Animals

Transgenic R6/1 mice [35] expressing the N-terminal exon-1 fragment of mHtt containing 115 CAG repeats, were used. HdhQ7 WT mice with 7 CAG repeats and HdhQ111 knock-in mice (KI), with targeted insertion of 109 CAG repeats that extends the glutamine segment in murine huntingtin to 111 residues, were also used [36]. Genotypes were determined by polymerase chain reaction (PCR) from ear biopsy. Microchips were implanted under the mice skin providing information about their birth, location, and genotype. All mice were housed in numerical birth order in a room kept at 19–22 °C and 40–60% humidity under a 12:12 light/dark cycle with access to water and food ad libitum. All experiments were conducted exclusively with male mice. WT littermates were used as a control group. Standard animal procedures were approved by the animal experimentation Ethics Committee of the Universitat de Barcelona (274/18) and Generalitat de Catalunya (10/20), in agreement with the Spanish (RD53/2013) and European (2010/63/UE) regulations for the care and use of laboratory animals.

Rabies virus-based monosynaptic tracing

To target the pre-synaptic inputs received by the striatum, we utilized a two-virus system adapted from a previous protocol [37]. Both viruses, the helper lentivirus containing the pBOB-Syn-hisGFP-TVA-rabiesG plasmid (10% dilution from titre: 9 × 107 TU/ml) and the pseudo-typed rabies virus EnvA-ΔG-mCherry (5% dilution from titre: 25 × 106 TU/ml) were provided by Dr. Malin Parmar from Lund University, Sweden. Stereotaxic surgery was performed in 4-week-old and 10-week-old R6/1 male mice and WT littermates. Mice were anesthetized with 3.5% isoflurane in an induction chamber. Then, mice were fixed in the stereotaxic apparatus being anesthetic status maintained with 1.5% isoflurane. A dose of 2 mg/kg of analgesic Metacam® was injected subcutaneously. 1 μl of viral vectors was injected targeting dorsal striatum following coordinates relative to Bregma (anteroposterior and lateral) and from skull (dorsoventral): AP: + 0.5 mm; L: + 1.8 mm and DV: – 2.55 mm were used for the 4-week-old animals. For the 10-week-old animals the coordinates were: AP: + 0.8 mm; L: + 2 mm and DV: – 2.7 mm. Viral vectors were injected with a 5 μl Hamilton Neuros syringe at an infusion rate of 100ηl/min. The needle was left in place for 5 min to ensure complete diffusion of the virus. After 7 days, animals were conducted to the second stereotaxic surgery and then infected with the pseudo-typed rabies virus, using an angle of 45º to avoid undesired infection of cortical neurons due to residual virus in the needle track. The following coordinates relative to Bregma (anteroposterior and lateral) and from skull (dorsoventral): AP: – 0.5 mm; L: + 3.45 mm and DV: – 2 mm were used for the 4-week-old animals. For the 10-week-old animals the coordinates were relative to Bregma (anteroposterior and lateral) and from skull (dorsoventral): AP: – 0.8 mm; L: + 3.65 mm and DV: – 2.35 mm. Seven days later, mice were intracardially perfused with 4% paraformaldehyde. After perfusion, the brain was isolated and post fixed with 4% paraformaldehyde overnight at 4 °C. The day after, brains were transferred to a 30% sucrose solution in PBS with 0.02% Sodium Azide, and kept at 4 °C. Then, 40 μm coronal brain sections were prepared using a microtome (Leica SM2010 R) and stored in a cryoprotectant anti-freeze solution (30% glycerol, 30% ethylene glycol and 15% Tris–HCl) at − 20 °C.

Brain sections were immunostained with antibodies against mCherry and GFP and mounted on glass slides. Slides were scanned with an Olympus BX51 microscope (Olympus, Ballerup, Denmark). Sections were analyzed using the Optical Fractionator technique and stereology software from visiopharm (version 7.0.3.3313). Primary somatosensory and motor cortices (S1-M1) and thalamic neuronal inputs to striatum were evaluated with 40 × objective by counting traced cells (expressing only mCherry) that fell within the different brain areas delineated based on DAPI staining and brain mouse atlas. Striatal starter neurons were evaluated by counting cells infected with both viruses (expressing nuclear GFP and cytoplasmic mCherry) that fell within the striatum. For each animal, the number of cortical and thalamic traced neurons was normalized dividing by the number of striatal starter neurons.

