordings from 98 IPN neurons (85/98 in the rostral IPN and 13/98 in the central IPN). The neurons were held at −65 mV. The traces were low-pass filtered at 3 kHz and digitized at 10 kHz (Axon Digidata 1322A, Molecular Devices). The electrophysiological data were analyzed with Clampfit 10.2 software (Molecular Devices). We used the membrane test function in pClamp to measure membrane capacitance (Cm) and resistance (Rm). Membrane time constant (Tau) was determined by Tau = Cm * Rm (Isokawa, 1997). We determined the resting membrane potential of IPN neurons by subjecting neurons to 10 mV voltage steps and choosing the point of zero current from data interpolation.
For optogenetic stimulation, an optical fiber (0.2 mm core diameter, NA = 0.22) linked to a 473 nm laser driver (MBL-III-473, Changchun New Industries Optoelectronics Technology) was submerged in ACSF and placed ∼0.3 mm from the recording site. The light intensity reaching the brain tissue was ∼9 mW/mm2.
The drugs were applied through perfusion by adding them to the ACSF following the dilution of a stock solution. The drug application onset indicates the switch in perfusion from standard ACSF to the drug-containing ACSF. The drugs include the following: baclofen (GABAB agonist, 1 μm; Sigma), picrotoxin (GABAA antagonist, 50 μm, Sigma), mecamylamine (Mec; nAChR antagonist, 5 μm; Sigma), hexamethonium bromide (HMT; nAChR antagonist, 50 μm; Sigma), and 6, 7-dinitroquinoxaline-2, 3-dione (DNQX; AMPA antagonist, 10 μm; Sigma).
For two-photon imaging, brain slices containing the IPN were prepared, and oxygenated ACSF was continuously perfused at a rate of 3 ml/min at room temperature (25 ± 2°C). GCaMP6m fluorescent signals were imaged with a 20× water immersion objective on a 2-photon microscope (FV1000, Olympus) at the rate of 1 Frame Per Second (FPS). We monitored the cholinergic axons in the IPN. To evoke Ca2+ transients, a bipolar electrode was placed in the IPN, and the electrical stimulation (1 ms per pulse, 50 or 100 μA) at 10 Hz 1 s was applied for each test condition. The stimulation levels (50 or 100 μA) were chosen to induce measurable Ca2+ signal changes within the imaging field in ACSF solution; they were comparable in WT mice and KCTD8/12/16 triple KO mice (WT: 50 μA for 5 fields and 100 μA for 2 fields; triple KO: 50 μA for 5 fields and 100 μA for 1 field). Fluorescence intensities of GCaMP were measured with ImageJ software. We used two slices per mouse and chose 1–2 imaging fields per slice. In every field, we manually identified puncta as the regions of interest (ROIs) where Ca2+ signal showed increased response to electrical stimulation in ACSF. The intensities were then processed and plotted using a custom-written MATLAB program. We measured the area under the (peri-event) curve (AUC) between electrical stimulation onset (0 s) and peak Ca2+ signals (3 s) to represent the response strength to the baclofen treatment. ROIs size in WT mice and KCTD8/12/16 KO mice was comparable.
Immunostaining and immunoblotting
Mice were deeply anesthetized with an overdose of pentobarbital and transcardially perfused with 0.9% saline, followed by paraformaldehyde (PFA, 4% w/v in PBS). Brains were removed and postfixed in 4% PFA for 4 h at room temperature. After samples were dehydrated in 30% sucrose solution, thin sections (35 μm) were prepared on a cryostat microtome (Leica CM1950). For antigen retrieval, we incubated the sections with sodium citrate buffer solution (10 μm; Sigma), adjusted to pH 6.0 with 1 m HCl at 95°C in a dry bath heater for 6 min (repeat twice with an interval of 15 min), then rinsed sections with PBS to cool down. After rinsing with 0.3% Triton X-100 in PBS (PBST) and blocking in 2% (w/v) BSA in PBST at room temperature for 1 h, the brain sections were incubated with primary antibodies anti-GABAB2 (1:200; catalog #G9920, Merck), anti-KCTD8 (1:200; catalog #ab110759, Abcam), anti-KCTD12 (1:200; catalog #15523-1-AP, Proteintech), anti-ChAT (choline acetyltransferase; 1:200; catalog #AB144P, Merck), and anti-Cav2.3 (1:200; catalog #C1853, Sigma) in the blocking solutions at 4°C for 48 h. After washing with PBS, the brain sections were incubated with fluorescent secondary antibodies (Cy2 or Cy3-conjugated donkey anti-goat, Cy2 or Cy3-conjugated goat anti-rabbit, Jackson ImmunoResearch) at room temperature for 2 h. Finally, PBS-washed sections were mounted with DAPI containing 50% glycerol. Fluorescent images were collected with a confocal microscope (Zeiss LSM880) or an automated fluorescent scanner (VS120Virtual Slide, Olympus). Images were analyzed and quantified using ImageJ.
For immunoblotting, protein extracts were prepared from the IPN of KCTD8/12/16 KO or WT mice, lysed in a strong denaturing buffer containing 50 mm Tris-HCl, pH 7.6, 150 mm NaCl, 2 mm EDTA, 1% Triton X-100, 0.5% Sodium deoxycholate, 0.1% SDS, and a protease inhibitor cocktail (Roche Molecular Biochemicals) at 4°C. After sonication, soluble fraction was obtained by centrifugation at 130,000 × g for 20 min at 4°C. Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Protein extracts were denatured by the SDS loading buffer and boiled for 15 min. Then boiled samples were separated on a 4–20% SDS-PAGE gel before transfer to a PVDF membrane (Millipore) for further immunoblotting analysis. We used antibodies for KCTD8 and KCTD12 as mentioned above and other antibodies for KCTD16 (1:200; catalog #ab185104, Abcam) and α-tubulin (T8328-100UL, 1:30,000, SigmaAldrich).
