Shank3 deletion in PV neurons is associated with abnormal behaviors and neuronal functions that are rescued by increasing GABAergic signaling

Pv-Cre^+/− Shank3^Wt/Fl mice were generated by breeding Shank3 floxed mice (kindly provided by Tobias Böeckers, from the Institute for Anatomy and Cell Biology of Ulm University, Germany) with Pv-Cre^+/− mice (PValb^tm1(cre)Arbr, The Jackson laboratory). To visualize PV⁺ interneurons, we generated Pv-Cre^+/− TdTomato^Fl/− Shank3^Fl/Wt mice using TdTomato^+/+ knock-in mice (Gt(ROSA)26Sor^{tm14(CAG−tdTomato)Hze}, The Jackson laboratory). The Shank3Δ11^−/− mice were generated as previously described by Schemeisser et al., 2012 [27] and re-derived in a C57BL/6 background (Charles River Laboratories, Calco, Italy). These mice have already been characterized by [30]. All experiments conducted for this project involved mice that were housed in an animal facility maintained at a constant temperature (22 +/− 1 °C) and humidity (50%), with a 12-h light/dark cycle, and provided with ad libitum access to food and water. The experiments were performed in accordance with the guidelines approved by the European Communities Council and the Italian Ministry of Health (Rome, Italy) for the ethical use of laboratory animals in research. All possible efforts were taken to minimize the number of mice used and reduce their suffering. Animals from both sexes were included in all experiments, and subsequently, the data were combined, as no statistical differences were observed between the two groups.

All primers were provided by Thermo Fisher; the REDExtract-N-Amp PCR Reaction Mix™

(Sigma-Aldrich) and the MyFi™ DNA Polymerase (Meridian Bioscience) were used for the polymerase reaction. PCR genotyping was performed using the following sets of oligonucleotide primers: parvalbumin for Wt allele forward 5′-CAGAGCAGGCATGGTGACTA-3′, for Wt allele reverse 5′-AGT ACCAAGCAGGCAGGAGA-3′, for mutant allele forward 5′-GCGGTCTGGCAGTAAAAACTATC-3′, for mutant allele reverse 5′-GTGAAACAGCATTGCTGTCACTT-3′; Shank3 floxed forward 5′-GTCTCTGTGGTTGGGGTGTC-3′, reverse 5′-CAGTGAAGAAGCCCCAGAAG-3′ for both Wt and mutant allele; TdTomato for Wt allele forward 5′- AAGGGAGCTGCAGTGGAGTA-3′, for Wt allele reverse 5′-CCGAAAATCTGTGGGAAGTC-3′, for mutant allele forward 5′- CTGTTCCTGTACGGCATGG -3′, for mutant allele reverse 5′- GGCATTAAAGCAGCGTATCC -3′; Shank3Δ11^−/− for Wt allele forward 5′-CAAGTTCATCGCTGTGAAGG-3′, for mutant allele forward 5′-CCTCTAGGCCTGCTAGCTGTT-3′, reverse 5′-AAGAAGCCCCAGAAGTGACA-3′ for both Wt and mutant allele.