Tissue preparation and immunofluorescence

Mice were euthanized by cervical dislocation. Brains were removed and fixed for 72 h with paraformaldehyde 4% (PFA), and then kept in PBS with 0.02% Sodium Azide at 4 °C until use. Coronal sections (40-μm) of the brain were obtained using a vibratome (Leica VT 1000S) and kept in cryoprotectant anti-freeze solution at − 20 °C until use. Free-floating brain sections were rinsed twice in PBS for min and incubated 2 times for 15 min each with 50 mM NH4Cl to reduce aldehyde-induced tissue autofluorescence. Sections were permeabilized twice for 10 min each with PBS containing 0.5% Triton X-100. Thereafter, sections were blocked with PBS containing 0.3% Triton X-100, 0.2% Sodium Azide and 5% normal donkey serum/goat serum (Pierce Biotechnology, Rockford, IL) for 2 h at room temperature. Then, sections were incubated overnight at 4 °C in the presence of primary antibody in PBS-T: rabbit Foxp2 antibody (1:100, Abcam #ab16046, Cambrige, UK), DARPP-32 (1:500; BD Transduction # 611,520, New Jersey, USA); parvalbumin (1:1000, Swant #PV27, Burgdorf, Switzerland), ChAT (1:500; Chemicon #AB144P, Burlington, USA), VGlut2 (1:1000, Millipore #AB2251-I, Burlington, USA) and PSD-95 (1:250, Cell Signaling #3450S, Danvers, USA). Sections were then washed three times and incubated for 2 h at room temperature with fluorescent secondary antibodies: Cy3 goat anti-rabbit (1:200) and/or AlexaFluor 488 donkey anti-mouse (1:200; both from Jackson ImmunoResearch, West Grove, PA, USA). No signal was detected in control sections incubated in the absence of the primary antibody.

Confocal imaging and analysis

Immunostained tissue sections (40-μm thick) containing ventrobasal thalamus were imaged using a Leica Confocal SP5-II (20× or 40× numerical aperture lens, 5 × digital zoom, 1-Airy unit pinhole). At least two slices per mouse, were analyzed, and up to two representative striatum images were obtained from each slice. Four frames were averaged per z-step throughout the study. Confocal z-stacks were taken at 1024 × 1024-pixel resolution every 2 μm. Foxp2 nuclei integrated optical density was quantified with NIH ImageJ freeware (Wayne Rasband, NIH). Colocalization of VGlut2 and PSD-95 double-positive puncta were quantified as previously described [38].

Western blot

Quantification of protein fraction was performed using the Detergent-Compatible Protein Assay (Bio-Rad, Hercules, CA, USA). Protein extracts (20 μg) were denatured in 62.5 mM Tris–HCl (pH 6.8), 2% Sodium dodecyl sulfate (SDS), 10% glycerol, 140 mM b-mercaptoethanol, and 0.1% bromophenol blue and heated at 100 °C for 5 min. Protein extracts were resolved in denaturing SDS–polyacrylamide gel electrophoresis (SDS-PAGE), with variable polyacrylamide concentration depending on the molecular weight of the protein of interest, at 35 mA/gel over 1 h. The Precision Plus ProteinTM Dual Color ladder (Bio-Rad) was loaded along with the protein samples to properly identify the protein of interest. Afterwards, proteins were transferred to a nitrocellulose membrane (Whatman Schleicher & Schuell, Keene, NH, USA) during 1.5 h at 90 V at 4 °C. Membranes were rinsed with Tris-buffered saline, 0.1% Tween 20 (TBS-T) to remove staining. Non-specific protein binding sites were blocked during 1 h incubation in blocking solution containing 10% non-fat powdered milk in TBS-T (50 mM Tris–HCl, 150 mM NaCl, pH 7.4, 0.05% Tween 20). Membranes were rinsed 3 times during 10 min in TBS-T and immunoblotted overnight at 4 °C with the primary antibody rabbit anti-Foxp2 (1:1000, Abcam #ab16046, Cambrige, UK). Membranes were then rinsed 3 times for 10 min each with TBS-T and incubated with the proper horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature.

