NaV1.2 haploinsufficiency in Scn2a knock-out mice causes an autistic-like phenotype attenuated with age

Mutations of the SCN2A gene, encoding the voltage gated sodium channel NaV1.2, have been associated to a wide spectrum of epileptic disorders ranging from benign familial neonatal-infantile seizures to early onset epileptic encephalopathies such as Ohtahara syndrome. These phenotypes may be caused by either gain-of-function or loss-of-function mutations. More recently, loss-of-function SCN2A mutations have also been identified in patients with autism spectrum disorder (ASD) without overt epileptic phenotypes. Heterozygous Scn2a knock-out mice (Scn2a+/−) may be a model of this phenotype. Because ASD develops in childhood, we performed a detailed behavioral characterization of Scn2a+/− mice comparing the juvenile/adolescent period of development and adulthood. We used tasks relevant to ASD and the different comorbidities frequently found in this disorder, such as anxiety or intellectual disability. Our data demonstrate that young Scn2a+/− mice display autistic-like phenotype associated to impaired memory and reduced reactivity to stressful stimuli. Interestingly, these dysfunctions are attenuated with age since adult mice show only communicative deficits. Considering the clinical data available on patients with loss-of-function SCN2A mutations, our results indicate that Scn2a+/− mice constitute an ASD model with construct and face validity during the juvenile/adolescent period of development. However, more information about the clinical features of adult carriers of SCN2A mutations is needed to evaluate comparatively the phenotype of adult Scn2a+/− mice.

Then, we investigated social behavior in freely moving pairs of unfamiliar male mice. In the reciprocal social interaction test with young mice, no significant differences were found between genotypes in the total time spent in social interaction initiated by the test mouse toward an unfamiliar wild-type stimulus mouse during the 10-min session (unpaired Student's t-test, t(29) = 0.54, p = 0.5952; Fig. 1I). We further analyzed social interactions by bins of 2 min. Both genotypes exhibited a gradual reduction in the amount of time engaged in social interaction over time (two-way repeated measures ANOVA, F(1, 29) = 0.736, p = 0.3981 for genotype; F(4, 116) = 80.7, p < 0.0001 for time; F(4, 116) = 3.03, p = 0.0202 for genotype x time interaction; Fig. 1J). However, the significant genotype by time interaction indicates that the time course of social interactions is genotype specific. Indeed, as can be seen in Fig. 1J, young male Scn2a +/− mice spent less time in social interaction compared to wild-type littermates during the first two minutes of the session. Heterozygous mice, however, did not differ significantly from wild-type mice in the latency to initiate the first social contact (unpaired Student's t-test, t(29) = −1.44, p = 0.1601; Fig. 1K). In contrast, adult male Scn2a +/+ and Scn2a +/− mice did not show any significant differences in the amount of time engaged in social interactions initiated by the test mouse toward the wild-type stimulus mouse during the 5-min session (unpaired Student's t-test, t(29) = −0.80, p = 0.4304; Fig. 1L) or the latency to establish the first social contact (unpaired Student's t-test, t(29) = −0.12, p = 0.9051; Fig. 1M). The time spent in social interaction during the first 2 min of the test was 32.19 ± 3.04 s for Scn2a +/+ and 38.45 ± 3.85 s for Scn2a +/− adult mice.
Stereotyped and repetitive behaviors are present in young but not adult Scn2a +/− mice. A further criterion for the diagnosis of ASD is the presence of restricted, repetitive and stereotyped patterns of behavior 27 . We therefore investigated if young male Scn2a +/− mice display stereotyped motor movements such as self-grooming behavior. Figure 3A shows that young Scn2a +/− mice spent significantly more time in self-grooming compared to young Scn2a +/+ mice during a 10-min session in an open-field arena (unpaired Student's t-test, t(29) = −2.24, p = 0.0328). We also used the marble burying test (Fig. 3B), in which young Scn2a +/− mice showed a significant increase in the number of buried marbles compared to wild-type littermates (unpaired Student's t-test, t(29) = −2.26, p = 0.0313), further supporting the presence of repetitive and perseverative behavior in these mutant mice 29 . (A) Time spent sniffing social or non-social odors (coffee) (with time sniffing water substracted) and total time investigating the three odors (water, coffee, social) in young mice. (B) Olfactory preference and total time spent sniffing the three odors in adult mice. (Paired t-test: ### p < 0.0005, #### p < 0.0001 vs coffee odor for WT and Scn2a +/− mice; n = 14/genotype).
Altogether, these results demonstrate that repetitive stereotyped behaviors are key features of young but not adult Scn2a +/− mice.