Aversive conditioning, fiber photometry, and intra-IPN infusion
We used KCTD8/12/16 KO, KCTD8/12 KO, and WT mice for the behavioral experiments. On day 0 (conditioning sessions), a mouse was placed into a conditioning chamber (24 × 24 × 30, L × W × H cm) with a metal fence floor and allowed to explore freely for 3 min. For fiber photometry experiments, 30 trials that consisted of footshock (0.6 mA scrambled, 0.5 s) delivered after an auditory tone (2 s, 7.5 kHz, 85–90 dB) with a 1 s delay were then conducted. For the intra-IPN drug injections, prolonged auditory tones (20 s, 7.5 kHz, 85–90 dB) were coupled to footshocks (0.7 mA scrambled, 1 s) and both stimuli coterminated for 10 trials. On days 1–3 (extinction sessions), mice were introduced into a new test chamber (40 × 30 × 30, L × W × H cm) that contained a white filter paper on the floor. Thirty or 10 auditory tones (same as the auditory tone for conditioning) were delivered after 2 min habituation without any footshock. Mice were returned to their home cages 3 min after the end of the last tone. Intertone intervals were randomly set between 30 and 50 s for fiber photometry experiments and 90–150 s for the intra-IPN drug injections.
For fiber photometry experiments, optical fiber implantation was conducted immediately after virus injection. A piece of optical fiber (catalog #FT200UMT, Thorlabs) was fit into a ceramic fiber ferrule (Shanghai Fiblaser). The optical fiber was implanted over the target brain areas with the tip 0.1 mm above the virus injection sites. The ceramic ferrule was supported with dental acrylic. All subsequent experiments were performed at least 3 weeks after virus injection to allow sufficient time for transgene expression and animal recovery. We recorded GCaMP signals using a fiber photometry system (ThinkerTech). We used an exciting beam from a 488 nm laser (0.01–0.02 mW; OBIS CORE 488LS; Coherent). The output signals were amplified and digitalized at 100 Hz. Auditory tone onset was used as the trigger event for data alignment. A custom MATLAB program and an Arduino R3 controller were used to generate the tone signal, and a Power 1401 digitizer simultaneously recorded tone timing and GCaMP signals (100 Hz sampling frequency). Fiber-photometry recording data were exported as MATLAB files from Spike2 software for further analysis. After smoothing the data with the MATLAB smooth function, we segmented the data based on behavioral events within individual trials. We derived the fluorescence change values by calculating the z score, where the baseline fluorescence signal was determined by averaging a 1.5 s long control time window 0.5 s before the cue initiation. The z score values are presented as heatmaps or as peri-event plots. The locomotion videos were analyzed using a custom-written MATLAB program. To examine the response strength of IPN neurons during the test sessions, we calculated the AUC of Ca2+ signal and locomotion for 3 s from conditioned stimuli onset (0 s) to represent the response in the aversion memory phase.
For the intra-IPN drug injections, a guide cannula (26 gauge, Plastics One) was implanted with its tip targeting the dorsal border of the IPN. After postsurgery recovery of 7 d, the mouse underwent the task of cued auditory fear conditioning. Before the extinction sessions in the following days, baclofen (15 pmol in 300 nl) or ACSF of equal volume was slowly infused into the IPN via the internal cannula (100 nl/min). Mice were allowed to rest 30 min to recover from the initial sedation. The mouse behavior was scored off-line by two trained observers without prior knowledge of mouse genotypes or drug treatments. Freezing was defined as the absence of movements, except for those related to respiration and slight head tremble. After the behavioral test, Texas Red conjugated dextran amine (MW = 3 kDa, catalog #D3328; Thermo Fisher Scientific) of equal volume (3%, 300 nl) was infused to the IPN via the cannula to identify the injection site and estimate the extent of the baclofen or ACSF injection.
Quantification and statistical analysis
We used MATLAB version 2019b, ImageJ 2.0.0, and GraphPad Prism 6 to perform the statistical analyses. For quantification of immunoreactivity, we manually identified the IPN regions of every slice as ROIs, and the signal intensity inside the ROIs were measured. To minimize the batch effects of the experiment repeats, we normalized the signal intensity of every section in different KCTD genotypes to the mean signal intensity of control mice in the same experimental batch (see Figs. 2A–D, 4B, WT mice) and mCherry-expressing control mice (see Fig. 5E,F). The precise statistical tests, exact p value, sample sizes (n) denoting the experimental replications, and the exact value of results are reported in the figure legends. All results were reported as means ± SEM. Detailed statistical analyses are shown in Table 1.
Summary of statistical analyses
Results
Triple KO of KCTD8/12/16 reduces GABAB presynaptic excitation
We first analyzed the effect of knocking out KCTD8/12/16 on GABAB receptors–mediated potentiation of the glutamate release from the MHb terminals. Our previous studies have shown that brief photostimulation of the so-called cholinergic habenula neurons in ChAT-ChR2 mice evokes rapid glutamate release, whereas prolong stimulation leads to slow volume transmission of acetylcholine (Ren et al., 2011; Hu et al., 2012; Zhang et al., 2016). By crossing the KCTD8/12/16 KO mice with the ChAT-ChR2 mice, we performed whole-cell recordings from IPN neurons to examine glutamatergic EPSCs on brief stimulation of ChR2-expressing axonal terminals (Fig. 1A). To isolate glutamate currents, we applied a blocker cocktail solution comprising picrotoxin, HMT, and Mec to inhibit GABAA receptors and nicotinic acetylcholine receptors (Ren et al., 2011; Zhang et al., 2016).
