Amygdala GluN2B-NMDAR dysfunction is critical in abnormal aggression of neurodevelopmental origin induced by St8sia2 deficiency

Aggression is frequently observed in neurodevelopmental psychiatric disorders such as schizophrenia, autism, and bipolar disorder. Due to a lack of understanding of its underlying mechanisms, effective treatments for abnormal aggression are still missing. Recently, genetic variations in Sialyltransferase 2 (St8sia2) have been linked to these disorders and aggression. Here we identify abnormal aggressive behaviors and concomitant blunted fear learning in St8sia2 knockout (−/−) mice. It is worth noting that the amygdala of St8sia2−/− mice shows diminished threat-induced activation, as well as alterations in synaptic structure and function, including impaired GluN2B-containing NMDA receptor-mediated synaptic transmission and plasticity. Pharmacological rescue of NMDA receptor activity in the amygdala of St8sia2−/− mice with the partial agonist d-cycloserine restores synaptic plasticity and normalizes behavioral aberrations. Pathological aggression and associated traits were recapitulated by specific amygdala neonatal St8sia2 silencing. Our results establish a developmental link between St8sia2 deficiency and a pathological aggression syndrome, specify synaptic targets for therapeutic developments, and highlight d-cycloserine as a plausible treatment.

between shocks was 20 s. The minimum current intensities required to elicit flinching and jumping in mice were measured. Open Field. Animals were placed in a rectangular arena (50 x 50 cm) and left to freely explore for 20 min. The light was adjusted to a level of 8-10 lux in the center of the arena. Video tracking of the animal's location was performed by a camera fixed above the arena, and images were transmitted at 5 Hz to a PC running Ethovision tracking system for further processing. The percentage of time spent in the center was taken as an indicator of anxiety.
Marble-burying test. To test the spontaneous burying behavior as an indication of anxiety-like behavior, mice were placed individually in a cage measuring 35 × 17 × 12 cm (L x W x H), containing bedding of 5 cm depth, with 12 glass marbles (2-3 cm diameter) evenly spaced on the surface of the bedding. Testing was conducted for 20 min, and buried marbles (i.e. at least onehalf covered with bedding) were counted at 1 min bins.
Locomotor activity. Animals were placed in individual new cages for a test duration of 24 hours.
After 2 hours of habituation, locomotor activity was assessed by the number of laser breaks in PhenoMaster system (TSE Systems GmbH, Germany).
Social preference test. The sociability test was carried out in a three-chambered box (the center compartment was 20 × 35 × 35 cm and the left and right compartments were 30 × 35 × 35 cm).
The dividing walls had retractable doorways that allowed access to each chamber. The test mouse was habituated to explore the entire apparatus for 10 min during 2 days. Each of the two side chambers contained an empty wire cage. The wire cages were 10 cm in height, with a bottom diameter of 9 cm and each bar spaced 1 mm apart. Juvenile mice (23 day-old C57BL/6J male mice) were habituated for 10 min to the wire cage for 2 days. In the third day (after habituation sessions), a test mouse was placed in the center compartment and allowed to explore the entire apparatus for 10 min. A juvenile mouse was enclosed in one of the wire cages, which was placed in one of the two sides of the social test box during the 10-min session. A dummy black mouse was placed in the other wire cage on the other side of the box. The time spent sniffing each wire cage was video-recorded and manually scored to evaluate the level of preference for the unfamiliar mouse compared with the object.
Quantification. Images were taken with confocal microscope (Zeiss LSM-700) using a 20X objective. Sample images were captured from different areas at the same coordinates for each animal using the mouse stereotaxic atlas as a reference. 2 Quantification was performed on original, unenhanced images only. Quantification of immunofluorescence LSM images were stitched together using the grid stitching plug-in for FIJI. The background intensity of each channel was measured at five different random areas and averaged to generate a mean background that was subtracted from each channel. Cells were delineated using a Triangle threshold to label only those stained with NeuN within 200-1000 pixels. The number of labeled cells that were co-labeled with phospho-ERK and the antibody of interest was counted and converted to a percentage of the total number of NeuN-stained cells for each section. Analyses were made blind to experimental conditions. GluN2A or GluN2B fluorescence intensities were measured in NeuN-positive cells.

Electrophysiological recordings.
The experiments described here were performed in 5-8 week-old mice. temperature. Spontaneous events were acquired for 5 min at -60 mV, starting from >5 min after the establishment of the whole-cell configuration, to allow the diffusion of the intracellular solution. The contribution of perisomatic inhibition was evaluated by selecting the events with fast rise time (< 3 ms; this limit was chosen based on the cumulative distributions of all events, that display a relative peak below this value). 3,4 To elicit asynchronous EPSCs (aEPSCs) at cortical inputs, extracellular CaCl2 was replaced by equimolar SrCl2. Cells were patched with the CsGluconate-based solution and held at -80 mV to amplify AMPAR-mediated currents. Only aEPSCs occurring between 20-500 ms after the onset of the first evoked EPSC were considered.
For detection of both mIPSCs and aEPSCs, traces were filtered at 1 kHz and analysed using the MiniAnalysis Program with a threshold corresponding to 2 times the baseline noise (Synaptosoft Inc., Decatur, USA).
Statistics. The number of animals/recordings per group was in agreement with the resource equation method to determine the sample size 5 and was guided by previous work of the lab with the same animal model. Power analysis was not performed, as animals typically underwent different experimental series (see Supplementary Figure 2), yielding multiple parameters with a group difference that was not known a priori. All groups consisted of males and were always matched per age among compared groups. For data exclusion, Grubbs' test for outliers was performed in GraphPad with an alpha level of 0.05, resulting in the exclusion of 1 animal from the cue fear learning series, 1 animal from mRNA analysis, and of 1 datapoint in the LTP control series in WT animals. Electrophysiological data are presented with n numbers representing recordings pooled from at least 3 animals per series, to take into account inter-animal variability.
Data from key experiments (such as AMPA/NMDA ratio, input-output curves, and LTP) were obtained from at least 5 animals. The animals used in this study were in general randomly distributed. Before any experiment, a number was allocated to each animal, and after all behavioral characterization and analyses, the genotype was uncovered. For the experiment involving treatment, trait anxiety of animals was measured as described in the methods, and subjects were assigned to control and drug groups in order to obtain groups with comparable mean anxiety. Data analysis was performed either by a researcher that was blind to genotype and treatment group or by an automated software.