Hyporeactivity to negative emotional stimuli is observed in young but not adult Scn2a +/− mice. Because anxiety and depression are among the most frequently diagnosed comorbidities in patients with ASD 34 , we assessed anxiety-and depression-related behavior in Scn2a +/− mice. In the open-field test (Fig. 4A), the total distance traveled by young male Scn2a +/− mice was not significantly different from that of control Scn2a +/+ littermates (Mann-Whitney test, U = 102, p = 0.5125), reflecting a normal exploratory behavior in heterozygous mice. However, young Scn2a +/− mice showed a significant increase in the distance traveled (unpaired Student's t-test, t(29) = −2.43, p = 0.0214) and time spent (Mann-Whitney test, U = 54, p = 0.0105) in the central area, along with a decrease in the time spent in the periphery (Mann-Whitney test, U = 180, p = 0.0163), compared to wild-type mice (Fig. 4B). In the elevated plus maze test (Fig. 4C,D,K), young Scn2a +/− mice displayed a significant higher percentage of time spent (Welch's t test, t(19) = −4.13, p = 0.0005) and number of entries (Mann-Whitney test, U = 6, p < 0.0001) in the open arms compared to wild-type littermates. As in the open-field test, no difference in locomotor activity was observed between the two genotypes as shown by the total distance www.nature.com/scientificreports www.nature.com/scientificreports/ traveled as well as the total number of entries in closed/open arms (Table 1). In the tail suspension test (Fig. 4I), young Scn2a +/− mice spent significantly less time immobile than WT mice during the 6-min session (two-way repeated measures ANOVA, F(1, 28) = 13.55, p = 0.0010 for genotype; F(2, 56) = 12.26, p < 0.0001 for time; F(2, 56) = 3.23, p = 0.0305 for genotype × time interaction; Bonferroni's post hoc test, 2-4 min: p = 0.0051; 4-6 min: p = 0.0023), suggesting that young heterozygous mice are less resigned. Taken together, these results indicate that young Scn2a +/− mice display reduced emotional reactivity to negative stimuli.
In adulthood, male Scn2a +/+ and Scn2a +/− mice showed similar exploratory/locomotor behavior in the open-field (Mann-Whitney test, U = 78, p = 0.1079; Fig. 4E) and elevated-plus maze tests (Table 1) Table 1). Similarly, in the tail suspension test (Fig. 4J), Scn2a +/+ and Scn2a +/− adult mice did not show any significant differences in the time spent immobile, although a tendency to decreased immobility time was observed for Scn2a +/− mice at the end of the 6-min session (two-way repeated measures ANOVA main, F (1,26)   www.nature.com/scientificreports www.nature.com/scientificreports/ Taken together, these data indicate that the reduced sensitivity to stressful stimuli found in young Scn2a +/− mice does not persist in adulthood. Importantly, it is unlikely that the hyporeactivity to negative emotional stimuli can contribute to the reduced social communication and stereotyped behaviors that we observed in young mutant mice since, on the contrary, increased anxiety results in lower social interactions and excessive motor stereotypies 29,35,36 . Young Scn2a +/− mice display impaired memory functions whereas adult Scn2a +/− mice show a tendency for slower spatial learning abilities. About 45% of subjects with autistic disorders have some degree of intellectual disabilities 34,37 . We therefore evaluated the presence of cognitive deficits in young and adult Scn2a +/− mice using different memory tests. We first assessed spontaneous alternation behavior in a Y maze. Young Scn2a +/− mice displayed a decrease in the percentage of spontaneous alternations at borderline significance compared to wild-type mice (unpaired Student's t-test, t(29) = 1.96, p = 0.0593; Fig. 5A), without change in the total number of visits (unpaired Student's t-test, t(29) = 0.83, p = 0.4134; Fig. 5B), suggesting that there are no major or very mild impairments in spatial working memory in heterozygous mice. In the novel object recognition task (Fig. 5C), young Scn2a +/+ mice spent more time exploring the novel object over the familiar object during the 5-min retention test, as indicated by a discrimination index higher than chance level (one sample t-test, t(14) = 4.08, p = 0.0011). In contrast, the discrimination index was not above chance level for young Scn2a +/− mice (one sample t-test, t(11) = 1.39, p = 0.1907), which failed to display a preference for the novel object compared to wild-type littermates (unpaired Student's t-test, t(25) = 2.13, p = 0.0435). The total time spent exploring the novel and the familiar objects did not differ between genotypes ( Table 2), indicating that the reduced discrimination index observed in young Scn2a +/− mice was not related to a decrease in exploration, in accordance with the normal exploratory behavior that we observed in the open-field test.