Mice were anesthetized with a solution of 20% urethane in physiological solution (0.9% NaCl) with a final dose of 0.8 ml/hg ([23, 24, 31, 32]; 1.6 g/kg). The depth of anesthesia was evaluated by monitoring the pinch withdrawal reflex and other physical signs (respiratory and heart rate). Additional doses (10% of initial dose) were intraperitoneally administered to maintain the level of anesthesia if necessary. The brain surface was routinely moistened with the addition of artificial cerebrospinal fluid (ACSF) at body temperature. The head of the mouse was fixed in a stereotaxic apparatus, and a portion of the skull overlying the visual cortex (0.0 mm anteroposterior and 2.7 mm lateral to the lambda suture) was drilled on the right hemisphere. A chamber was created around the craniotomy applying a thin layer of dental cement (Vertex Dental). The mouse was placed with its left eye in front of the monitor at 30 cm, oriented 45° with respect to the medial sagittal plane of the animal. Local field potentials (LFP) were recorded with glass micropipettes (impedance ~ 2 MΩ), filled with ACSF solution and connected to the amplifier head stage with an Ag/AgCl electrode. The microelectrode was positioned into the visual cortex at a depth of 250–300 μm (II/III layer) with a motorized micromanipulator. A common reference Ag/AgCl electrode was placed on the cortical surface in the ACSF cortical bath. Electrophysiological signal was amplified 1000-fold with a multichannel differential amplifier (EXT-02F, NPI), band-pass-filtered (0.1–1000 Hz) and sampled at 2 kHz with 16-bit precision by a National Instruments (NI-usb6251) ADC board controlled by custom-made LabView software. Line frequency 50 Hz noise was removed by means of a linear noise eliminator (Humbug, Quest Scientific). LFP recording lasted around 45 min, after which some animals have been superfused over the visual cortex with Midazolam (5 mg/ml in ACSF; 200 µl volume; BI- Istituto Biochimico Italiano Giovanni Lorenzini S.p.A.) or its vehicle (ACSF). After administration, the mouse was left resting, in the dark, for 1 h. Then, a second round of LFP recording was performed. Visual stimuli were provided via an LCD monitor (120 Hz frame rate, Asus). The stimuli were generated with a custom-made MATLAB program based on Psychophysics Toolbox 3 [33]. The stimulus consisted in a black and white contrast-reversing checkerboard (0.5 Hz; 0.04 cycles/degree). The checkerboard contrast was changed between different traces. The contrast values were calculated using the formula:where K indicates contrast; L and l indicate, respectively, the maximum and minimum luminance of the stimulus; L_max and l_max indicate, respectively, the maximum and minimum luminance of the highest luminance stimulus and L_min and l_min the maximum and minimum luminance of the lowest luminance stimulus. Screen luminance was measured using a photometer (Konica Minolta). Six contrast values were used from maximum to minimum and randomly alternated (100%, 28%, 5.7%, 2.3%, 1.8% and 0%).

Analysis of slow-wave oscillations was performed computing the short-time root mean square (RMS) power of the data band passed in the gamma band (25–80 Hz). The time course of the RMS power was computed on a moving window of 250 ms width and 150 ms overlap. The distribution of the logarithm of the time-resolved gamma power was bimodal, reflecting high gamma activity during USs and low gamma activity during down states. This distribution was then fitted with a double Gaussian function and the threshold for the discrimination of USs was chosen, minimizing false positives and false negatives, by computing the ROC curve. A cutoff in the minimum up (down) states duration was set to 100 ms, and up (down) state shorter than the cut-off was assigned to the ongoing down (up) state (see [32] for details).

Spectrograms have been computed using the mtspecgramc function of the Chronux toolbox, with an overlapping Hammer window of 150 ms width and 50 ms overlap. Multitaper power spectra have been calculated on 90-s traces recorded from anesthetized mice with no visual stimulation, using the mtspectrumc function of the Chronux toolbox [25].

The visual evoked potential (VEP) of each mouse was calculated as the average of 120–180 trials. The peak amplitude was measured in a window from 90 to 500 ms after the stimulus, with respect to a baseline level computed in the 30–80 ms window. Contrast transfer functions obtained for every mouse have been fitted with a Michaelis Menten function:\(A(k) =\frac{{A}_{\mathrm{max}}*k}{{k}_{H}+k}\), where A is the response amplitude, k is contrast, A_max is the response at "saturating" contrast levels and k_H is the contrast at which the amplitude is equal to half M. The VEP slope was calculated as the angular coefficient of a line fitting the VEP in a window of 80 ms centered on the half-peak value. Gamma band power (25–80 Hz) during the response was calculated for every trial as the RMS power of the filtered trace in two different time windows. The windows were 30–80 ms and 90–500 ms after the stimulus, called, respectively, "baseline" and "response"; then the base-10 logarithm of the ratio between response and baseline gamma power has been calculated and averaged among trials.