Triple KO of KCTD8/12/16 reduces the GABAB receptors–mediated potentiation of synaptic glutamate release and Ca2+ entry of MHb terminals. A, The schematic shows the habenulo–interpeduncular pathway in a coronal section and the recording from IPN neurons in response to photostimulation of ChR2-expressing axonal terminals. B, In IPN neurons of ChAT-ChR2 (left) and KCTD8/12/16 KO::ChAT-ChR2 (right) mice, a brief pulse of photostimulation (5 ms duration, 473 nm) of MHb cholinergic terminals elicited glutamatergic EPSCs that were potentiated by the GABAB agonist baclofen (1 μm). Raw traces and time-series plots are shown. Gray indicates basal responses, and black and blue indicate responses in baclofen application in the two mouse lines. EPSCs were recorded in the presence of a cocktail solution comprising GABAA blocker (50 μm picrotoxin) and nAChR blockers (50 μm HMT and 5 μm Mec). C, Averaged time-series plot of EPSC amplitudes to compare the baclofen-induced potentiation between ChAT-ChR2 mice (IBasal = 66.3 ± 20.1 pA; IBaclofen = 1087.2 ± 164.7 pA; n = 11 cells from 5 mice) and KCTD8/12/16 KO::ChAT-ChR2 mice (IBasal = 62.8 ± 15.1 pA; IBaclofen = 355.6 ± 70.6 pA; n = 12 cells from 4 mice). ****p < 0.0001; Two-way ANOVA test for the difference of EPSCs between ChAT-ChR2 and KCTD8/12/16 KO::ChAT-ChR2 mice. D, Comparison of resting membrane potentials (RMP; WT: −63.6 ± 2.3 mV, n = 11; 3KO: −61.6 ± 1.4 mV, n = 15), Cm (WT: 44.1 ± 5.6 pF, n = 8; 3KO: 36.6 ± 4.3 pF, n = 13), Rm (WT: 268.0 ± 28.8 MΩ, n = 8; 3KO: 318.4 ± 38.0 MΩ, n = 13), and Tau (WT: 11.1 ± 1.3, n = 8; 3KO: 10.7 ± 1.5, n = 13) of IPN neurons recorded from ChAT-ChR2 and KCTD8/12/16 KO::ChAT-ChR2 mice. n.s., Not significant. Unpaired t test and Mann–Whitney test (with Welch's correction). Each dot indicates an individual cell, and the number of dots indicate sample size. E, Example images show AAV transduction of GCaMP6m in MHb neurons (top) and expression of GCaMP6m in the axonal terminals within the IPN (bottom) of WT mice and KCTD8/12/16 KO mice. F, G, Two-photon imaging of GCaMP6m fluorescence changes in the habenular terminals within the IPN of WT mice (F) and KCTD8/12/16 KO mice (G). Left, Example image shows the GCaMP6 expression in the IPN. Right, Pseudocolor images show the GCaMP6 signal changes within the same area by application of the baclofen (1 μm) following electrical stimulation (10 pulses at 10 Hz). H, I, Heatmaps illustrate that baclofen increased GCaMP6 signals for 83 terminal areas of WT mice (7 imaging fields from 3 mice) and 50 terminal areas of KCTD8/12/16 KO mice within (6 imaging fields from 3 mice). J, Average GCaMP6 signals from WT (black) and KCTD8/12/16 KO (blue) mice following electrical stimulations. Dashed line indicates basal, solid line indicates baclofen, and shaded area indicates SEM. K, L, Comparison of AUC of GCaMP signals (from electrical stimulation onset, t = 0 s, to peak GCaMP6 signals, t = 3 s) between WT and KCTD8/12/16 triple KO mice during baseline condition (WT, 10.1 ± 3.0; 3KO, 8.4 ± 1.2; K) and in the presence of baclofen (WT, 93.9 ± 10.2; 3KO, 59.5 ± 7.7; L). Number of individual dots indicate the sample size. n.s., Not significant. *p < 0.05; Mann–Whitney tests. Data are presented as means ± SEM (Table 1). Scale bars: E–G, 100 μm.
We observed a strong decrease in GABAB-mediated presynaptic excitation in the KCTD8/12/16 triple knock-out mice. In the control ChAT-ChR2 mice, GABAB receptors agonist (1 μm baclofen) produced a striking increase in the amplitude of light-evoked EPSCs (Fig. 1B,C). In KCTD8/12/16 KO::ChAT-ChR2 mice, terminal stimulation produced similarly small inward currents, which was also potentiated by baclofen (Fig. 1B,C). However, compared with the control ChAT-ChR2 mice, baclofen-induced potentiation of EPSCs was significantly reduced in KCTD8/12/16 KO::ChAT-ChR2 mice (67.3% reduction in fold increase; Fig. 1C). We note that the intrinsic properties (e.g., resting membrane potential, membrane capacitance, resistance, and membrane time constant) of IPN neurons were not altered in the KCTD8/12/16 KO mice (Fig. 1D). These results suggest that KCTD8, 12, and 16 are involved in the GABAB receptors–mediated potentiation of glutamate release from the MHb.
As GABAB receptors are known to exert excitation via amplifying presynaptic Ca2+ entry through Cav2.3 channels (Zhang et al., 2016), we next tested whether KCTD8/12/16 KO had an effect on Ca2+ entry into the MHb axonal terminals. We used recombinant AAV vectors to express a genetically encoded Ca2+ indicator (GCaMP6m) in MHb neurons and measured GCaMP fluorescence changes from the MHb axonal terminals within the IPN brain sections using two-photon microscopy (Fig. 1E). Baclofen strongly increased presynaptic Ca2+ signals following electrical stimulation (10 pulses at 10 Hz) in wild-type (WT) C57BL/6N mice (Fig. 1F,H). Similarly, baclofen increased presynaptic Ca2+ signals in KCTD8/12/16 KO mice (Fig. 1G,I), suggesting that KCTD8/12/16 KO mice maintained GABAB receptors–mediated amplifying Ca2+ entry of the MHb axonal terminals. Compared with WT mice, we found that changes of Ca2+ signals following electrical stimulation showed no difference without baclofen but were significantly reduced after the application of baclofen in KCTD8/12/16 KO mice (36.6% difference in baclofen-mediated fold increase; Fig. 1J–L). These results thus demonstrated that KCTD8/12/16 KO reduced GABAB receptors–mediated presynaptic Ca2+ entry of the MHb axonal terminals. Our results collectively indicated that the genetic deficiency of KCTD8/12/16 reduced GABAB receptors–mediated presynaptic excitation.
KCTD8/12/16 KO reduces GABAB receptors expression in MHb axonal terminals
Previous studies have shown that KCTDs interact with GABAB receptors and with Cav2.3 (Gassmann and Bettler, 2012; Bhandari et al., 2021). We explored whether KCTD8/12/16 KO influenced axonal expression of Cav2.3 and GABAB receptors by immunostaining of IPN brain sections. We did not detect any significant changes in the expression of Cav2.3 in the MHb axonal terminals between WT and KCTD8/12/16 KO mice (Fig. 2A). Also, the immunoreactivity of choline acetyltransferase (ChAT)—a marker of cholinergic neurons—was unaltered in the MHb axonal terminals of KCTD8/12/16 KO mice (Fig. 2B). Surprisingly, the expression of GABAB2 receptor subunit was significantly reduced in the MHb axonal terminals within the IPN in KCTD8/12/16 KO mice (Fig. 2C). Knocking out KCTD8/12/16 did not change the level of GABAB2 expression in the MHb, where the somata and dendrites of MHb neurons are located (Fig. 2D).
