Neuronal signature of social novelty exploration in the VTA: implication for Autism Spectrum Disorder

Novel stimuli attract our attention, promote exploratory behavior, and facilitate learning. Atypical habituation and aberrant novelty exploration have been related with the severity of Autism Spectrum Disorders (ASD) but the underlying neuronal circuits are unknown. Here, we report that dopamine (DA) neurons of the ventral tegmental area (VTA) promote the behavioral responses to novel social stimuli, support preference for social novelty, and mediate the reinforcing properties of novel social interaction. Social novelty exploration is associated with the insertion of calcium-permeable GluA2-lacking AMPA-type glutamate receptors at excitatory synapses on VTA DA neurons. These novelty-dependent synaptic adaptations only persist upon repeated exposure to social stimuli and sustain social interaction. Global or DA neuron-specific inactivation of the ASD risk gene Neuroligin3 alters both social novelty exploration and the reinforcing properties of social stimuli. These behavioral deficits are accompanied by an aberrant expression of non-canonical GluA2-lacking AMPA-receptors at excitatory synapses on VTA DA neurons and an occlusion of novelty-induced synaptic plasticity. Altogether, these findings causally link impaired novelty exploration in an ASD mouse model to VTA DA circuit dysfunction.


Introduction
From infancy, we encounter an array of diverse stimuli from the environment.
Stimulus repetition can result in habituation whereas novel stimuli trigger elevated behavioral responses. Habituation and novelty detection allow us focusing attention on what is un-known, promote exploratory behavior, facilitate learning and are predictive of cognitive function later in life 1 . Several neuropsychiatric disorders are characterized by deficits in habituation and novelty exploration. In autism, young ASD patients show prolonged attention to depictions of objects but reduced attention to social stimuli 2 . Moreover, ASD patients exhibit aberrant habituation to social stimuli and reduced responses to social novelty 3,4 . Such alterations in novelty responses and habituation appear to be observed in a significant number of individuals with ASD, as they have been reported in clinical studies using diverse stimuli and read-outs [5][6][7][8][9] . However, the circuits and neuronal mechanisms underlying this specific aspect of the ASD phenotype remain largely unknown.
One system that may contribute to social novelty responses and habituation are dopamine (DA) neurons in the ventral tegmental area (VTA) and substantia nigra (SN). DA neurons increase their activity in response to novel environments 10 , stimuli of positive or negative value 11 , and natural rewards, such as food 12 . Interestingly, these neurons also respond to non-rewarding novel stimuli and their responses habituate when the stimulus becomes familiar 13,14 . This has led to the proposal that novelty by itself may be rewarding. Notably, several studies highlight decreased social reward processing in patients with ASD 15,16 and these alterations have been hypothesized to precipitate further developmental consequences in social cognition and communication 17 .
In rodents, VTA DA neurons increase their activity in response to rewarding stimuli 18 , unfamiliar conspecifics or unfamiliar objects and this activity is necessary to promote social, but not object exploration 19 . Moreover, VTA DA neuron outputs modulate inter-peduncular nucleus (IPN) activity resulting in an enhanced exploration of familiar social stimuli 20 . Basal ganglia circuits, oxytocin, endocannabinoid, and dopamine signaling have been shown to regulate social reward behaviors [21][22][23] .
However, the circuitry and synaptic plasticity events contributing to social novelty responses remain incompletely understood.
Glutamatergic synapses onto DA neurons undergo several forms of synaptic plasticity that may contribute to the modification of social interactions in response to experience. Specific synaptic adaptations have been described during development, after drug exposure, cue-reward learning, reciprocal social interactions and after repeated burst stimulation of DA neurons 18,[29][30][31][32] . Furthermore, glutamatergic transmission is altered in several ASD animal models 33 , and we have recently shown that deficits in the postnatal development of excitatory transmission onto VTA DA neurons lead to sociability deficits 34 . Whether specific forms of synaptic plasticity in the VTA are induced by novelty exposure and whether aberrant plasticity associated with social novelty detection in the VTA is related to the maladaptive responses to social novel stimuli in ASD models is still largely unknown.
In this study, we parsed the response to social novel stimuli, social novel preference and the reinforcing properties of novel social stimuli as specific aspects of sociability controlled by DA neurons. We demonstrate that intact VTA DA neuron excitability is necessary to drive preference for social novelty but not for novel objects. Additionally, we developed a social novelty conditioned place preference protocol and show that VTA DA neuron function is required for social noveltyinduced contextual reinforcement learning. Mice lacking the expression of the ASDrisk factor Neuroligin 3 (Nlgn3) exhibit aberrant social novelty and habituation processing. These phenotypes are recapitulated by VTA DA neuron-specific downregulation of Nlgn3, thus providing a cell type-and circuit-based perspective on specific aspects of sociability dysfunctions in ASD. Finally, we discovered a form of novelty-induced synaptic plasticity at glutamatergic inputs onto VTA DA neurons that sustains social interactions and is impaired in Nlgn3 KO and Nlgn3 VTA DA knockdown mice.
Virus infusions led to mCherry expression in 50% of TH + (Tyrosine hydroxylase, an enzyme necessary for DA synthesis) VTA neurons and in only very few (2%) of TH + cells in the neighboring substantia nigra pars compacta (SNc; Fig. S1a), confirming preferential targeting of the VTA. Application of the hM4Di ligand Clozapine-noxide (CNO) decreased the neuronal excitability of VTA::DA hM4Di neurons compared to VTA::DA mCherry , measured ex vivo, as a reduction in the number of action potentials fired at increasing amplitude steps of current injections (Fig. S1b).
To assess the effect of the reduced excitability of DA neurons on socialnovelty exploration, we conducted a social habituation/social novelty task. To enable the identification of enduring alterations in synaptic function and plasticity (see further below) we modified the classic short-term social memory paradigm 25,36 by increasing the length of each interaction trial, performing 1 trial of 15 minutes a day over 5 days (Fig. 1b). To compare responses to social as well as non-social stimuli, we examined the behavioral responses not only for same-sex conspecifics but also objects. When repeatedly exposed to the same social ( Fig. 1c) or object stimulus ( Fig.   S1c; s1 and o1, respectively), VTA::DA hM4Di animals injected with vehicle show reduced interaction with the stimuli over days. We refer to this process as "long-term habituation". After 4 habituation days, the animals increased their exploratory behavior towards either a novel social stimulus (s2; Fig. 1d) or a novel object (o2, To study the role of VTA DA neurons in this behavioral trait, DA neuron excitability was decreased by intra-peritoneal (i.p.) injection of CNO in VTA::DA hM4Di before the exposure to the novel stimulus at day 5. We found that VTA::DA hM4Di animals decreased their exploratory behavior toward the new social stimulus. By contrast, VTA::DA mCherry mice treated with CNO showed unaltered novelty exploration (s2, Fig. 1e, f). Interestingly, when exposed to a novel object ! 5! (o2), both VTA::DA hM4Di and VTA::DA mCherry animals treated with CNO exhibited novelty exploration (Fig. S1e, f). Thus, reducing VTA DA neuron excitability alters social novelty-induced increases in exploratory behavior but does not affect the investigation of an unfamiliar object, suggesting a differential requirement of DA neuron activity for driving exploration of social and inanimate stimuli.

