Skip to main content

Data for: The cerebellum regulates fear extinction through thalamo-prefrontal cortex interactions in male mice


Frontera, Jimena et al. (2023), Data for: The cerebellum regulates fear extinction through thalamo-prefrontal cortex interactions in male mice, Dryad, Dataset,


Fear extinction is a form of inhibitory learning that suppresses the expression of aversive memories and plays a key role in the recovery of anxiety and trauma-related disorders. Here, using male mice, we identify a cerebello-thalamo-cortical pathway regulating fear extinction. The cerebellar fastigial nucleus (FN) projects to the lateral subregion of the mediodorsal thalamic nucleus (MD), which is reciprocally connected with the dorsomedial prefrontal cortex (dmPFC). The inhibition of FN inputs to MD in male mice impairs fear extinction in animals with high fear responses and increases the bursting of MD neurons, a firing pattern known to prevent extinction learning. Indeed, this MD bursting is followed by high levels of the dmPFC 4 Hz oscillations causally associated with fear responses during fear extinction, and the inhibition of FN-MD neurons increases the coherence of MD bursts and oscillations with dmPFC 4 Hz oscillations. Overall, these findings reveal a regulation of fear-related thalamo-cortical dynamics by the cerebellum and its contribution to fear extinction.


1. Animals

Adult male C57BL/6N mice (Charles River, France,MSR Cat# CRL_27,RRID:IMSR_CRL:27 ) were housed in groups of 4 mice per cage, with free access to food and water, maintained at 21–22°C, with a 12 h : 12 h light/dark cycle. Adult male wild-type mice 8–12 weeks of age were used for all the experiments. Male and female mice are equally suitable for fear conditioning analysis, but male mice are bigger and have therefore more strength to carry the implant and preamplifier in electrophysiological experiments. Animal care and experimental procedures followed the European Community Council Directives (authorization number APAFIS#1334-2015070818367911 v3 and APAFIS #29793-202102121752192).

2. Stereotactic surgeries

Surgeries for tracing, chemogenetic, optogenetic, and electrophysiological experiments were performed in 7–8 weeks-old mice, placed in a stereotaxic frame (Kopf Instruments) after buprenorphine administration, and anaesthetized with isoflurane 2% along the whole procedure. In addition, a local analgesia was administered subcutaneously above the skull with 0.02% of lidocaine. The body temperature was monitored and maintained at 36°C with a heating pad and a rectal thermometer.

For viral injections and electrodes and optic fiber implantations, the stereotaxic coordinates used for FN (AP: -6.37, ML: ±0.70, DV: -2.12), MD (AP: -1.12, ML: ±0.65, DV: -3.15) and dmPFC (AP: +2.34, ML: ±0.25, DV: -1.5) were taken relative to the bregma, and the depth was considered from the brain surface. Glass capillaries used for viral infusions were kept at the injection site for 5 minutes after viral infusion before being slowly withdrawn. 

3. Neuroanatomical tracing and histology

For neuroanatomical tracing, a group of mice were injected with 100 nl of anterograde AAV1-CB7-Cl-mCherry-WPRE-rBG (Upenn Vector) in the FN (n = 6), 100 nl CTB-alexa 488 (Invitrogen) in the dmPFC (n=6), or retrograde AAV-Syn-eGFP (Addgene) in the MD (n = 6). For trans-synaptic labelling, a group of mice was injected with 75 nl of AAV9-CMV-PI-Cre-rBG (Addgene) in the FN and 100 nl of AAV1-CAG-Flex-tdTomato-WPRE-bGH in the MD (n = 3), and MD sections were used to Calbindin immunostaining.

Mice were deeply anesthetized with ketamine 80 mg kg-1 and xylazine 10 mg kg-1 i.p., and transcardially perfused with formalin (Sigma) 3–4 weeks after the surgery to assess the viral expression. Brains were dissected and kept overnight in formalin at 4°C and then placed in PBS solution until slicing. Coronal brain sections of 50 or 90 μm were made using a vibratome (Leica VT 1000S), then dried and mounted with Mowiol (Sigma) or Fluoroshield with DAPI medium (Sigma). The slices were analyzed and imaged using a confocal microscope (Leica TCS Sp8).