The effects of KCTD8/12/16 KO on the expression of Cav2.3, ChAT, and GABAB2 in the axonal terminals of MHb neurons. A, Images show the immunoreactivity of Cav2.3 in the anterior and posterior brain sections of the IPN of WT mice (top) and KCTD8/12/16 KO mice (bottom). Right, Bar plot shows the summary data of the normalized mean pixel intensity of Cav2.3 immunoreactivity (WT: 1.00 ± 0.04, n = 22 slices from 3 mice; 3KO: 1.01 ± 0.03, n = 22 slices from 3 mice). n.s., Not significant. Unpaired t test (with Welch's correction). B, Images and summary plot show the ChAT immunoreactivity within the IPN of WT mice and KCTD8/12/16 KO mice (WT: 1.00 ± 0.06, n = 25 slices from 3 mice; 3KO: 0.91 ± 0.04, n = 23 slices from 3 mice). Same conventions as in A. C, Images and summary plot show significantly lower GABAB2 immunoreactivity within the IPN of KCTD8/12/16 KO mice (WT: 1.00 ± 0.03, n = 25 slices from 5 mice; 3KO: 0.72 ± 0.04, n = 25 slices from 5 mice). ****p < 0.0001; unpaired t test (with Welch's correction). D, Images and summary plot show no difference in GABAB2 immunoreactivity in the soma area of MHb neurons between WT and KCTD8/12/16 KO mice (WT: 1.00 ± 0.03, n = 27 slices from 3 mice; 3KO: 0.97 ± 0.03, n = 29 slices from 3 mice). n.s., Not significant. Unpaired t test (with Welch's correction). E, Raw traces of cell-attached recordings show the action potential firing of MHb neurons before, during, and after baclofen (1 μm) in the presence of a cocktail solution comprising DNQX, picrotoxin, HMT, and Mec. Left, WT mice. Right, KCTD8/12/16 KO mice. F, Group data show that firing rates of MHb neurons are reversibly inhibited by baclofen in WT (n = 9 cells) and KCTD8/12/16 KO (n = 7 cells) mice. *p < 0.05, **p < 0.01, ***p < 0.001; paired t test for difference between basal and baclofen and between baclofen and wash. Two-way ANOVA for the difference between WT and KCTD8/12/16 KO mice. Data are presented as means ± SEM (Table 1). Scale bars, A–D, 100 μm.
Next, we conducted cell-attached recordings from MHb neurons to test whether knocking out KCTD8/12/16 influenced the function of GABAB receptors in the somata of MHb neurons. Baclofen significantly reduced the firing rates of the MHb neurons in both WT and KCTD8/12/16 KO mice (Fig. 2E,F). There was no difference in the baseline firing rates and the effects of GABAB-mediated inhibition between WT and KCTD8/12/16 KO mice (Fig. 2F), indicating that KCTD8, 12, and 16 do not have an obvious effect on GABAB-mediated inhibitory responses in the somata of MHb neurons.
These results thus suggested that certain isoforms of KCTD8/12/16, either functioning separately or in combination, regulate the expression of GABAB receptors in the axonal terminals but not somata/dendrites of habenula cholinergic neurons.
KCTD8 and 12 are important for presynaptic GABAB expression and signaling
Given the physiological and neurochemistry phenotypes of the triple knock-out mice, we asked which of these three KCTD proteins might be involved in the axonal expression and presynaptic excitation of GABAB receptors. We examined the expression of KCTD isoforms in the cholinergic pathway from the MHb to the IPN. Immunohistochemistry revealed strong KCTD8 expression and moderate KCTD12 expression in the somata and axonal terminals of ChAT-expressing habenula cells (Fig. 3A–D). The immunoreactivity was absent in the KCTD8/12/16 triple KO mice (Fig. 3E–G), which confirmed the validity of our antibodies. These observations are also consistent with the mRNA in situ hybridization data in the Allen Brain Atlas (Fig. 3H). The KTCD16 antibody was not appropriate for immunohistochemistry, but KCTD16 mRNA level was very low in the MHb according to the Allen Brain Atlas (Fig. 3H). Altogether, these results demonstrated that KCTD8 and KCTD12 rather than KCTD16 were expressed in habenula cholinergic neurons.
Distributions of KCTDs in the habenulo–interpeduncular cholinergic pathway. A, B, Images show the colocalizations of KCTD8 (A) or KCTD12 (B) with ChAT in the IPN of a WT mouse. Top, The immunoreactivities of KCTD8 or KCTD12 (magenta) and ChAT (green). Bottom, zoomed-in views of the dashed rectangular areas. C, D, Images show the distribution of KCTD8 (C) or KCTD12 (D) in the MHb of a WT mouse. Top, The immunoreactivities of KCTD8 (magenta) and ChAT (green), Bottom, Zoomed-in view of the dashed rectangular area. E, F, Images show the immunoreactivities of KCTD8 (magenta, E), KCTD12 (magenta, F) and ChAT (green) in the IPN of a KCTD8/12/16 KO mice. G, Immunoblotting show the presence of KCTD8, 12, and 16 in the IPN of WT and KCTD8/12/16 KO mice. H, Images from Allen Brain Atlas show the mRNA expression pattern of KCTD8, 12, and 16 in the MHb and the IPN. Scale bars: 100 μm; A–D, top and bottom, 20 μm; E, F, 100 μm; and H, 400 μm.
We tested the isoform-specific contribution of KCTDs by analyzing the immunoreactivities of GABAB2 receptor subunit in the cholinergic terminals within the IPN of KCTD8 KO, KCTD12 KO, and KCTD16 KO mice. We observed no significance difference in the GABAB2 immunoreactivity of the axonal terminals within the IPN between the WT mice and the single knock-out mice, but a significant decrease of the GABAB2 immunoreactivity in KCTD8/12 double KO mice (Fig. 4A,B).