Intact VTA DA neuron excitability is necessary for preference for social novelty
When comparing stimuli of different nature, ASD patients show reduced attention to social stimuli but pay attention to non-social stimuli 37 . To assess the role of VTA DA neuron excitability in mediating both the orienting toward and the exploration of a novel social stimulus over an inanimate object or a familiar social stimulus, VTA::DA hM4Di and VTA::DA mCherry mice were subject to the 3-chamber test 38 under vehicle and CNO conditions. To this end, the test was performed twice: first, the animals received either vehicle or CNO and, after 1 week of washout, the test was repeated and the pharmacological treatment was counterbalanced (Fig. 2a).
To monitor potential off target effects of CNO 39 we also included VTA::DA mCherry mice treated with CNO as controls. During the task, the animals were given a choice between an object (o1) versus an unfamiliar mouse (s1 or s3; social preference, SP) and subsequently a choice between a familiar (second exposure to s1 or s3) versus an unfamiliar social stimulus (s2 or s4; preference for social novelty, SN). Previous studies define sociability in this assay as longer time spent in the chamber with the same-sex target mouse rather than in the chamber with the object, and more time spent sniffing the same-sex mouse rather than sniffing the object 40,41 .
According to these criteria, VTA::DA hM4Di mice treated with vehicle (Fig. S2a, d), VTA::DA mCherry mice treated with CNO (Fig. S2b, e) as well as in VTA::DA hM4Di mice treated with CNO (Fig. S2c, f) exhibited sociability. In the social preference phase of the assay, we observed a decreased distance moved upon CNO-mediated reduction of DA neuron excitability (Fig. S2g). However, despite the reduced locomotion, experimental subjects still expressed social preference.
During the second phase of the 3-chamber task, preference for social novelty was defined as longer time spent in the chamber with the same-sex novel mouse rather than in the chamber with the familiar mouse and more time spent sniffing the ! 6! same-sex novel mouse rather than sniffing the familiar mouse, particularly during the first 5 minutes of the test 38 . While preference for social novelty was exhibited by VTA::DA hM4Di mice treated with vehicle (Fig. 2b, e) and by VTA::DA mCherry mice treated with CNO (Fig. 2c, f), it was absent in VTA::DA hM4Di mice treated with CNO ( Fig. 2d, g). As for the social preference phase, CNO treated VTA::DA hM4Di mice displayed a reduction in distance moved (Fig. 2h). Additionally, to compare preference for social novelty across groups, we calculated a preference score, named here "social novelty index", as time spent sniffing the novel social stimulus minus time spent exploring the familiar target 42 , in the first and last 5 minutes of the assay.! We found that the social novelty index was reduced by CNO injections in VTA::DA hM4Di mice compared to both CNO treated VTA::DA mCherry and vehicle treated VTA::DA hM4Di (Fig. 2i). Altogether, these findings indicate that reducing the excitability of DA neurons decreases preference for novel social stimuli when given a choice to explore either a familiar or a novel social stimulus.