The same procedure was used to verify electrode and optic fiber implantation sites and viral expressions for each experiment, mice were perfused as detailed above, and mice with no viral expression or misplacement of the viral infusion, optic fiber or electrode were excluded from the analysis. Electrode placements are summarized in Supplementary Figure 6.

4. Immunohistochemistry

For the immunostaining, 50 µm slices containing MD were washed with PBS-0.3% Triton X-100, and blockage of nonspecific sites was assessed by 2 h of incubation with 3% normal donkey serum (NDS). Sections were then incubated in a solution containing monoclonal mouse anti-Calbindin-D-28K (1:300, Sigma, C9848) in PBS-0.3% Triton X-100 with 1.5% NDS at 4°C for 24 h. After first antibody incubation, slices were rinsed and sections were incubated at room temperature for 2 h with secondary antibody, donkey anti-mouse IgG conjugated to Alexa Fluor 488 (1:300, Invitrogen). Slices were mounted with Mowiol and analyzed with the confocal microscope.

5. Fear conditioning and extinction protocol

Previous to behavioral experiments, mice were handled daily during one week. Fear conditioning and extinction protocol were performed in two different chamber configurations (Context A or B), placed inside of a sound-attenuating box (Ugo Basile). The fear conditioning was performed in Context A, which consisted of a rectangular-shaped Plexiglas chamber, 17x17x25 cm (L x W x H) dimension, with a grid floor and black and white checker walls, scented with peppermint soup. On fear conditioning day (day 1), mice were left for 5 min to habituate to context A, then they were exposed to 5 presentations of a tone (30 s, 80 dB, 2.7 kHz, representing the CS), that co-terminated with a mild foot-shock (0.5 sec, 0,4 mA, representing the US), and with an interval between CS-US presentations of 120 sec. During the following 3 days, mice underwent the fear extinction training, which consisted in 3 days of extinction sessions (EXT1-3). Extinction sessions were performed in Context B, composed of a yellow cylindrical chamber (17 cm diameter, 25 cm H), grey Plexiglas floor, and vanilla scented.  Each extinction session started with 5 min of habituation period in the novel context B, followed by 25 consecutive un-reinforced 30s-long CS presentations, with an interval between CS of 30 s.

Mice were video-tracked using a high-definition video camera, positioned above the testing chambers. Stimuli administration was controlled by the EthoVision XT 14 software (Noldus Information Technology), which assessed also the freezing behavior during the experiments as inactive periods (the threshold of inactivity was set to not detect the breathing movements as active periods). All mice were returned to their home cage after the experiments. Analysis of freezing levels in each session was performed by comparisons between groups of the first CS and CS5 for FC, and averaged of CS1-CS5 representing “early” EXT, CS10-CS14 as “middle” EXT, and CS21-CS25 as “late” EXT.

6. Anxiety tests

Open field, Elevated plus maze and Dark-light box tests were performed in order to analyze anxiety-like behavior. One week before subjecting the mice to fear conditioning, they were subjected on successive days to anxiety tests, starting with the less anxiogenic, open field test, followed by Elevated plus maze and Dark-light box. Open field test consisted in a 38 cm diameter circular arena (Noldus), under 40-50 lux luminosity. Each mouse was placed in the center of the arena and allowed to explore freely for 10 min. Frequency of entries to the center of the arena, time spent in center, and distance moved in center were measured. Elevated plus maze test was realized using an elevated (52 cm above the floor), plus-shaped apparatus (Noldus) with 2 open arms (36x6 cm L x W) and 2 closed arms (36x6x25 cm L x W x H), under 40–50 lux luminosity in the open arms. Mice were left to freely explore the arena for 5 min, while assessing the frequency of entries in open and closed arms, time spent in open arms, and total distance moved. In the dark-light box test, each mouse was placed in the light zone (luminosity 500–600 lux, 40x20x20 cm L x W x H dimension) of the apparatus (Noldus) and left to freely explore for 5 min, while assessing the frequency of entries in light zone, time spent in light zone, latency to enter in dark zone (0 lux, 20x20x20 cm L x W x H dimension), and total distance moved. Mice were video tracked, and behavior was analyzed using EthoVision XT 14.