Knocking out both KCTD8 and KCTD12 reduces the axonal expression and presynaptic excitation of GABAB receptors in habenula cholinergic neurons. A, Images show the immunoreactivity of GABAB2 receptors in the anterior, center, and posterior sections of the IPN of WT, KCTD8 KO, KCTD12 KO, KCTD16 KO, and KCTD8/12 double KO mice. Scale bars: 100 μm. B, Summary data show the mean intensity of GABAB2 immunoreactivity in the IPN of WT (1.00 ± 0.03, n = 30 slices from 6 mice), KCTD8/12 KO (0.78 ± 0.02, n = 26 slices from 4 mice), KCTD8 KO (0.98 ± 0.04, n = 20 slices from 3 mice), KCTD12 KO (0.89 ± 0.04, n = 16 slices from 3 mice), and KCTD16 KO (0.90 ± 0.03, n = 20 slices from 3 mice) mice. Data are normalized to the mean pixel intensity of WT mice within the same batch. n.s., Not significant. ****p < 0.0001; one-way ANOVA and Sidak's post hoc multiple comparisons test. C, Summary data show the peak amplitudes of evoked EPSCs before (1), during (2), and after (3) baclofen treatment in the presence of a cocktail solution comprising picrotoxin, HMT, and Mec from IPN neurons of ChAT-ChR2 (IBasal = 112.9 ± 44.7 pA; IBaclofen = 1152.6 ± 177.1 pA; IWash = 94.9 ± 43.9 pA; n = 13 cells from 5 mice), KCTD8/12 KO::ChAT-ChR2 (IBasal = 104.4 ± 37.4 pA; IBaclofen = 455.7 ± 128.3 pA; IWash = 66.2 ± 27.2 pA; n = 7 cells from 3 mice), KCTD8 KO::ChAT-ChR2 (IBasal = 124.9 ± 23.9 pA; IBaclofen = 738.8 ± 156.9 pA; IWash = 104.3 ± 25.9 pA; n = 7 cells from 3 mice), KCTD12 KO::ChAT-ChR2 (IBasal = 80.6 ± 30.2 pA; IBaclofen = 723.9 ± 101.7 pA; IWash = 98.9 ± 28.4 pA; n = 9 cells from 3 mice), and KCTD16 KO::ChAT-ChR2 (IBasal = 133.8 ± 74.3 pA; IBaclofen = 957.2 ± 284.8 pA; IWash = 175.1 ± 99.6 pA; n = 8 cells from 3 mice) mice. n.s., Not significant. p > 0.05; *p < 0.05; two-way ANOVA and Sidak's post hoc multiple comparisons test. Data are presented as means ± SEM (Table 1).
We crossed various KCTD KO genotypes mice with ChAT-ChR2-EYFP mice and performed whole-cell recordings from IPN neurons to examine any potential changes in the baclofen-induced potentiation of EPSCs. In accordance with the reduced axonal GABAB receptors expression in KCTD8/12 KO mice, we observed impaired baclofen-induced potentiation of EPSCs (60.5% reduction; Fig. 4C). As expected, single KCTD KO showed no difference in baclofen-induced potentiation of EPSCs with ChAT-ChR2 mice (Fig. 4C). Altogether, these experiments thus revealed that both KCTD8 and KCTD12 are important for facilitating the expression and presynaptic excitation of GABAB receptors in the axonal terminals of habenula cholinergic neurons.
The phenotypes of knocking out both KCTD8 and KCTD12 suggest that (1) both isoforms critically contribute to presynaptic signaling possibly by forming KCTD hetero-oligomers, or (2) either isoform is dispensable and the total expression levels of the two isoforms are important. We tested the two hypotheses by overexpressing a single KCTD in the habenula neurons of KCTD8/12/16 KO mice. We infused AAV to express mCherry, KCTD8-mCherry, or KCTD12-mCherry into the MHb of the triple KO mice and examined the axonal expression and presynaptic excitation of GABAB receptors. AAV transduction produced strong expression of KCTD8 or KCTD12 (Fig. 5A,C). Compared with mCherry-expression controls, overexpressing either KCTD8 or KCTD12 significantly increased the axonal expression of GABAB receptors (Fig. 5B,D–F) and baclofen-induced potentiation of EPSCs (Fig. 5G). Therefore, overexpressing either KCTD8 or KCTD12 alone is sufficient to facilitate the presynaptic expression and signaling of GABAB receptors in habenula cholinergic neurons.
Overexpressing KCTD8 or KCTD12 in the MHb neurons of KCTD8/12/16 KO mice rescues axonal GABAB expression and presynaptic excitation. A, B, Example images show AAV transduction of AAV-mCherry (A, top) and KCTD8-P2A-mCherry (A, bottom) in MHb neurons and expression of mCherry in the axonal terminals within the IPN of KCTD8/12/16 KO mice. AAV-KCTD8-P2A-mCherry but not AAV-mCherry in MHb neurons of KCTD8/12/16 KO mice expressed KCTD8 (A) and increased the expression of GABAB2 in the axonal terminals within the IPN (B). C, D, The effect of overexpressing KCTD12 in habenula neurons on GABAB2 expression within the IPN of KCTD8/12/16 KO mice. Same conventions as in A, B. E, F, Summary plots show the effect of overexpressing KCTD8 (mCherry: 1.00 ± 0.05, n = 14 slices; KCTD8: 1.41 ± 0.04, n = 22 slices; E) or KCTD12 (mCherry: 1.00 ± 0.02, n = 21 slices; KCTD12: 1.19 ± 0.04, n = 19 slices; F) on the mean intensity of GABAB2 immunoreactivity in the IPN. ***p < 0.001, ****p < 0.0001; unpaired t test (with Welch's correction). G, Summary data show the effect of overexpressing mCherry (IBasal = 45.6 ± 8.9 pA; IBaclofen = 567.3 ± 113.3 pA; n = 17 cells from 4 mice), KCTD8-P2A-mCherry (IBasal = 65.6 ± 17.0 pA; IBaclofen = 1179.5 ± 90.9 pA; n = 7 cells from 3 mice), and KCTD12-P2A-mCherry (IBasal = 87.2 ± 14.6 pA; IBaclofen = 1147.3 ± 239.4 pA; n = 9 cells from 5 mice) on basal and baclofen-potentiated, light-evoked EPSCs in the IPN neurons of KCTD8/12/16 KO::ChAT-ChR2 mice. *p < 0.05; two-way ANOVA and Sidak's post hoc multiple comparisons test. Data are presented as means ± SEM (Table 1). Scale bars: 100 μm.