Intra-thalamic injections of adeno-associated viruses

Stereotaxic surgery was carried out in 10–12-week-old WT mice by ketamine–xylazine-induced anesthesia. Mice were anesthetized with 3.5% isoflurane in 100% oxygen in an induction chamber. Then, fixed in the stereotaxic apparatus anesthetic status was maintained with 1.5% isoflurane. Mouse head was shaved and cleaned with ethanol. Then iodine and local anesthesia were applied (lidocaine 2.5% and prilocaine 2.5% EMLA®, AstraZeneca), and a dose of 2 mg/kg of analgesic Metacam® was injected subcutaneously. 500ηl of viral vector were injected in each hemisphere targeted bilaterally the thalamus. The following coordinates relative to Bregma (anteroposterior and lateral) and from skull (dorsoventral) were used to target the ventrobasal thalamus: AP: – 1.7 mm; L: ± 1.5 mm and DV: – 3.5 mm. Viral vectors (AAV9-CamKIIa-eGFP-2A-mFoxp2-WPRE (Vector Biolabs, Pennsylvania, USA), AAV5-CamKIIa-eGFP-2A-WPRE (UNC Vector Cre, North Carolina, USA), AAV8-mCherry-U6-mFoxp2-shRNA (#shAAV-259,597, Vector Biolabs, Pennsylvania, USA) and AAV5-mCherry-U6-scrmb-shRNA (#1781, Vector Biolabs, Pennsylvania, USA); titers: 7–8.5 × 1012) were injected with a 5 μl Hamilton syringe at an infusion rate of 100ηl/min. We used the CamKIIa promoter for our AAVs since we had previous very good viral transduction rates [26] and since CamKIIa is highly expressed in the thalamus [39]. The needle was left in place for 5 min to ensure complete diffusion of the AAVs. Mice were return to their home cage after fully recovery. All mice subjected to surgery that survived and showed no ethical and healthy problems (such as head inclination or > 15% of body weight loss) were the ones used for behavioural characterization.

Novel whisker-dependent texture discrimination test

Novel whisker-dependent texture discrimination test was conducted as previously reported [40]. It consisted of two consecutive days of habituation in the open field, and one day for testing divided in two sessions: training and testing session. A gray open-top square arena (40 cm × 40 cm long; with 30 cm high walls) was placed in a room with dim light (20–25 lx). On testing day, for the first session mice were placed in the testing apparatus with two identically smooth-textured objects at the center of the arena. Mice were allowed to explore the objects for 5 min and then removed and held in a cage for 5 min. Before the second session, the two objects in the arena were replaced with a third one identically textured object and a novel object with a different texture. Mice were then placed back into the arena for the second session, the test phase, and allowed to explore for 1 min. The amount of time mice spent actively investigating the objects was recorded. Animal tracking was recorded via a CCD camera. Image data acquisition and analysis were performed automatically using Panlab SMART Video Tracking System (3.0).

Accelerating rotarod

The accelerating rotarod training procedure (ARTP) was conducted as previously described [41]. To evaluate mouse motor learning and performance, mice were tested on the accelerating rotarod over 2 days. Animals were placed on a horizontal rotating rod (30 mm diameter) with an increasing gradually speed (4–40 RPM) over 5 min. Latency to fall was recorded as the time mice spent in the rod before falling. The testing session consisted of 3 trials a day for 2 consecutive days, with a 1 h inter-trial interval.