VTA DA neuron excitability mediates social novelty reinforcing properties
To investigate whether novel stimuli have reinforcing properties in mice, we designed a novelty conditioned place preference (nCPP) protocol. The protocol modifies previously used social-CPP paradigm 21,43 as follows: mice are housed jointly with familiar mice throughout the protocol and are conditioned with novel stimuli in a test apparatus. After the Pre-TEST, we performed 4 days of repeated conditioning where wild-type (WT) mice learn to associate one compartment of the apparatus with the presence of either a co-housed social (familiar, f1), a novel social (s1) or a novel object (o1) stimulus while the other compartment is left empty (Fig.   3a, b). At day 5 (Post-TEST) the preference of mice to explore the two compartments, in the absence of social or object stimuli, was quantified and compared to Pre-TEST.
While with this nCPP protocol no significant preference was developed with the familiar social stimulus ( Fig. 3c and Fig. S3a), mice exhibited preference to explore the compartment associated with the novel social stimulus ( Fig. 3d and Fig. S3b), and an avoidance for the novel object stimulus associated chamber ( Fig. 3e and Fig. S3c).
Interestingly, across conditioning sessions, we observed habituation to all the stimuli ( Fig. 3f-h). However, when the time of interaction with the stimulus during the first ! 7! and the last day of conditioning were plotted, we observed a higher interaction with novel social stimulus compared to the other stimuli at both time points (Fig. 3i).
These data suggest that a novel social stimulus remains salient over days and promotes contextual associative learning.
To assess the role of VTA DA neuron excitability in mediating reinforcing properties of novel social stimuli, both control VTA::DA mCherry and VTA::DA hM4Di received injections of CNO before each conditioning session and were treated with vehicle before the Post-TEST (Fig. 3j). Control VTA::DA mCherry but not VTA::DA hM4Di mice developed a preference for the compartment associated with novel social stimulus exposures ( Fig. 3k and Fig. S3d, e). These observations suggest that the excitability of DA neurons mediates both the interaction with a novel social stimulus as well as its reinforcing properties.  26,[49][50][51] . We examined male-male interactions in the social habituation and social novelty exploration test (Fig. 4a). Nlgn3 KO mice exhibited overall lower interaction times, no significant habituation, and lacked the increased response to social novelty seen in wild-type littermates (Fig. 4b, c and Fig. S4a-d).
However, Nlgn3 KO mice showed normal habituation and novelty response to objects ( Fig. 4d, e) and preference for novel objects in a novel object recognition task ( Fig.   4f-h). This indicates that both novelty preference and memory for objects are unaltered. In addition to impaired social novelty response, Nlgn3 KO mutants exhibit alterations in motor activity ( Fig. 4i) and marble burying (Fig. 4j). In an olfactory discrimination test 52 , Nlgn3 KO male mice showed normal response and habituation to a social odor (Fig. S4e). However, the mutant mice had a significantly decreased response when subsequently presented to a second (novel) social odor (Fig. S4e) mice did not develop a preference for the social compartments whereas wild-type mice did (Fig. 4k, l, and Fig. S4f, g). These findings suggest that Nlgn3 KO mice exhibit altered social interactions and defects in social reward behaviors.