KCTDs regulate neuronal responses to aversion-predicting cues
Given that GABAB-mediated presynaptic excitation in the MHb–IPN pathway regulates aversive memory expression (Soria-Gómez et al., 2015; Zhang et al., 2016), we examined the involvement of KCTDs in the cued-aversion response. We trained mice with an aversive pavlovian conditioning paradigm by presenting a 2 s auditory tone [conditioned stimulus (CS)] with the delayed (1 s) delivery of a brief footshock [0.5 s; unconditioned stimulus (US); Lu et al., 2020]. We expressed GCaMP6m in the IPN neurons—the downstream of habenula cholinergic neurons—and implanted an optical fiber with its tip in the IPN for fiber photometry of GCaMP fluorescence changes during the conditioning sessions (Fig. 6A,B).
IPN neurons in KCTD8/12/16 triple KO mice display enhanced Ca2+ signal responses to aversion-predicting cues. A, Expression of GCaMP6m and the placement of recording optical fiber (top), and schematic drawings of the optical fiber tips (bottom) in anterior and center and posterior sections of the IPN, corresponding approximately to bregma 3.4, 3.52, and 3.64 mm, respectively. Scale bars: 100 μm. B, Schematic diagram of the experimental design for the cued-aversion conditioning and extinction sessions. The conditioning sessions consisted of the deliveries of 30 discrete tone-footshock pairs, whereas the extinction sessions consisted of 30 tone deliveries. C, Trial-by-trial heatmaps and the peri-event plot of average Ca2+ signals illustrate example neuronal responses in cued-aversion conditioning sessions and extinction sessions in WT and KCTD8/12/16 KO mice. The color code for the plots of Ca2+ signals, light green for trials 1–2, green for trials 3–5, and red for trials 26–30. D, Averaged Ca2+ signals of WT and KCTD8/12/16 KO mice during the cued-aversion conditioning and extinction sessions. Shaded area indicates SEM. E, AUC of averaged Ca2+ signal responses (3–5 s) to footshocks in WT and KCTD8/12/16 KO mice. Each data point represents the average of five consecutive trials, n.s., Not significant. ****p < 0.0001; two-way ANOVA and Sidak's post hoc multiple comparisons test for the difference between the first data point and those of the following trials (n.s., no differences across all trials in WT or KCTD8/12/16 KO mice), and for the difference of AUC between WT and KCTD8/12/16 KO mice. F, AUC of averaged Ca2+ signal responses (0–3 s) to the auditory tones in WT and KCTD8/12/16 KO mice. Each data point represents the average of five consecutive trials during cued-fear conditioning sessions and the extinction sessions on day 1–3, n.s., Not significant. *p < 0.05, ***p < 0.001; two-way ANOVA for the difference of AUC between WT and KCTD8/12/16 KO mice. Two-way ANOVA for the difference of AUC between extinction day 1 and extinction day 2, 3 in WT mice (*p < 0.05 between day 1 and day 2; **p < 0.01 between day 1 and day 3). Two-way ANOVA for the difference of AUC between extinction day 1 and extinction day 2, 3 in KCTD8/12/16 KO mice (n.s., p > 0.05 between day 1 and day 2, ***p < 0.001 between day 1 and day 3). G, AUC of averaged locomotion (0–3 s) during cued-fear conditioning sessions and the extinction sessions on day 1–3 in WT and KCTD8/12/16 KO mice; *p < 0.05, ****p < 0.001. Two-way ANOVA for the difference of AUC between WT and KCTD8/12/16 KO mice. H, Summary graphs depict the total locomotion in 10 min before fear condition without cue in the WT and KCTD8/12 KO mice; *p < 0.05, unpaired t test (with Welch's correction). Data are presented as means ± SEM (Table 1).
Initially, the US but not the CS induced Ca2+ transients at the population level in both WT and KCTD8/12/16 KO mice. Within trials 3–5, the CS began to induce an increase in Ca2+ signal intensity (Fig. 6C,D), whereas the US-evoked Ca2+ signals remained unchanged (Fig. 6E). As the conditioning developed, the CS elicited an increasing trend in the intensity of Ca2+ signals, and the CS-evoked Ca2+ signals lasted through the delay period, reaching a peak following delivery of the US (Fig. 6C,D). These observations support that cued-aversion conditioning rapidly shaped the response of IPN neurons. Compared with WT mice, KCTD8/12/16 KO mice responded to the CS and US with stronger Ca2+ signals during aversion conditioning sessions, suggesting a high sensitivity to aversive stimuli.
We tested whether KCTDs contributed to the expression and extinction of cued-aversion memory in the IPN. On the extinction days, the mice were introduced into a new test chamber with the same CS but omitting the US. The CS-evoked Ca2+ signal intensities were strong on extinction day 1 and then gradually decreased during the subsequent extinction days; the intensities were reduced significantly on extinction day 2 in WT mice, whereas on extinction day 3 in KCTD8/12/16 KO mice (Fig. 6C,D,F). We monitored the locomotor activity of the mice. In the beginning trials, mice exhibited lower locomotion following the CS, indicating freezing levels during cued-aversive extinction sessions. After 10 trials of extinction, their locomotor activity during the CS period became higher than the initial level (Fig. 6G). These observations indicate the correlation of IPN neuronal activities with physiological freezing during extinction sessions. Thus, the fiber photometry recording suggests that KCTD8/12/16 KO mice displayed enhanced neuronal responses to aversion-predicting cues in IPN neurons. Compared with wild-type mice, KCTD8/12/16 triple KO mice showed hyperlocomotion (Fig. 6H), and exhibited lower locomotive activity during CS delivery in the cued-aversion conditioning and extinction sessions (Fig. 6G), which suggested changes in locomotor control and behavioral sensitivity to the formation and expression of cued-aversion memory.