7. Hot plate and tail immersion tests

Pain sensitivity was analyzed using hot plate and tail immersion assays performed two weeks after the end of the other experiments. Mice were placed on a hot plate (55°C) (Harvard apparatus), until they jumped or licked their hind paw. The latency to the first reaction was measured. After the hot plate test, tail immersion test was performed. The mice tale tips were immersed in hot water (50°C) and the latency of the tail flick or withdrawal was measured. All experiments were video tracked.

8. Chemogenetics

Specific chemogenetic inhibition of the FN neurons projecting to MD was carried out by bilateral injection of 250 nl of a retrograde virus expressing a cre-recombinase in the MD (CAV2-cre-GFP, from Platforme de Vectorologie de Montpellier) in combination with the infusion of 200 nl of inhibitory cre-dependent DREADD (AAV‐hSyn‐DIO-hM4Di-mCherry, Addgene) bilaterally in FN, adapted from previous works in our team, which demonstrated a strong reduction in cerebellar nuclei firing rate upon injection of CNO (1mg/kg). The behavioral experiments started 10–14 days after the surgery, to ensure the expression of the receptors and the recovery of the mice. Transient inhibition of the FN MD-projecting neurons was assessed by intraperitoneal administration of the clozapine N-oxide (CNO, Tocris Bioscience) dissolved in saline solution (1.25 mg/kg), while a control group received only saline administration, 30 min before the first 2 extinction sessions (EXT1 and EXT2) or 30 min before anxiety or nociceptive tests.

Additional controls were performed to verify the specificity of the pathway inhibition and the DREADD-modulation effect. To corroborate that the CNO does not have an effect per se in our experimental conditions, a batch of mice (named “Sham” mice) underwent the same surgical procedure with the injection of AAV‐hSyn‐mCherry instead of the DREADD vector. During behavioral experiments, the Sham mice received either an intraperitoneal injection of saline or CNO solution (1.25 mg/kg), 30 min before the tests, and freezing levels were analyzed in absence of DREADD expression. On the other hand, to verify that the effect observed under DREADD inhibition was specific of the FN terminals in MD, mice were implanted with 30G cannulas bilaterally in the MD to allow the local infusion of 200 nl of CNO or saline intracranially. During extinction sessions, mice expressing inhibitory DREADD received either sterilized filtered PBS or CNO (0.5 mM) infusion, and another group of Sham mice received CNO (0.5 mM). The intracranial infusions were performed 10 min before the beginning of each session, by using a pump at an infusion speed of 100 nl/min, and a total volume of 250 nl. Freezing levels were analyzed and the positions of the cannulas were corroborated histologically.

9. Electrophysiological recordings

Extracellular recordings were assessed using standard 16 channels electrode interface boards (EIB-16; Neuralynx), to which attached 2 cannulas were attached, one for each recorded region, MD and dmPFC. The electrode bundles consisted of nickel chrome wires (16μm diameter, Coating ¼ Hard PAC, KANTHAL Precision Technology) twisted in groups of 6, gold-plated to 100–400 kΩ (cyanure-free gold solution, Sifco). Two bundles were inserted per cannula (stainless steel, 30 Gauge, Phymep). The recording electrodes were then progressively lowered until they reached the targeted brain structures and the electrode interface boards were cemented to the skull (Dental Parkell Adhesive Resin Cement Super-Bond C&B). Miniature stainless steel screws were implanted on the left parieto-occipital suture, serving as electrical reference and ground. The skin ridges were sutured and mice were allowed to recover in their home cage for at least 10 days.

10. Optogenetics and electrophysiological recordings

The neural activity in the MD and dmPFC was recorded during the optogenetic activation of the contralateral FN neurons. Expression of channelrhodopsin 2 (ChR2) was assessed by injection of an anterograde AAV8-Syn-ChR2-H134R-EYFP (Addgene) in the FN, and an optical was fiber implanted above (0.22 aperture, fixed in a stainless steel ferula, Thor Labs) and contralateral to the recording sites. Electrodes were placed through cannulas in the MD and dmPFC and the EIB were fixed to the skull. The optical stimulations were performed three or four weeks after the surgery (to ensure the ChR2 expression) in freely moving mice. After a 5 min habituation period in an open field (38 cm diameter), the activation of the FN projections was done using light pulses of 100 ms, at 0.25 Hz, 1 mW, for 30 min (with a 2 min break every 10 min block). The choice of light intensity to stimulate FN neurons was based on the calibration performed in our previous work. The activation of MD and dmPFC neurons by optogenetic stimulation of FN neurons was assessed by computing the Peri-Stimulus Time Histogram (5 ms bins) around the optogenetic stimulation. This PSTH was then normalized using a classical Z-score (subtracting by the mean value of the PSTH during the 100 ms before the stimulation and dividing it by its standard deviation). Normalized PSTH reaching an absolute Z-score value superior to 3 during the 100 ms stimulation were considered responsive.