Nlgn3 in VTA DA neurons is required for social novelty and social interaction
The diverse alterations in social but also non-social behaviors in Nlgn3 KO mice, indicate that multiple different systems might contribute to their phenotype. To test whether any alterations are due to Nlgn3 functions in VTA DA neurons we generated microRNA-based knock-down vectors for conditional suppression of Nlgn3 expression ( Fig. S5a, b). Cre-dependent AAV-based vectors were injected into the developing VTA of DAT-Cre mice at postnatal days 5-6 and mice were analyzed using a battery of behavioral tests (AAV2-DIO-miR Nlgn3 in DAT-Cre mice: VTA::DA NL3KD , Fig. 5a, b, and see VTA::DA NL3KD mice showed preference to novel objects in the novel object task ( Fig.   5h-j). Thus, there is a specific requirement for Nlgn3 in VTA DA neurons for appropriate social novelty responses and for the reinforcing properties of social interaction. By contrast, motor activity, marble burying, and social olfaction that are altered in global Nlgn3 KO mice were not modified in the VTA::DA NL3KD mutants ( Fig.   5k, l, Fig. S5l). Interestingly, we observed that knock-down of Nlgn3 in VTA-DA neurons of adult mice produced a similar but less pronounced social interaction phenotype as in developing animals, with reduced habituation and reduced social novelty response (Fig. S6). Thus, Nlgn3 is required for normal function and/or plasticity in VTA DA cells, even in fully developed circuits.

A synaptic signature of saliency detection in VTA DA neurons
Several experiences strengthen synaptic transmission and drive the insertion of GluA2-lacking AMPARs at excitatory inputs onto DA neurons. This form of plasticity can be assessed by calculating a rectification index (RI) and has been observed both in the VTA 30,31 and in dorsal raphe 53 . We used the long-term habituation/novelty task to test whether novelty exploration induced specific forms of long-lasting synaptic plasticity at excitatory inputs onto DA neurons in the VTA. In WT mice, the RI increased at synapses 24 hours after exploration of either a novel mouse or a novel object when compared to RI calculated from home caged mice (Fig.   6a). By contrast, the RI was unchanged after the exposure to a new context and AMPA/NMDA ratios were unchanged for any of the above conditions (Fig. 6b).
When AMPAR EPSCs were recorded after repeated exposure (over 4 days) to object stimuli, the RI was normalized to control condition (Fig. 6c). A subsequent exposure to a new object (o2) promoted the increase in RI (Fig. S7a). By contrast, GluA2lacking AMPARs were detected in mice repeatedly exposed to a social stimulus (s1) over a four-day period and were still present at these synapses after ten days of repeated exposure (Fig. 6c). Remarkably, the AMPA/NMDA ratio was significantly elevated after 4 days of social (s1) repeated exposure relative to baseline but was normalized after 10 days of repeated exposure (Fig. 6d), while the Paired-Pulse Ratio (PPR) remained unchanged throughout (Fig. S7b). Taken together, these data indicate that repeated exposure to a novel social stimulus, but not an object stimulus, transiently increases synaptic strength (AMPA/NMDA) and produces a stable insertion of GluA2-lacking AMPARs at VTA DA neuron excitatory inputs.
To understand the functional role of non-canonical AMPARs inserted during repeated social novelty exposure, we infused the GluA2-lacking AMPAR blocker NASPM into the VTA starting from the second day of interaction with either social or object stimuli (Fig. 7a,b). NASPM infused mice reduced the interaction with a social stimulus upon repeated exposure (Fig. 7c); by contrast, the infusions did not alter long-term habituation to an object (Fig. 7d), social interaction in the home cage between two familiar mice or distance moved in an open field (Fig. S7c- (Fig. 7e,f). This non-contingent burst activation increased RI in photocurrent positive neurons (I ChR2 + ; Fig. 7g) and blocked long-term habituation to social stimuli (Fig. 7h). Altogether, these data indicate that GluA2-lacking AMPARs might represent a synaptic signature of social stimulus saliency and, once inserted, their activity counteracts habituation.