Golgi and Nissl staining and dendritic spine analysis and volumes estimations

Fresh brain hemispheres were processed following the Golgi–Cox method as described elsewhere [42]. Secondary dendrites from stellate neurons in the layer IV of the somatosensorial cortex and in the ventrobasal thalamus were photographed, with a maximum of two–three dendrites per neuron and from at least 3 slices per animal. Z-stacks from 0.2 μm sections were obtained in bright field at 63 × resolution on a Widefield AF6000 Monochrome Camera Leica Microscope. Images were analyzed with the ImageJ software. Secondary dendritic segments (> 20 μm long) were selected and traced. The total number of spines was obtained using the cell counter tool from ImageJ. Dendritic spine density was obtained after dividing the number of spines by the length of the segment (n° spines/μm). At least 60 dendrites in cortical neurons and 20 dendrites in thalamic neurons per group from at least four mice per genotype were counted. The head diameter and the spine length were measured as previously described [42]. Nissl staining protocol was performed as previously reported [43]. Brain regions (striatum and ventrobasal thalamus) volumes were estimated using the Cavalieri method as previously described [44].

Unbiased stereology

Striatal and thalamic volumes of Cytochrome C Oxidase (see below for further details about the immunohistochemical protocol) and Nissl-stained brain sections were estimated using the Cavalieri method as previously described [44].

Surgical procedures for in vivo electrophysiology

Anesthesia was induced with an intraperitoneal injection of urethane (1.6 g/kg) in 16-week-old WT and R6/1 mice. During recording sessions, the level of anesthesia was monitored by the presence of delta frequency waves (1–4 Hz) of high amplitude (> 50 µV) in the local field potentials (LFP) as well as by the absence of both spontaneous whisker movements and pinch withdrawal reflex. Atropine (0.3 mg/kg), methylprednisolone (30 mg/kg), and mannitol (40 mg/kg) were administered subcutaneously to avoid respiratory secretions and edema. The animal was placed on a water-heated pad (RWD Life Science, China) set at 37 °C to keep body temperature stable, and the head was positioned in a rodent stereotaxic frame (SR-6 M, Narishige, Japan). Local anesthetic (lidocaine 1%) was applied to all skin incisions and a craniotomy was performed using mini-rongeurs over the left hemisphere from − 3.0 mm to + 3.0 mm relative to bregma and + 3.0 mm relative to midline. This broad craniotomy was selected to access a large area of the targeted hemisphere.

Electrical stimulation in vivo

Electrical stimulation in the thalamus, with the aim of evoking responses in the striatum and the cortex, was applied by means of a stainless-steel bipolar electrode (210 μm spacing between tips; FHC Inc., USA), with a constant current isolated stimulator (DS3, Digitimer Ltd., UK) controlled by Spike2 software using a CED Power 1401 interface (Cambridge Electronic Design, UK). The tip of the stimulation electrode was stereotaxically positioned into the ventral posteromedial thalamic nucleus (VPM) following coordinates according to the Gaidi mouse atlas (AP: − 1.9 mm, LM: − 1.7 mm; DV: − 3.4 mm) in WT animals. These coordinates were adapted for R6/1 mice (AP: − 1.7 mm, LM: − 1.6 mm; DV: − 3.3 mm). For each protocol applied, 50 single square pulses (0.3 ms duration) were delivered at 0.1 Hz with intensity currents ranging from 40 to 320 µA (steps of 40 µA) to elicit thalamo-striatal responses and ranging from 20 to 160 µA (steps of 20 µA) to elicit thalamo-cortical responses.

Electrophysiological recordings in vivo

Spontaneous and evoked local field potentials (LFP) were recorded in the striatum and the cortex. First, a tungsten microelectrode (2–4 MΩ; FHC Inc., USA) was lowered stereotaxically into the striatum following the coordinates AP: − 0.9 mm, LM: − 3.3 mm; DV: − 2.25 mm for WT mice and AP: − 0.8 mm, LM: − 3.2 mm; DV: − 2.25 mm for R6/1 mice. The tungsten microelectrode was previously coated with a fluorescent dye (Vybrant® DiI or Vybrant® DiO; Invitrogen) to enable the visualization and location of the tip under the microscope. A total of 500 s of spontaneous activity were recorded prior to the stimulation protocol, which was applied in increasing current intensities. When all the striatal recordings were completed, the electrode was slowly removed from the brain. Then, a 32-channel multielectrode array ([45]; 550 μm spacing between recording points) was positioned in the cortical surface covering the motor and the primary and secondary somatosensory areas. In the cortex, 500 s of spontaneous activity were recorded and then electrical stimulation was applied for all increasing current intensities. For both striatal and cortical recordings, the signal was amplified by 100 and high-pass filtered above 0.1 Hz (Multichannel Systems, GmbH), digitized at 5 kHz and fed into a computer via a digitizer interface (CED 1401 and Spike2 software, Cambridge Electronic Design, UK). Mice were sacrificed with an overdose of anesthetics (see below) immediately after recordings were completed.