In another group of mice, the specific activation of the FN-MD pathway was assessed by injecting a cre-dependent anterograde adenovirus expressing ChR2 (AAV-Dio-ChR2-EYFP, Addgene) in the FN, together with a retrograde adenovirus expressing cre-recombinase (AAV-Cre-mCherry, Addgene) in the contralateral MD (left side). The implantation and surgical procedures were identical to the ones described above. The FN MD-projecting neurons were stimulated by pulses of 10 ms, at 0.5 Hz and 1 mW, during 10 min. Luminous stimuli were administered and neural signals were recorded using Tucker Davis Technology System 3 acquisition system (25 Hz sampling rate, RZ2, RV2, Tucker-Davis Technologies) and the spike sorting was performed using Matlab scripts. Viral infusions and implant positions were confirmed histologically when the experiment ended. Analysis of the spike-sorted data was performed using Python. The specific activation of MD and dmPFC neurons by optogenetic stimulation of FN MD-projecting neurons was assessed by computing the Peri-Stimulus Time Histogram (5 ms bins) around the optogenetic stimulation. This PSTH was then normalized using a classical Z-score (subtracting by the mean value of the PSTH during the 100 ms before the stimulation and dividing it by its standard deviation). Normalized PSTH reaching an absolute Z-score value superior to 3.5 were considered responsive.

In order to assess the validity of the optostimulation experiments, we performed a control experiment to verify that FN illumination in absence of ChR2 expression does not affect spike firing in the areas recorded. We implanted an optic fiber in the FN together with electrodes in FN, MD and dmPFC, in mice injected with retrograde AAV-GFP (without the expression of ChR2) in the MD, and performed 100 ms illuminations at 0.25 Hz, 1 mW, for 30 min (with a 2 min break every 10 min block), while recording in the FN, MD and dmPFC (Supplementary Fig. 2A-B). There was no variation of firing rate at the population level in the FN, MD and dmPFC (Supplementary Fig. 2C), although we observed 1 neuron in the FN and 3 neurons in the MD reaching the threshold of significant variation of firing rate during the 100 ms of illumination (Supplementary Fig. 2D). However, the transient nature of the mild increase in firing rate suggests a coincidental classification as responsive cells.

Since the wireless transmitter used in the electrophysiological experiments represented a significant hindrance for the movement and was associated with atypical immobile postures (nose down and/or tilted head to rest the preamplifier on the floor or against the wall) which could indifferently reflect freezing or resting, we could not score freezing in these mice.

11. Chemogenetics and electrophysiological recordings

Mice expressing cre-dependent inhibitory DREADDs in FN MD-projecting neurons (see chemogenetics section) were implanted with electrodes in the MD and dmPFC (as described in Electrophysiological recordings section). The neuronal activity of the MD and dmPFC was examined during FC and extinction learning, under chemogenetic inhibition of the FN-MD projections during EXT1 and without manipulation during EXT3. After experiments were concluded, viral expression and electrode positions were confirmed by histology.

12. Quantification and statistical analysis

Behavioral data analysis

For freezing analysis, the data were analyzed by repeated-measure ANOVA computed with R (lme package version 3.1-3), and posthoc tests were performed with the package emmeans (version 1.7.0). During extinction, Early, Middle, and Late values of freezing respectively correspond to the average of the 5 first, 5 middle, and 5 last CS freezing scores. For other behavioral measures, t-test were used (Graph Pad Prism® version 5). All tests used are two-sided (when applicable).