Nlgn3 loss-of-function impairs novelty-induced plasticity
Nlgn3 has been implicated in the regulation of AMPARs at glutamatergic synapses 49,54 . We therefore hypothesized that defects in DA neuron synaptic function could represent the mechanism underlying the aberrant habituation and response to novel social stimuli in VTA::DA NL3KD mice. We explored glutamate receptor function in VTA DA neurons of global Nlgn3 KO and conditional VTA::DA NL3KD mice.
Notably, we observed increased RI of AMPAR-mediated currents indicating the aberrant presence of GluA2-lacking AMPARs at excitatory inputs onto VTA DA neurons in both Nlgn3 loss-of-function models (Fig. 8a). Given the abnormal elevation of GluA2-lacking AMPARs in naïve VTA::DA NL3KD mice, we hypothesized that in these mice social novelty-induced plasticity might be occluded. Indeed, GluA2-lacking AMPARs in VTA DA neurons were not further increased 24 hours after novelty exposure in VTA::DA NL3KD mice (Fig. 8b). Thus, aberrant plasticity of GluA2-lacking AMPARs in VTA DA neurons is associated with an impaired response to a social novel stimulus.

Discussion
In this study, we established that VTA DA neurons drive social novelty exploration and preference, two aspects of social behavior. Novel stimuli, independent of their nature, leave a plasticity trace at glutamatergic synapses in the VTA, which persists upon repeated exposure to social stimuli and supports sustained social interactions. We used deletion of Nlgn3 [44][45][46]48  The specific synaptic signatures observed in response to social novelty responses might occur in dedicated circuits, within the DA system, responsive for processing highly salient social stimuli in a temporally-defined manner.
Electrophysiological recordings from the VTA have pointed to the fact that DA neurons constitute a heterogeneous population in terms of intrinsic properties, projection specificity and neurotransmitter/neuromodulator release [73][74][75] . This diversity is thought to subserve the processing of reinforcing and aversive experiences in an input-output specific manner [76][77][78] . VTA DA neurons projecting to the nucleus accumbens (NAc), but not prefrontal cortex (PFC), control social interaction 19 while IPN-projecting DA neurons are involved in preference for social novelty expression 20 . Therefore, the synaptic adaptations reported in response to repeated exposure to social novelty exposure might occur in DA neurons projecting to either the NAc, the IPN or both. At the same time, given the intrinsic diversity of sensory and emotional information provided by social vs inanimate stimuli, it is also conceivable that synaptic plasticity occurs at specific inputs to defined subclasses of VTA DA neurons. Additional investigations of synaptic properties of defined inputs to projection-specific DA neuron subclasses is needed to further understand the circuits and the synaptic mechanisms underlying both novelty and saliency processing associated with social and inanimate stimuli.

14!
Altered social interactions and communication are defining aspects of the autism phenotype. However, such alterations may arise from a plethora of neuronal processing defects, ranging from alterations in perception, sensory processing, multisensory integration, or positive and negative valence assigned to social stimuli! 17,79,80 .
In this work, we specifically explored neuronal circuitry relevant for social novelty responses. We chose this domain, as studies in children with ASD demonstrated altered habituation and responses to novel stimuli 81,8 . Notably, in toddlers, a slowed habituation to faces but normal habituation to repeatedly viewed objects has been reported to coincide with more severe ASD symptoms 3 . Several rodent models of ASD exhibit altered social novelty responses 25, 28, ! 27, 82-84 ! and such alterations have been suggested to reflect changes in social memory or discrimination. However, brain areas and circuit elements contributing to these changes in habituation and social novelty responses in mice and humans are largely unknown. Our rodent work not only highlights a contribution of VTA DA neurons to this process but also takes steps toward identification of the synaptic basis of social novelty responses and habituation.
Considering the complexity of ASD behavioral dysfunctions, we propose that fractionating the autism phenotype according to specific behavioral domains based on neuronal circuit elements will provide a productive stratification criterion for patient populations. Thus, we speculate that in a sub-population of individuals with ASD alterations VTA DA function might contribute to the social interaction phenotype whereas in other sub-groups of patients alterations in social interaction may arise for different reasons. A prediction from this hypothesis is that stratification of patient populations based on an assessment of novelty responses, habituation, and social reward may help to identify sub-groups of patients that would particularly benefit from interventions targeting function and plasticity of the VTA-DA circuit elements.