In the figure legends are reported the number of replicates (n) and the statistical analyses used for every experiment. Based on the number of condition and the number of groups in the comparison, appropriate statistical tests were used to analyze the data. First was analyzed the data distribution by the Shapiro–Wilk normality test. For the comparisons, parametric tests including Student t-tests and one-way and two-way analysis of variance (ANOVA) with appropriate post hoc test were used if the data were normally distributed; nonparametric test including the Mann–Whitney test and Kruskal–Wallis test with appropriate post hoc test were used for data with non-normal distribution. Results are presented as box plots presenting the 1^st and 3^rd quartile, the median and the mean. Since each data point represents an experimental mouse, the total number of data points is limited and the box plots do not include whiskers, but all data points are presented, thus allowing a proper understanding of scattering and range. Statistics were computed as described in the figure legends.

All electrophysiological data analysis was performed with custom-made software suite written in MATLAB (https://​github.​com/​DidiLamers/​PCDH19_​ZebraExplore) [34].

Mice were anesthetized and perfused transcardially with sucrose 5% and paraformaldehyde 4%. Brains were removed and post-fixed overnight in paraformaldehyde 4%. The day after three washes with PBS were performed and brains were incubated for at least eight hours in 30% sucrose. Finally, brains were included in cryomolds with Tissue-Tek OCT compound (Bio-Optica) and put at − 80 °C until cryostat sectioning. Brains were cut with a cryostat and 20 μm-thick coronal slices were collected on polysine microscope adhesion slides (ThermoFisher). Slices were incubated first in blocking solution (3%BSA, 10% goat serum, 0.4% Triton-X-100, diluted in PBS) for at least 30 min and then they were incubated with primary antibodies overnight at 4 °C. Subsequently, three wash (10 min each) with PBS were performed and brain slices were incubated with fluorophore-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) for 1 h at room temperature. After the antibody incubation, brain slices were washed and 4',6diamidino-2-phenylindole (DAPI) staining (ThermoFisher) was carried out (DAPI diluted in PBS to a final concentration of 0.5 μg/ml). Finally, another washing step is performed before mounting the coverslips with Mounting Medium (Vecta Shield). Primary antibodies used were: anti-Parvalbumin (Swant, GP72), Shank3 (homemade SHANK3rb [35]).

Brain areas were selected according to the mouse brain atlas of Paxinos and Franklin (Paxinos & Franklin, 2008) from each of the following brain regions at the following bregma coordinates: prefrontal cortex (from 2.4 to 1.7 mm), visual cortex (from − 2.06 to − 2.70 mm), hippocampus (from − 1.06 to − 2.2 mm). Confocal images were obtained using LSM800 confocal microscope (Carl Zeiss) with Zeiss 10 × objective at a resolution of 1024 × 1024 pixels, under the same condition across different mice and brain slices. PV-positive cells were counted in the bilateral areas of each section using ImageJ software.

For Shank3 staining, imaging was performed on a laser-scanning microscope (Leica DMi8) with 63 × oil DIC immersion objective (ACS Apo, NA 1.3) with xy resolution 159 nm and z resolution 345 nm with a stack size of 1.73 µm. Deconvolution of the images was performed using Huygens Essentials 22.04 software (Scientific Volume Imaging B.V.). Subsequently, an analysis of the images was performed with surface and spot tool with Imaris 9.9 Software (Oxford Instruments). Parvalbumin-positive neurons within a specific region of interest (ROI) were selected with the surface tool by tdTomato signal. With the spot tool Shank3-positive spots close to the surfaces (If < or equal to 0.5 µm distance) were determined. Number of spots close to surface within one ROI was normalized on the surface volume within the ROI. Two sections (3 ROIs per section) were analyzed for each hemisphere per animals.