Data analysis from in vivo electrophysiological recordings

In the spontaneous activity for both striatal and cortical recordings, the detection of the transitions between silent (Down) and active (Up) states was performed using a z-score normalized multivariate time series, composed by the raw signal (LFP), a logarithmically-scaled estimation of 200–1500 Hz power of the LFP (the Multi-Unit Activity or MUA) [46], and the envelope of the variance of the gamma-filtered LFP [47]. For each LFP signal, a highly processed data time sequence was obtained resulting from the linear combination of the three-time sequences described above, using principal component analysis (PCA) to weigh the contribution of each of them. PCA was applied to the time series and a bimodal distribution originated from the projections over the first principal component, whose peaks corresponded to samples of activity belonging to Up or Down events. This bimodal distribution allowed us to select a threshold that optimally separated the two modes, enabling the creation of a binary signal that contained 1’s in those time samples where logMUA was over the threshold (Up state) and 0’s where below (Down state). A minimum duration of 80 ms for Up and Down state was set to avoid the detection of random signal fluctuations. Every detection was visually examined and validated to be included in the analysis. After Up and Down state detection, mean Up state durations were obtained [48]. The average firing rate for Up and Down states was computed as the mean value of the logMUA for all those time samples belonging to 1 and 0 respectively. The Power Spectral Density (PSD) was computed using Welch’s method from scipy (scipy.org, signal.welch) over the z-scored normalized LFP with a 2-s window and 0.1 Hz bin size.

Evoked potentials elicited by thalamic stimulation were extracted in the striatum and the cortex by calculating the average from responses to 50 stimuli. Given the complexity of the evoked responses and the difficulties in automating the detection of the positive and negative peaks of the generated waveforms, a customized Python pipeline was designed, with some small graphical widgets (ipywidgets) embedded in it. This pipeline allowed a semi-automatic detection of the peaks that was later validated by the experimenter. The mean of the peak latency, peak amplitude and the area under the curve were calculated from the evoked potentials. A baseline normalization was applied to all evoked responses such that the mean of all baselines was set to 0. The detection of the peaks was computed over the trial-averaged evoked responses for all intensities, according to the spatial coordinates of the recording array in the case of the cortical recordings. When performing the detection, the median was also displayed in case some outliers were observed. No significant differences between the mean and the medians were found. The first positive and negative peaks, if any, were detected. The transition times, considered as the points where either the positive or negative responses started to increase or decrease, were detected to compute the overall magnitude of the response (e.g., the area under the curve and peak amplitude). Peaks were only detected if they had an absolute amplitude value higher than 20 µV and occurred in the range of 3–50 ms after the stimulus onset. Then, the peak amplitude and the area under the curve (AUC) metrics were computed. The amplitude was taken as the absolute difference between the detected peak of the response and the first transition point from the same response. The AUC was computed as the absolute value of the integral of the response, whose boundaries were marked by the response’s transition points. In addition, long-lasting baseline z-score normalized evoked responses were also analyzed using a 600 ms post-stimuli time window. AUC and the standard deviation of the 32-channel-averaged evoked response were computed. Given the sample size and the comparison across only three groups in these experiments, we sought a balance between identifying true significant differences and the risk of false positives. We opted for the Benjamini-Hochberg [49] procedure to investigate the false discovery rate (FDR). The approach revealed an overall FDR of 0.125 across all our tests, suggesting that approximately 12.5% of the detected significant results might be false positives.