Tzanoulinou for insights on behavioral experiments and Lorena Jourdain and Caroline
Bormann for technical support.

Author Contributions
Ex vivo electrophysiology experiments were performed by S.B., C.B and S.M.

Animals
The study was conducted with wild-type (WT) and transgenic mice in C57BL/6J  Between experiments, the cannula was protected by a removable cap. All animals underwent behavioral experiments 1 -2 weeks after surgery.

3-chamber test
A three-chambered social preference test was used, comprising a rectangular

Intra-peritoneal injection of saline and Clozapine-N-oxyde (CNO)
The mice were weighted before each experiment and intra-peritoneal (i.p.) injection.
The CNO dose was based on previous publications 88 and a concentration of 5mg/kg -1 was used for all the experiments. The CNO was , iluted in saline to obtain a concentration of 0.5mg.mL -1 to inject a reasonable volume of solution. The volume of saline (vehicle) injection was comparable to the volume of CNO solution.

Long-term habituation/novelty exploration
An experimental cage similar to the animal's home cage was used for this task. The bedding was cleaned after each trial and water and food were available. A novel social stimulus (s2) was placed with the experimental mouse in the cage for 15 minutes to allow direct interaction. In total, the experimental mice were exposed to 2 different social stimuli: one social stimulus repeatedly presented from day 1 to day 4 (habituation phase, s1) and a second mouse at day 5 (novelty phase, s2) The same protocol as described above was used for object habituation/novelty. The VTA::DA hM4Di experimental mice received injection of saline and were exposed to the same object (der klein kaufman tanner; Germany, o1) from day 1 to day 4 (habituation phase). On day 5 the animals were injected with either saline or CNO, and were exposed to a novel object stimulus (novelty phase, o2).
To exclude pharmacological effects of the CNO dose on the behavioral parameters analyzed in this task, VTA::DA mCherry mice underwent the habituation/novelty and received an intraperitoneal injection of saline from day 1 to day 4 and CNO on day 5.

27!
The social and object habituation/novelty task performed with Nlgn3 KO and VTA::DA NL3KD was performed as described above. The test was done in a cage similar to the mice home cage containing food and water; the same cage was used for the duration of the trial. 3 -4 weeks old C57Bl/6J male mice were used as stimulus mice, lego blocks and a small plastic toy were used as object. The animals were left to freely interact with the stimulus mouse or object for 15 minutes. For 4 consecutive days the experimental mouse was exposed to the same stimulus (s1 or o1). Day 5 consisted of the novelty phase (s2 or o2). At the end of each trial, the experimental and stimulus mice were returned to their home cage.
During the social habituation/novelty exploration task, non-aggressive social interaction was scored (experimenter blind to genotype and treatment group) when the experimental mouse initiated the action and when the nose of the animal was oriented toward the social stimulus mouse only. A non-aggressive interaction only initiated and maintained by the social stimulus mouse was not scored. During the object habituation/novelty exploration, the interaction was scored when the nose of the animal was oriented toward the object stimulus. The time interaction was used to calculate the Novelty Index as: !"#$%&'#()"!!"# ! − !"#$%&'#()"!!"# ! , both for social and object habituation/novelty exploration task.
The experimental cage was cleaned with 5% ethanol solution and the bedding was changed between sessions.
For the experiments with pharmacological agents, mice were cannulated to allow the infusion of either saline or 1-Naphthylacetyl spermine trihydrochloride (NASPM), directly in the VTA. The habituation task was performed as previously described. NASPM or saline were infused using a Minipump injector (pump Elite 11, Harvard apparatus, US) with 500 nL of saline (2 minutes of active injection at 250 nL/min rate, and 1 minute at rest), 10 minutes before each trial. At day 1, mice received saline. From day 2 to day 4 of the habituation phase, mice received either 4 µg of NASPM dissolved in 500 nL of saline or 500 nL of saline only (at 250 nL/min) before each trials. This dose has been previously used to obtain GluA2-lacking AMPARs block in vivo 89 . After at least 1 week, the animals were re-tested to habituation/novelty and the pharmacological treatment was counterbalanced. The ! 28! scoring of the social or object interaction was made as previously described. The experimental cage was cleaned with 5% ethanol solution and the bedding was changed after every session. To assess the cannula placement, experimental subjects were infused using Chicago Sky Blue 6B (1 mg/mL), sacrificed 1 -2 hours later and transcardially perfused as previously described. At day 5, during the Post-TEST, experimental mice freely explored the CPP apparatus, without any stimulus for 15 minutes and the preference score was measured. The CPP apparatus was cleaned with 1% acetic acid, rinsed with distilled water and dried between each experimental subject.