Cortex and hippocampi were dissected from mouse brain of both sexes. Tissues were homogenized in buffer containing 10 mM Hepes pH 7.4, 2 mM EDTA, protease inhibitors (Roche), and phosphatase inhibitors (Sigma, P8340). Samples were centrifuged at 500 × g for 5 min at 4 °C. Resulting supernatants were centrifuged at 10,000 × g for 15 min at 4 °C. After the centrifugation, pellets were resuspended in buffer composed by 50 mM Hepes pH 7.4, 2 mM EDTA, 2 mM EGTA, 1% triton-X-100 (Sigma), protease inhibitors, and phosphatase inhibitors and centrifuged at 20,000 × g for 80 min at 4 °C. Finally, pellets were resuspended in buffer containing 50 mM Tris pH 9, 1% sodium deoxicholate (Sigma, D6750). Samples were quantified by BCA protein assay (EuroClone) to assess protein concentration. Equal amounts of each sample were separated using SDS-PAGE and subsequently blotted on nitrocellulose membranes using the Trans- Blot Turbo System (BioRad). Membranes were washed in Tris-buffered saline-Tween (TBS-T) (200 mM Tris pH 7.4, 1.5 M NaCl (both Sigma-Aldrich) and 0.1% Tween 20% (BioRad). After 1 h blocking at room temperature with 5% bovine serum albumin or milk in TBS-T, membranes were incubated overnight at 4 °C with primary antibody (Shank3, Santa Cruz Biotechnology, cat. H-160; GABA-A-R-alpha1, Neuromab, cat. 75136; Parvalbumin, Swant, cat. GP72; β3-tubulin, Sigma, cat. T8578). Membranes were washed in TBS-T and then incubated with HRP-conjugated secondary antibodies (Jackson ImmunoResearch) for 1 h at room temperature. After three washes, chemiluminescence was induced using an ECL Western Blotting Substrate kit and further detected using a ChemiDoc XRS + machine. All signals were quantified using ImageLab software and normalized against the values of the respective signal for βIII-tubulin.

The spontaneous self-grooming behavior was assayed as described in [36]. Single mouse was placed into a standard cylinder (46 × 23.5 × 20 cm). Cylinders were empty to eliminate digging in the bedding, which is a potentially competing behavior. The room was illuminated at about 40 lx. A front-mounted CC TV camera (Security Cameras Direct) was placed at circa 1 m from the cages to record the sessions. Sessions were video-taped for 20 min. The first 10 min of habituation was not scored. Cumulative time spent grooming all the body regions during the second 10 min of the test session was measured.

The novel object recognition test was performed in an open plastic arena (60 × 50 × 30 cm). The test had three phases: the habituation on the first day, when mice were accustomed to the test arena for 10 min; the familiarization; and the novel object recognition the day after. In the familiarization phase, two identical objects were placed in the middle of the arena equidistant from the walls and from each other. Mice were placed between the two objects until it had completed 30 s of cumulative object exploration (20 min cut-off). Experimenter measured the time mice was within approximately 1 cm of an object with its nose toward the object. Climbing the object or pointing the nose toward ceiling near the object were not considered exploring behaviors. After familiarization, mice returned to the home cage until they were tested for novel recognition after 5 min, 120 min and 24 h. In the novel recognition phase, a novel object (never seen before) took the place of one of the two familiars. Scoring of object recognition was performed as during the familiarization phase. For each mouse, the role (familiar or new object) as well as the relative position of the two objects were randomly permuted. The objects used for the test were white plastic cylinders and colored plastic Lego stacks of different shapes. The arena was cleaned with 70% ethanol after each trial. Performance was analyzed by calculating the discrimination index (N − F/N + F), where N = the time spent exploring the novel object, and F = the time spent exploring the familiar object.

Spatial object recognition test was performed in an arena according to [30], with minor modifications that consisted in an opaque white Plexiglass cage (58 × 50 × 43 cm) that was dimly lit from above (27 lx) and two visual cues that were placed above two adjacent walls. The visual cues consisted of a black and white stripped pattern (21 × 19.5 cm) that was affixed to the center of the northern wall and a black and gray checkered pattern (26.5 × 20 cm) that was placed in the center of the western wall. Objects were placed across the visual cues. Mice were habituated to the arena for 10 min the day before the test. Twenty-four hours later, it is initially performed a training session where mice were allowed to familiarize with two different objects. The experimenter measured the time spent in sniffing both objects until the mouse completed 30 s in exploring objects (cut-off 20 min). Exploring behavior was defined as mouse having its nose directed toward the object and within approximately 1 cm of the object [37]; climbing or sitting were not considered exploration behaviors. After 5 min, 120 min and 24 h, mice were allowed to re-explore the cage where one object was moved in a new position. Between two sessions, mice returned to their home cage. Cage and object were carefully cleaned with 70% ethanol before and after all behavioral procedures. Performance was analyzed by calculating a discrimination index (N − F/N + F), where N = the time spent exploring the moved object during the test, and F = the time spent exploring the unmoved object during the test.