Histological processing of mice from electrophysiological recordings in vivo

After electrophysiological recordings were completed, all animals were perfused transcardially, and brains processed for histology as described above. One of the series was used to analyze the location of the microelectrode in striatum coated with fluorescent dye. All nuclei were labeled with Bisbenzimide (Hoescht, Thermo Fisher Scientific; Waltham, MA, USA). The second series was stained using the cytochrome oxidase histochemistry protocol (CyO; [50]) to help cytoarchitectonic localization of both stimulation and recording electrodes. Both series were mounted onto gelatin-coated glass slides, air dried, dehydrated in graded ethanol, defatted in xylene and coverslipped with DePex (Serva). These sections were analyzed and photographed under an epifluorescence light microscope (Eclipse 600; Nikon).

In vivo microdialysis

Extracellular glutamate and GABA concentration were measured by in vivo microdialysis as previously described [51, 52]. Briefly, one concentric dialysis probe (Cuprophan membrane; 6000 Da molecular weight cut-off; 1.5 mm long) was implanted in the striatum (AP, 0.5; LM, – 1.7; DV, – 4.5 in mm) of isoflurane-anesthetized mice (n = 5–6 mice per group). Microdialysis experiments were conducted in freely moving mice 24 and 48 h after surgery. Probes were perfused with artificial cerebrospinal fluid (aCSF in mM: NaCl, 125; KCl, 2.5; CaCl2, 1.26 and MgCl2 1.18) at 1.5 μL/min. Following an initial 150 min stabilization period, five baseline samples were collected (20 min each) before local drug application by reverse dialysis and then successive dialysate were recovered. The concentration of glutamate and GABA in dialysate samples was determined by an HPLC system consisting of a Waters 717 plus autosampler, a Waters 600 quaternary gradient pump, and a Waters XBridge Shield RP18 5 μm 100 × 3 mm column. Dialysate samples were pre-column derivatized with OPA reagent by adding 90 μL distilled water to the 10 μL dialysate sample and followed by the addition of 15 μL of the OPA reagent. After 2.5 min reaction, 80 μL of this mixture was injected into the column. Detection was carried out with a Waters 2475 scanning fluorescence detector using excitation and emission wavelengths of 360 nm and 450 nm, respectively. The mobile phase was pumped at 0.8 μL/min and consisted of two components: solution A, made up of 0.05 M Na2HPO4, 28% methanol, adjusted to pH 6.4 with 85% H3PO4, and solution B, made up of 100% methanol/H2O (8:2 ratio). After the elution of glutamate peak at 2,10 min with 100% solution A, a gradient was established going from 100% solution A to 100% solution B in 2 min. After washing out eluting peaks (3 min), next phase returned to initial conditions (100% solution A) in 2 min, GABA elution were found at 11,6 min. The detection limit for glutamate and GABA was 0.2 pmol (signal-to-noise ratio 3). Quantification of Glu and GABA was carried out by comparison to a daily standard curve comprising the concentration of neurotransmitters expected in dialysate samples. Microdialysis data are expressed as femtomoles per fraction (uncorrected for recovery) and are shown in figures as percentages of basal values (individual means of 5 pre-drug fractions).

All reagents used were of analytical grade and were obtained from Merck (Germany). (S)-AMPA (AMPA, alpha-amino-3-hydroxy-5-methyl-4-is-oxazole-4-propionate) and NBQX (2,3-dihydroxy-6-nitro-7-sulfamoyl-ben-zo(f)quinoxaline) were from Sigma/RBI. Drugs were dissolved in the perfusion fluid. Concentrated solutions (1 mM; pH adjusted to 6.5–7 with NaHCO3 when necessary) were stored at – 80 °C, and working solutions were prepared daily by dilution in aCSF and administered by reverse dialysis at the stated concentrations.

Statistical analysis

Statistical analyses were carried out using the GraphPad Prism 8.0 software. Sample sizes were chosen using a power analysis: 0.05 alpha value, 1 estimated sigma value and 75%. No methods of randomization were used to allocate animals to experimental groups. For single comparisons and normally distributed data (the Shapiro–Wilk normality test), we used two-tailed Student's t test (95% confidence). For multiple comparisons we used two-way ANOVA. For multiple comparisons, we used Kruskal–Wallis test plus Dunn’s multiple comparisons or Tukey’s test as a post hoc tests when required. A p value < 0.05 was considered significant. Non-normally distributed data were compared with Mann–Whitney U test for independent samples. All statistical information is further described in supplementary statistics.