KCTDs regulate baclofen-mediated fear extinction in the habenulo–interpeduncular pathway
There were some potential limitations for our data shown in Figure 6. First, we used the brief cue–footshock paradigm (2 s cue with 1 s delay before footshock), which differs from the standard fear conditioning paradigm that measures freezing behavior following long cues (20 s). Second, we recorded Ca2+ signals from the triple knock-out mice, whereas only KCTD8 and 12 are important for axonal expression of GABAB receptors in habenula cholinergic neurons. Third, knocking out KCTD16 alone impairs the extinction of cued aversion memory (Cathomas et al., 2017). The hyperlocomotor phenotypes of the KCTD8/12/16 triple knock-out mice might further confound the behavioral roles of KCTDs in the habenulo–interpeduncular pathway.
Considering that GABAB activity facilitates fear extinction in the MHb–IPN pathway (Zhang et al., 2016) and mainly KCTD8 and KCTD12 are expressed in MHb neurons, we examined the behavioral phenotypes of KCTD8/12 double knock-out mice using the standard fear conditioning and extinction paradigm (10 trials of 20 s cue with or without footshock for conditioning and extinction, respectively; Fig. 7A). Baclofen (15 pmol) or ACSF control was infused into the IPN through an implanted cannula (Fig. 7B). Confirming that activating GABAB receptors in the MHb–IPN pathway facilitates fear extinction (Zhang et al., 2016), baclofen infusion into the IPN significantly reduced the freezing time of WT mice in response to the cue during the extinction sessions on the 3 consecutive days (Fig. 7C,D). In contrast, baclofen infusions into the IPN of KCTD8/12 double knock-out mice only mildly affected the freezing behavior during the extinction session and did not significantly change the overall freezing time (Fig. 7C,D). ACSF control infusions did not reveal a strong difference between WT mice and the mutant mice both during the condition sessions and the extinction sessions (Fig. 7C,D). In addition, the KCTD8/12 double knock-out mice exhibited similar locomotor activity pattern at rest (Fig. 7E). These results thus suggested that KCTDs regulate the activity of GABAB receptors in the habenulo–interpeduncular pathway to facilitate fear extinction rather than general locomotor activity.
Knocking out KCTD8/12 reduces the effect of baclofen on facilitating fear extinction. A, Experimental timeline (left) and the behavioral paradigms (right) for fear conditioning and fear extinction on the following 3 consecutive days. Mice were conditioned or tested for 10 trials in each session. B, Left, Image shows the cannula placement and the extent of the Texas Red dextran amines infusion following the behavioral tests. Scale bars: 200 μm. Right, Images show the drug injection sites in anterior and posterior sections of the IPN, corresponding to 3.52 and 3.64 mm from bregma, respectively. C, Freezing responses across trials during cued-fear conditioning and extinction sessions of WT mice (left) and KCTD8/12 KO mice (right), with the pretreatment of ACSF control (black) or baclofen (red; 15 pmol) during the extinction sessions. n.s., Not significant. *p < 0.05, ***p < 0.001, ****p < 0.0001; two-way ANOVA for the difference of AUC between ACSF and Baclofen. D, Total freezing durations of WT mice (left) and KCTD8/12 KO mice (right) in each behavioral session. n.s., Not significant. ****p < 0.0001; one-way ANOVA for the difference of AUC between ACSF and Baclofen in the WT and KCTD8/12 KO mice groups. E, Summary graphs depict the total locomotion in 3 min before fear condition without cue in the WT and KCTD8/12 KO mice. n.s., Not significant. Unpaired t test (with Welch's correction). Data are presented as means ± SEM (Table 1).
Discussion
The richly expressed GABAB receptors represent the principal inhibitory neurotransmitter receptors (Nishikawa et al., 1997; Couve et al., 2000; Chalifoux and Carter, 2011; Benke et al., 2012; Gassmann and Bettler, 2012; Pin and Bettler, 2016; Fritzius and Bettler, 2020). In the MHb–IPN pathway, GABAB receptor is first reported to mediate presynaptic excitation via Cav2.3 (Zhang et al., 2016). The exact molecular mechanism of this presynaptic excitation remain to be dissected. Several KCTD isoforms assemble with native GABAB receptors (Schwenk et al., 2010, 2016), but their roles in mediating GABAB receptors–mediated presynaptic expression and excitation remain unclear. By analyzing the effect of knocking out KCTD8, 12, and 16 separately or in various combinations, in this study we find that KCTD8 and KCTD12 in habenula cholinergic neurons are important for facilitating the axonal expression and presynaptic excitation of GABAB receptors. We further show that the KCTDs in the habenulo–interpeduncular pathway are important for regulating physiological responses during aversion memory formation and expression.
Our findings expand our understanding of functions of KCTD-GABAB interactions in adult brains. KCTDs tightly assemble with GABAB receptors to regulate the activation and desensitization of their effectors, including the G-protein-coupled inwardly rectifying potassium channels, the opening of which inhibits neurons. In the habenulo–interpeduncular pathway, GABAB receptors mediate the presynaptic excitation via the R-type calcium channel Cav2.3 (Zhang et al., 2016). A recent study reveals that KCTD8 KO, KCTD12b KO, and KCTD8/12b double KO do not affect GABAB receptors–mediated presynaptic excitation, whereas KCTD8 and KCTD12b modulate Cav2.3-mediated release probability in the MHb–IPN pathway in a GABAB receptors–independent manner (Bhandari et al., 2021). Here, we characterized the phenotypes of single, double, and triple knock-out of KCTD8, 12, and 16. Consistently, single knock-out of KCTD8 does not affect the presynaptic excitation of GABAB receptors. Rather, knocking out both KCTD8 and KCTD12 is required to produce a strong decrease in the potentiation effect of GABAB receptors on neurotransmitter release and presynaptic Ca2+ entry. Overexpression experiments indicate that either isoform sufficiently rescues the knock-out phenotypes. Thus, our results, together with those by Bhandari et al. (2021), illustrate isoform-specific roles of KCTDs in synaptic transmission via interacting with GABAB receptors in the brain.