The beam apparatus consisted of 1 m beams with a flat surface of 12 mm or 6 mm width resting 50 cm above the table top on two poles. A black box was placed at the end of the beam as the finish point. Nesting material from home cages was placed in the black box to attract the mouse to the finish point. A lamp (with 60-W light bulb) was used to shine light above the start point and served as an aversive stimulus. A video camera was set on a tripod to record the performance. On training days, each mouse crossed the 12 mm beam 3 times and then the 6 mm beam 3 times. The time required to cross to the escape box at the other end (80 cm away) was measured with a stopwatch. The stopwatch started when the nose of the mouse began to cross the beam, and stopped when the animal reaches the escape box. Once the mice are in the safe box, they are allowed some time (~ 15 secs) to rest there. Before the next trial the mice rest for 10 min in their home cages between training sessions on the two beams. On the test day, times to cross each beam were recorded. Two successful trials in which the mouse did not stall on the beam are averaged. The beams and box were cleaned with towels soaked with 70% ethanol and then water before the next beam was placed on the apparatus.

The rotarod apparatus (Ugo Basile, Biological Research Apparatus, Varese, Italy) was used to measure fore and hindlimb motor coordination and balance. During the training period, each mouse was placed on the rotarod at a constant speed (12 and 32 rpm) for a maximum of 120 s, and the latency to fall off the rotarod within this time period was recorded. Mice received four trials per day for 4 straight days. The fourth trial of each day was evaluated for statistical analysis.

In the pole test, the mouse was placed on a vertical wire-mesh pole (90 cm length, 1 cm diameter) with its head facing upwards. Mice were habituated to descend the pole in 2 trials per day for 2 days. On test day (third day) mice were subjected to 5 trials: the total time taken to turn the body and to descend the pole was recorded according to [38]. A cutoff of 60 s was given. Data were shown as mean of 5 trials evaluated during the test day.

The limb force was tested by positioning a mouse on the top of a wire cage lid (19 × 29 cm) that was turned upside down at approximately 25 cm above a surface with the bedding material. The grip of the mouse was ensured by gently waving three times before rotating the lid as described in [39]. The latency to fall onto the bedding was recorded over a maximum period of 300 s.

The sociability tests were performing in an apparatus called three-chambered box. It is formed by a rectangular transparent polycarbonate apparatus with three-chamber (width = 42.5 cm, height = 22.2 cm, central chamber length = 17.8 cm, and side chamber lengths = 19.1 cm) as previously described in [36]. The three-chamber box was illuminated by diffuse incandescent lighting (15 lx), and a video camera was located directly over the center of the open field.

First, the test includes a habituation phase during which the mouse was placed in the middle chamber free to explore all the compartments for 10 min. After this period, novel conspecific mouse, that had no previous interaction with the test mouse, was placed in one of the side chambers. The doors were unblocked and the subject mouse was given the possibility to explore for 10 min either an empty chamber or a chamber containing the stranger mouse. In the 10-min trial, the duration of contact with the counterpart was recorded. Immediately after sociability test, without cleaning the apparatus, it was performed the social novelty test putting an unfamiliar mouse in the empty wire cage. Time spent in each chamber was recorded. The data were expressed in sociability index (SI) and social novelty preference index (SNI) as follows: SI = (time exploring novel mouse 1 – time exploring empty cage) / (time exploring novel mouse 1 + time exploring empty cage) and SNI = (time exploring novel mouse 2 – time exploring familiar mouse) / (time exploring novel mouse 2 + time exploring familiar mouse).

Pv-Cre^+/− Shank3^Wt/Wt, Pv-Cre^+/− Shank3^Fl/Wt, Shank3 Wt and Shank3 KO mice were treated with a selective positive allosteric modulator of GABA_A receptors: Ganaxolone. Ganaxolone (Tocris Cat. No. 2531) was resuspended in ethanol and betaciclodextrin (Sigma-Aldrich 332607). Mice received an intraperitoneal injection of ganaxolone (5 mg/Kg) or vehicle 30 min before each behavioral test.