Electrophysiological data analysis

Burst analysis

We quantified the occurrence of bursts in MD, detected using the Robust Gaussian surprise algorithm 1. Spike trains were processed in the following way: Inter-Spike Intervals (ISI) were calculated, then transformed to log (ISI). A central set of ISI was defined as the portion of this distribution lying under [E-1.64*MAD; E+1.64*MAD], where E is the midpoint of top and bottom 100*p percentile of the log (ISI)s with p = 0.05. The central location of this distribution was defined as the median of the central set previously mentioned, and the distribution of log (ISI) was then normalized by subtracting the central location. This normalization process was performed on the entire spike train using a sliding window of half-width 0.2*N/2, where N is the number of spikes in the spike train. The burst-threshold is set as 0.5 percentile of the central distribution. Bursts seeds were defined as normalized ISI being below the burst-threshold. Those seeds were then extended by recursively trying to add the previous and the next normalized ISI; if the addition led to an increase of the associated p-value, the concatenation process was stopped, otherwise, it was continued. Overlapping bursts strings were cleaned by keeping the strings with the lowest associated p-value, thus making them mutually exclusive. Burst occurrence was computed as the number of bursts strings divided by the duration of the recorded spike trains. The average burst firing rate was calculated as the number of spikes contained within a burst divided by the total duration of burst strings in the spike train.

Local field potential, spectral analysis and phase locking

In order to process the LFP, the raw electrophysiological traces sampled at 25 KHz were filtered between 1 Hz and 200 Hz, then downsampled to 1 kHz. All signals were filtered using the filtfilt function from the package scipy, with zero-phase distortion 5th order Butterworth filters. Power Spectral Densities (PSD) were computed using the welch function from the scipy package. The fraction of the PSD representing 4Hz oscillations was defined the integral of the PSD in the 4Hz range (2-6Hz) divided by the integral of the PSD from 1 to 100 Hz. To assess the relationship between LFP in the dmPFC and in the MD, Generalized Partial Directed Coherence analysis was performed on the processed LFP traces using the package spectral_connectivity from Eden Kramer Lab ( For the specific analysis of 4 Hz LFP, the processed LFP traces were filtered in the desired frequency band (2-6 Hz) using the filtfilt function from the scipy package, with zero-phase distortion 5th order Butterworth filters. To study the relationship between the timing of MD neuronal activity and the dmPFC 4 Hz oscillations, Hilbert’s transform was performed using the function hilbert from the package scipy, allowing the extraction of the instantaneous phase and amplitude of envelope estimated at every sample point of the signal. Phases were expressed in radians, with a phase of 0 corresponding to a negative peak in the 4 Hz LFP. Phase locking analysis of bursting activity was performed by considering the first spike of each bursts. Considering spikes occurring in periods associated to weak 4 Hz oscillations would result in the inclusion of uninformative phases in the analysis. So, in order to prevent this bias, only burst occurring in periods of high 4 Hz were considered. Periods of high 4 Hz were defined as periods were the amplitude of envelope was superior to the median amplitude of envelope during the baseline before the Extinction session plus one median absolute deviation (MAD). In other words, we considered the periods where the Robust Z-score of the amplitude of envelope was superior to 1, creating a spike train corresponding to MD bursting during periods of high 4 Hz oscillations in dmPFC. Von Mises distributions fit on phase distributions were performed using the function vonmises from the package scipy, allowing the computation of the parameter Kappa, indicative of the concentration of a circular distribution. Rayleigh tests and circular V-tests with a preferred direction of pi were performed using the package pingouin, respectively using the functions circ_rayleigh and circ_vtest. In order to assess the relationship between MD bursting and the coherence between dmPFC and MD LFP, we used the function spectral_connectivity included in the package mne, For this, LFP snippets of 2000ms centered on the bursts of each individual MD neurons were extracted from the channels displaying the highest fraction of the PSD corresponding to 4 Hz oscillations in the MD and in the dmPFC, and the coherence between MD and dmPFC LFP was calculated using Morlet wavelets of the first order, yielding a time frequency coherence analysis centered on MD bursts.


Fondation pour la Recherche Médicale, Award: DPP20151033983

Fondation pour la Recherche Médicale, Award: EQU202103012770

Agence Nationale de la Recherche, Award: ANR-21-CE37-0025

Agence Nationale de la Recherche, Award: ANR-17-CE37-0009

Agence Nationale de la Recherche, Award: ANR-17-CE16-0019

Agence Nationale de la Recherche, Award: ANR-21-CE16-0017

Agence Nationale de la Recherche, Award: ANR-10-INBS-04

Agence Nationale de la Recherche, Award: ANR-10-LABX-54