Intriguingly, GABAB receptors mediate excitation in the axonal terminals but inhibition in the somata/dendrites of habenula cholinergic neurons. We show that KCTD8 and KCTD12 together promote GABAB expression and its excitatory effects in presynaptic terminals but are indispensable for GABAB expression and its inhibitory effects in the somata. It has been reported that KCTD12 promotes receptor expression at the cell surface by reducing GABAB receptors internalization in COS-1 cells and cultured hippocampal neurons (Bartoi et al., 2010; Ivankova et al., 2013). Moreover, transfection of KCTD12 into cultured hippocampal neurons increases the amount of GABAB receptors primarily in axonal terminals but not in dendrites (Bartoi et al., 2010). This study reveals that both KCTD8 and KCTD12 contribute to GABAB expression and signaling in the axonal terminals but not somata/dendrites, thus suggesting the presence of two distinct signaling pathways in MHb neurons, resulting in opposite function of GABAB receptors in the axonal terminals and somata/dendrites. KCTD8 and12 are integral constituents of GABAB receptors (Schwenk et al., 2010), which suggests that they may have a site-specific impact on the transport or surface stability of GABAB receptors. Of note, knocking out both KCTD8 and KCTD12 reduces but does not abolish presynaptic excitation. This suggests the presence of a yet to be identified non-KCTD signaling component that switches the presynaptic GABAB receptors in habenula cholinergic neurons from the common inhibition mode to the unique excitation mode.
The requirement of knocking out both KCTD8 and KCTD12 isoforms to reduce axonal GABAB receptors expression indicates a compensatory effect of KCTD8 and KCTD12 in the MHb–IPN pathway. Bhandari et al. (2021) report that knocking out KCTD12b induces a compensatory increase of KCTD8 at the active zone of the MHb terminals. It is thus possible that the genetic deficiency of a given isoform in the single KCTD KO mice might be compensated by other KCTD isoforms of similar functions. Indeed, the N-terminal regions of KCTD8 and 12 are homologous to the BTB (BR-C, ttk and bab) domain which has adapted to several different modes of self-association (Stogios et al., 2005). This is further supported by our observation that overexpressing either KCTD8 or KCTD12 was sufficient to rescue the axonal expression and presynaptic excitation of GABAB receptors in KCTD8/12/16 KO mice. Therefore, although different KCTD isoforms may have distinct molecular features, the redundancy among various isoforms may require simultaneous knock-out to reveal their functional significance.
Our fiber photometry of behaving mice reveals the IPN neuronal response to aversion memory. Lesions of the habenulo–interpeduncular pathway impair the performance of the avoidance response and the consolidation and retrieval of aversive memories (Thompson, 1960; Zhao-Shea et al., 2015; McLaughlin et al., 2017; Vincenz et al., 2017; Khatami et al., 2018). Using fiber photometry of Ca2+ signals in the IPN (Huang et al., 2019), we provide the first demonstration that cued-aversion conditioning can rapidly shape the response of IPN neurons, forming the physiological correlate of aversive memories. In KCTD 8/12/16 triple knock-out mice, IPN neurons exhibit stronger Ca2+ signals to both the cue and the footshock. This observation seems to be the opposite of what would be expected from the finding that the triple KO mice exhibit a decrease in neurotransmitter release from the habenula to the IPN. This might be reconciled by the fact that a majority of IPN neurons, including interneurons, are GABAergic (Zhao-Shea et al., 2013). The extensive GABAergic local circuits in the IPN might form a disinhibition mechanism that transforms a decrease in glutamatergic signal from the MHb to an enhancement in output from the IPN neurons of the triple knock-out mice. Future cell type-specific recordings and manipulations would help testing this scheme of microcircuitry in the IPN.
Our behavioral assays also indicate that KCTDs contribute to the regulation of aversion memory in the MHb–IPN pathway. In accordance with impaired aversion extinction in GABAB conditional knock-out in cholinergic neurons (Zhang et al., 2016), here we show that knocking out KCTD8 and KCTD12 reduces the effect of intra-IPN baclofen infusion on facilitating fear extinction. Unlike GABAB knock-out mice (Zhang et al., 2016), the double knock-out mice do not exhibit a decrease in fear extinction in the control condition, possibly because the genetic deficiency of KCTD8/12 reduces but does not abolish GABAB presynaptic excitation. In addition, multiple GABAB-associated KCTDs, such as KCTD12 and KCTD16, are expressed in numerous brain areas and regulate aversive memory (Cathomas et al., 2015, 2017). Therefore, although our physiological and behavioral assays indicate an important role of KCTDs in regulating neurotransmission within the habenulo–interpeduncular pathway and its associated behavioral functions in cued-aversion memory, KCTD8 and KCTD12 outside the habenulo–interpeduncular pathway may also play a role in regulating aversive memory processes. Conditional knock-out of both KTCD8 and KCTD12 precisely in the habenula cholinergic neurons of adult mice would help dissecting the precise behavioral functions of these two KCTD isoforms in the MHb.
Altogether, this study reveals that KCTD8/12 auxiliary subunits modulate the expression and function of GABAB receptors in habenula cholinergic neurons, thereby affecting aversion memory processing in adult mice. Malfunctions in GABAB receptors, KCTDs, and their interactions are involved in several neuropsychiatric disorders (Gassmann and Bettler, 2012; Cathomas et al., 2015, 2017). For example, KCTD8 is associated with brain size and modulates the adverse effects of smoking during pregnancy on brain development (Paus et al., 2012). The KCTD12 gene expression is associated with bipolar 1 disorder (Lee et al., 2011), depressive-like state (Sibille et al., 2009; Surget et al., 2009), and schizophrenia (Benes, 2010), whereas KCTD16 is related to congenital partial epilepsy syndrome (Angelicheva et al., 2009). Given that the habenulo–interpeduncular pathway and the GABAB signaling in this pathway regulates aversion-memory-associated behaviors in animals, our results suggest that regulating KCTD8/12-associated signaling in this neural pathway may have relevance to therapeutic interventions in related mental disorders.
Footnotes
This work was supported by the Beijing Municipal Government. We thank Bernhard Bettler (University of Basel) for the KCTD8/12/16 KO mice, Bin Li (National Institute of Biological Sciences, Beijing) for advice on statistical tests, and J. H. Snyder and other Luo lab members for comments.
The authors declare no competing financial interests.
- Correspondence should be addressed to Minmim Luo at luominmin{at}nibs.ac.cn