Social conditioned place preference
Mice were tested at P30 -P45 and were group housed before the test. The test apparatus was a custom-built cage measuring 46x24x22 cm divided into three chambers. The two outer chambers (23x18x22 cm) had vertical or horizontal striped pattern on the walls and flooring consisting or black rubber mats with different patterns (stripes vs squares). The outer chambers were joined together by a smaller chamber (23x10 cm) with white walls and floor with a 7x7 cm opening at the base to the outer chambers that can be closed. The cage was cleaned with 70% ethanol between each trial. During the pre-trial, mice were left to freely explore the cage for 30 minutes. After the pre-trial, all mice were single housed for the remainder of the test and one chamber was assigned the social chamber and one the isolation chamber.
All mice received one social and one isolation condition session (30 minutes each) per day for 4 days, with a two-day rest between the 2 nd and 3 rd conditioning day. Mice were socially conditioned for 30 minutes together with their cage-mates followed by conditioning in the isolation chamber for 30 min. After the 4th conditioning day, mice were tested in a 30 minutes post-conditioning trial. The time spent freely exploring the chambers for 30 minutes was manually scored by an investigator blinded to the genotype. The preference score was calculated as the time spent in the social chamber divided by the combined time spent in the social and isolation chamber. Animals were excluded by pre-established criteria if they exhibited a strong preference for one chamber (more than 2x preference for one chamber).

Olfactory habituation/dishabituation test
The olfactory habituation/dishabituation test was performed as previously described 52

Open field, object recognition task, and marble burying
On day 1, mice were placed individually in the center of a square open field arena (50x50x30 cm) made of grey plastic for 7 minutes. Velocity (cm.sec -1 ) was analyzed using EthoVision10 system (Noldus). The arena was cleaned with 70% ethanol between trials. 24 hours later, mice were placed back in the arena containing two identical objects (culture flask filled with sand) for a 5-minutes acquisition trial.
Object recognition memory was tested 1 hour later during a 5-minutes test trial in the arena containing a familiar and novel object (Lego block). The trial was recorded with a video camera and the time spent investigating was scored manually, the experimenters were blinded to the genotype. Investigation of the object was considered when the mouse nose was sniffing less than a centimeter from or touching the object. The discrimination ratio was calculated as following: .   Imaging System and Li-Cor, Odyssey) and images were analyzed using ImageJ.

Statistical analysis
No statistical methods were used to predetermine the number of animals and cells, but  interaction. Data are represented as the mean ± s.e.m. and the significance was set at P < 0.05.

Statistical Analysis of the 3-chamber task
According to the original developer of the 3-chamber task, this assay is a yes-or-no test in which animals display sociability/social novelty or they do not 90  For the time sniffing, we performed a RM two-way ANOVA followed by Bonferroni post-hoc test. By performing this within-group analysis, we considered the animals to express sociability or preference for social novelty if they spent more time in the social/novel social chamber compared to the object/familiar mouse chamber and if they were engaged for longer time in social interaction or novel social interaction compared to object and familiar stimulus interaction, respectively 90 . Additionally, we ! 36! calculated social novelty index, or preference score S2-S1 or S4-S3, to allow between group comparisons 42 after a significant RM two-way ANOVA (P< 0.05 for main effects and interaction) followed by Bonferroni post-hoc test.!

Data availability
The data supporting this study are available upon request to the corresponding author.