As the visual cortex serves as a prototypical model for studying changes in brain circuitry associated with various forms of ASD [40‐43], we decided to analyze sensory processing in the visual cortex in Shank3Δ11^−/− [27] (referred as Shank3 KO) mice by conducting local field potential (LFP) recordings. The mice were anesthetized with urethane to induce a slow-wave-like sleep as in [32, 44]. During this state, the brain cortex exhibited oscillations between states of very low neuronal activity (known as down states, DS) and states of high neuronal activity (known as up states, US). Our results demonstrate that USs in Shank3 KO mice had a longer duration and a higher gamma power (25–80 Hz) compared to controls. However, there were no significant differences in the frequency or median amplitude of the up states between the two groups (Fig. 1A and Additional file 1: Fig. 1). This result is consistent with an enhancement of the spectral power for the frequency band 4–100 Hz, as represented in Fig. 1B, and suggests a reduction of the inhibitory drive. To further investigate the role of Shank3 in regulating inhibition, we measured visual contrast gain control, which is heavily reliant on the inhibitory feedback of fast-spiking interneurons on pyramidal neurons. This feedback is essential for adapting the limited dynamic range of pyramidal cell firing to the large dynamic range of sensory stimuli [45, 46]. We recorded local field potentials (LFP) in response to a checkerboard contrast-reversal stimulus and measured the amplitude of the visual evoked potentials (VEPs) in relation to contrast levels (Fig. 1C). Our findings indicate that the response of Shank3 KO mice reached a saturation level for lower contrast values, compared to the wild-type mice (Fig. 1D). The gain of the visual response was estimated for each mouse by fitting the contrast curves with a Michaelis–Menten function. The distribution of the two parameters (A_M, asymptotic response amplitude and K₅₀, half-saturating contrast) was compared between Shank3 KO and Wt mice (Additional file 1: Fig. 2A–C). We found that K₅₀ was lower in the Shank3 KO compared to Wt mice, indicating the presence of saturation at a lower contrast and increased contrast gain in Shank3 KO. Additionally, the recorded VEPs were steeper in Shank3 KO mice compared to Wt mice, as represented in the analysis of the slope (Additional file 1: Fig. 2C). Spectrograms of the VEP recordings show a power increase during visual response in Shank3 KO mice (Additional file 1: Fig. 2D–E). We quantified the power increase in the gamma spectral band both during response and in presence of a blank stimulus as a function of the stimulus contrast (Additional file 1: Fig. 2E). These curves appear to be similar to those obtained when measuring VEP amplitudes (Fig. 1D), confirming that contrast gain is greater in the Shank3 KO mice. The increased gain we measured in response to visual stimuli is suggestive of an impairment of inhibition from fast-spiking PV interneurons, which are responsible for gain control in the visual cortex [45]. Therefore, our data suggests that the major cause of the alterations in the visual response is due to inhibitory impairment. To test this hypothesis, we measured contrast gain in Shank3 KO mice before and after superfusion with the GABA_A receptor agonist midazolam (5 mg/ml). As shown in Fig. 1E–G, pharmacological activation of GABA_A receptor was sufficient to rescue gain control in Shank3 KO mice compared to Shank3 KO mice treated with vehicle only (ACSF). Midazolam superfusion per se acts on resting activity reducing the high excitability apparent in Shank3 KO mice as shown in Additional file 1: Fig. 3A–D, with no differences in US duration, frequency and amplitude. Given that Shank3 is localized on excitatory synapses only, we can rule out that the inhibition impairment originates in decreased number or strength of inhibitory synapses. Far more likely is that the excitatory drive on PV interneurons is defective, leading to reduced recruitment of PV interneurons and to impaired excitation/inhibition balance.