Fig.1c
Friedman test followed by Dunn's test for planned multiple comparisons Fig.1d Wilcoxon test Fig.1e RM two-way ANOVA followed by Bonferroni post-hoc test Fig.1f Kruskal-Wallis test followed by Dunn's multiple comparisons test Fig.2b RM one-way ANOVA followed by Holm-Sidak post-hoc test for planned comparisons Fig.2c RM one-way ANOVA followed by Holm-Sidak post-hoc test for planned comparisons Fig.2d RM one-way ANOVA followed by Holm-Sidak post-hoc test for planned comparisons Fig.2e RM two-way ANOVA followed by Bonferroni post-hoc test Fig.2f RM two-way ANOVA followed by Bonferroni post-hoc test Fig.2g RM two-way ANOVA followed by Bonferroni post-hoc test Fig.2h One-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.2i RM two-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.3c Paired t-test Fig.3d Paired t-test Fig.3e Paired t-test Fig.3f Friedman test Fig.3g Friedman test Fig.3h Friedman test Fig.3i Kruskal-Wallis test followed by Dunn's test for planned comparisons Fig.3k RM two-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.4b RM two-way ANOVA followed by Bonferroni post-hoc test Fig.4c Unpaired t-test Fig.4d RM two-way ANOVA followed by Bonferroni post-hoc test Fig.4e Mann-Whitney test Fig.4g Paired t-test Fig.4h Unpaired t-test Fig.4i Unpaired t-test Fig.4j Unpaired t-test Fig.4l RM two-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.5d RM two-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.5f RM two-way ANOVA followed by Bonferroni post-hoc test Fig.5g Unpaired t-test Fig.5i Paired t-test Fig.5j Unpaired t-test Fig.5k Mann-Whitney test Fig.5l Mann-Whitney test Fig.6a One-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.6b One-way ANOVA Fig.6c One-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.6d One-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.7c RM two-way ANOVA Fig.7d RM two-way ANOVA Fig.7g Mann-Whitney test ! 37!

Fig.7h
RM two-way ANOVA Fig.8a One-way ANOVA followed by Bonferroni post-hoc test Fig.8b Unpaired t-test Fig.S1b RM two-way ANOVA Fig.S1c One-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.S1d Paired t-test Fig.S1e RM two-way ANOVA Fig.S1f One-way ANOVA Fig.S2a RM one-way ANOVA followed by Holm-Sidak post-hoc test for planned comparisons Fig.S2b RM one-way ANOVA followed by Holm-Sidak post-hoc test for planned comparisons Fig.S2c RM one-way ANOVA followed by Holm-Sidak post-hoc test for planned comparisons Fig.S2d RM two-way ANOVA followed by Bonferroni post-hoc test Fig.S2e RM two-way ANOVA followed by Bonferroni post-hoc test Fig.S2f RM two-way ANOVA followed by Bonferroni post-hoc test Fig.S2g One-way ANOVA followed by Bonferroni post-hoc test for planned comparisons Fig.S3a RM two-way ANOVA by both factor Fig.S3b RM two-way ANOVA by both factor followed by Bonferroni post-hoc test Fig.S3c RM two-way ANOVA by both factor Fig.S3d RM two-way ANOVA by both factor followed by Bonferroni post-hoc test Fig.S3e RM two-way ANOVA by both factor Fig.S4a Friedman test followed by Dunn's post-hoc test for planned multiple comparisons Fig.S4b RM one-way ANOVA followed by Bonferroni post-hoc test for planned multiple comparisons Fig.S4c Wilcoxon test Fig.S4d Paired t-test Fig.S4e RM two-way ANOVA for within and between genotype followed by Bonferroni post-hoc test Fig.S4f RM two-way ANOVA by both factor followed by Bonferroni post-hoc test Fig.S4g RM two-way ANOVA by both factor followed by Bonferroni post-hoc test Fig.S5d RM two-way ANOVA followed by Bonferroni post-hoc test Fig.S5e Unpaired t-test Fig.S5f RM two-way ANOVA by both factor followed by Bonferroni post-hoc test Fig.S5g RM two-way ANOVA by both factor followed by Bonferroni post-hoc test Fig.S5h Friedman test followed by Dunn's post-hoc test for planned multiple comparisons Fig.S5i RM one-way ANOVA followed by Bonferroni post-hoc test for planned multiple comparisons Fig.S5j Paired t-test

Fig.S5k
Paired t-test Fig.S5l RM two-way ANOVA for within and between virus injection followed by Bonferroni posthoc test Fig.S6c RM two-way ANOVA followed by Bonferroni post-hoc test Fig.S6d Mann-Whitney test Fig.S6e Friedman test followed by Dunn's post-hoc test for planned multiple comparisons Fig.S6f RM one-way ANOVA followed by Bonferroni post-hoc test for planned multiple comparisons Fig.S6g Wilcoxon test Fig.S6h Paired t-test Fig.S7a Unpaired t-test Fig.S7b One-way ANOVA Fig.S7d Wilcoxon test                    b a n a n a 1 b a n a n a 2 b a n a n a  b a n a n a 1 b a n a n a 2 b a n a n a