Our results align with substantial evidence suggesting a role for PV interneurons in the etiology of neuropsychiatric disorders [16, 47, 48]. Therefore, to determine the specific contribution of Shank3 deletion in these neurons to the pathogenesis of PMS and ASD, we utilized Cre-lox technology to generate mice with selective deletion of Shank3 in PV neurons. We first bred Shank3 floxed mice with transgenic mice carrying TdTomato gene which can only be expressed after the excision of a loxP-flanked STOP cassette by Cre-mediated recombination. Subsequently, we crossed these mice with the line expressing Cre-recombinase under the PV promoter. This allowed us to easily visualize PV neurons with the deletion of Shank3 using TdTomato. PCR analysis confirmed the genotype of these mice (Fig. 2A). Furthermore, immunofluorescence showed that TdTomato expression increased during brain development (Fig. 2B) and that cells expressing TdTomato were also positive for parvalbumin expression (Fig. 2C). We confirmed that Shank3 was downregulated in Pv-Cre^+/− TdTomato^Wt/fl Shank3^Wt/Fl mice compared to controls (Fig. 2D). Subsequently, we examined the impact of selective deletion of Shank3 in PV neurons on mouse behavior. Surprisingly, our analysis revealed that the deletion of a single Shank3 allele in PV neurons was sufficient to induce behavioral changes, which contrasts with the absence of cognitive deficits in Shank3 full KO mice in heterozygosity [30]. Pv-Cre^+/− Shank3^Wt/Fl spent significantly more time in doing grooming compared to Pv-Cre^+/− Shank3^Wt/Wt (referred as control mice) suggesting that Shank3 deletion in PV neurons is sufficient to cause repetitive behavior (Fig. 3A). Memory performance was also affected by selective deletion of Shank3 in PV neurons, as Pv-Cre^+/− Shank3^Wt/Fl exhibited deficits in both the novel object recognition test (Fig. 3B) and the spatial object recognition test (Fig. 3C). Furthermore, Pv-Cre^+/− Shank3^Wt/Fl displayed impaired motor coordination taking longer to cross the beam than control mice (Fig. 3D) and displayed some motor problems, as demonstrated by the results of the rotarod test (Fig. 3E). However, they did not exhibit any impairments during the pole test (Fig. 3F) nor a reduction in muscular strength (Fig. 3G). The three-chamber test did not reveal any differences between Pv-Cre^+/− Shank3^Wt/Fl mice and control mice in sociability or social novelty (Fig. 3H). Social performances were also unaffected in the homozygous Pv-Cre^+/− Shank3^Fl/Fl mice, indicating that Shank3 deletion only in PV neurons was insufficient to affect mice sociability (Additional file 1: Fig. 4A). Finally, we investigated whether homozygous Shank3 deletion in PV neurons exacerbates the phenotype observed in Pv-Cre^+/− Shank3^Wt/Fl mice. However, as demonstrated by the results of the grooming test and novel object recognition test, we observed only a slight deterioration in behavior compared to heterozygous mice, which did not appear to be significant (Additional file 1: Fig. 4B and C).

Next, we recorded the resting state LFP in the visual cortex of Pv-Cre^+/− Shank3^Wt/Fl mice under urethane anesthesia. Power spectra analysis showed that Pv-Cre^+/− Shank3^Wt/Fl mice had a significantly higher spectral power compared to controls at both low (< 5 Hz) and high frequencies (> 100 Hz Fig. 4A–C). Further detailed analysis of the SWA revealed that there was no significant difference in the frequency and duration of up states between the two groups (Fig. 4E, F). However, the amplitude of up states was larger in Pv-Cre^+/− Shank3^Wt/Fl mice, suggesting higher neuronal firing during up states in this group (Fig. 4 G). Additionally, the measured power during USs was significantly higher for Pv-Cre^+/− Shank3^Wt/Fl mice in HF band (> 100 Hz) (Fig. 4I). These findings are all indicative of a higher degree of excitability, suggesting that selective deletion of Shank3 in PV neurons caused a specific excitation/inhibition (E/I) balance that leads to cortical hyperexcitability.

We subsequently conducted a quantitative analysis to determine the density of parvalbumin (PV) neurons in different brain regions that exhibit high levels of Shank3 expression. We conducted an analysis of the number of Tdt + and Pv + cells and the percentage of colocalization across multiple brain regions, including the mPFC, CA1, CA3, DG, and visual cortex. Our results indicate that in all brain regions examined, 77% of Tdt + cells also express Pv + , suggesting high specificity of the model. In the hippocampus of Pv-Cre^+/− Shank3^Wt/Fl mice, we observed a significant decrease in the number of PV neurons specifically in the dentate gyrus, whereas no significant difference was detected in the CA1 and CA3 regions (Fig. 5A). Similarly, in the medial prefrontal cortex (mPFC) of Pv-Cre^+/− Shank3^Wt/Fl mice, we observed a reduction in the number of PV neurons when compared to Pv-Cre^+/− Shank3^Wt/Wt mice (Fig. 5B). However, in the visual cortex, no discernible difference was observed between Pv-Cre^+/− Shank3^Wt/Fl and Pv-Cre^+/− Shank3^Wt/Wt mice (Fig. 5C). Similar results were also obtained when we focused on the cells that express both TdTomato and Parvalbumin (Tdt + Pv + cells) (Additional file 1: Fig. 5) confirming the Cre specificity.

We discovered that the deletion of Shank3 in PV neurons leads to brain hyperexcitability. To investigate whether positive allosteric modulation of GABA_A receptors by the neurosteroid ganaxolone could improve the deficits in Pv-Cre^+/− Shank3^Wt/Fl mice we administered ganaxolone (5 mg/Kg) or vehicle intraperitoneally to mice before conducting behavioral tests. Due to their broad availability and safety, benzodiazepines and other positive allosteric modulators of GABA_A receptors could be a near-term strategy to improve the symptoms of PMS patients. However, one difficulty in using benzodiazepines is their sedative properties. For this reason, for the in vivo experiments we chose ganaxolone because it has fewer side effects. We first assessed whether ganaxolone could ameliorate stereotyped behavior by performing the grooming test and found that ganaxolone administration resulted in a significant reduction in grooming time compared to the vehicle-injected mice (Fig. 6A). In addition, ganaxolone treatment improved memory deficits in Pv-Cre^+/− Shank3^Wt/Fl mice as shown by the results of novel object recognition test (Fig. 6B) and spatial object recognition test (Fig. 6C). Moreover, ganaxolone treatment rescued motor coordination deficits in Pv-Cre^+/− Shank3^Wt/Fl (Fig. 6D).

Finally, to better characterize how Shank3 deletion in the entire brain affect PV neurons we quantified the number of PV neurons in the hippocampus, medial prefrontal cortex (mPFC), and visual cortex of Shank3 knockout (KO) mice. Surprisingly, we did not observe any significant differences in the number of PV neurons in these brain regions compared to wild-type (Wt) mice (Fig. 7A–C). However, we did observe a significant reduction in the expression of the GABA-A-R-alpha1 subunit in the postsynaptic density (PSD)-enriched fraction from the hippocampus of Shank3 KO mice compared to Wt and Pv-Cre^+/− Shank3^Wt/Fl (Fig. 7D), suggesting that ganaxolone treatment might have a beneficial effect also in Shank3 KO mice. In PSD-enriched fraction from the cortex we did not detect any alteration in GABA-A-R-alpha1 subunit (Fig. 7E); moreover, similar to the total cortex, we did not observe any significant differences in GABA-A-R-alpha1 expression between the groups analyzed in PSD-enriched fraction from the visual cortex (data not shown). To evaluate the effects of ganaxolone treatment on Wt and Shank3 KO mice, we injected the mice with ganaxolone (5 mg/Kg i.p.) or vehicle 30 min before conducting behavioral tests. Our results demonstrate that ganaxolone treatment rescued memory impairment in Shank3 KO mice at 5-min, 120-min and 24-h delay, as evidenced by the results of novel object recognition test (Fig. 7F). Additionally, acute treatment with ganaxolone was also able to rescue repetitive behavior in Shank3 KO mice to the levels of Wt mice (Fig. 7G). In conclusion, our findings demonstrated that PV neurons activity is sensitive to Shank3 ablation and that restoring their function with ganaxolone may represent a possible pharmaceutical approach for ameliorating Shank3-related behavioral alterations.