Overall, these data suggest that young Scn2a +/− mice exhibit impairment in recognition memory and a tendency to reduced spatial working memory, while adult heterozygous mice have intact spatial working and long-term memory, as well as intact object recognition memory, showing only a tendency to slower spatial learning.

Discussion
SCN2A mutations have been identified as one of the most significant risk factors for ASD 8,38 . It has been proposed that these mutations, which are often missense, cause nearly complete loss-of-function of Na v 1.2 15 . Our study demonstrates that Na v 1.2 haploinsufficiency in mice -modeling heterozygous Na v 1.2 complete loss-of-function associated with mutations identified in humans-causes a self-limited autistic-like phenotype, which is attenuated with age. Thus, young Scn2a +/− mice exhibit deficits in social communication and interaction as well as repetitive stereotyped behaviors, recapitulating core symptoms of ASD, whereas adult heterozygous mice display only reduced social communication. In addition to autistic-like behaviors, young mutant mice display impaired memory and low reactivity to stressful stimuli, alterations that in our study do not persist in adulthood.   Table 2. Additional measures of exploration in the novel object recognition task. Comparisons between genotypes were performed using unpaired Student's t-test.
The genetic studies that identified loss-of-function SCN2A mutations associated with ASD without overt epilepsy were performed on patients aged from 3 to 17 years 7-9,16,39-41 . Among these studies, a few reported other symptoms than the core ASD features. Intellectual disability, ranging from moderate to severe, was the most frequently observed comorbidity in these children or adolescents with ASD 7,9,16,40,41 . Abnormal gait and/or fine coordination was observed in two autistic children 9,40 , attention deficit hyperactivity disorder in one adolescent 41 , and obsessive-compulsive disorder associated to anxiety in one child with Asperger's syndrome 40 . The autistic-like behaviors and comorbid cognitive deficits that we observed in Scn2a +/− mice during the juvenile/adolescent period of development are therefore consistent with the symptoms described in these young patients. However, no clinical information, to our knowledge, is available about adult carriers of SCN2A mutations suffering of non-syndromic ASD. We could not, therefore, evaluate whether adult Scn2a +/− mice exhibit behavioral phenotype similar to those observed in humans, especially the attenuation of autistic symptoms. Of particular interest, autistic symptoms follow heterogeneous developmental trajectories (six patterns were identified by Fountain et al. 42 ) www.nature.com/scientificreports www.nature.com/scientificreports/ and good outcome has been described in adulthood for a subset of individuals diagnosed with autism in childhood 42-45 . Such longitudinal studies in patients with loss-of-function SCN2A mutations may be useful to correlate the mild phenotype that we observed in adult heterozygous mice.
Young Scn2a +/− mice display strong deficit in social communication but mild impairment in social interaction associated to several motor stereotypies. It is not excluded that, in other experimental conditions (long time isolation before test, pairs of mutant mice) or in other tests like the three-chamber sociability test, social interaction deficits might be more pronounced. Importantly, if lack of social interactions is a symptom shared by other disorders such as schizophrenia or depression, reduced social communication associated to repetitive behaviors early in childhood are specific to ASD. Dysfunction in striatal circuits, especially cortico-striatal pathways, has been postulated to underlie autistic behaviors 46,47 . A marked reduction in Na v 1.2 expression in these circuits during development might therefore contribute to the autistic phenotype seen in young Scn2a +/− mice. Interestingly, it has been recently shown that cortico-striatal pathways are involved in the generation of absence-like seizures in Scn2a +/− mice 24 .
Intriguingly, young mutant mice show reduced anxiety in the open-field and plus maze tests along with decreased resignation in the tail suspension test. Autistic patients often exhibit anxiety or depression and some ASD mouse models, such as Shank3B −/− , Chd8 +/− or En2 −/− mice, display anxiety-like 33,48 or depression-like 49 phenotypes. Interestingly, our results are very similar to those described in BTBR mice 50,51 . This inbred strain with autistic-like behaviors displays lower levels of anxiety in the elevated plus maze or zero maze tests (although these results have not been always replicated), and low depression-related behavior in the tail suspension test. These behaviors were associated to alterations in the hypothalamic-pituitary-adrenal (HPA) axis. It is therefore tempting to hypothesize that young Scn2a +/− mice have similar altered HPA axis activity which could explain their lower sensitivity to stressful stimuli. Altered HPA axis activation might be the consequence of decreased Na v 1.2 expression in hypothalamic or limbic circuits involved in the regulation of this axis, such as activatory pathways from the amygdala, where Na v 1.2 channels are expressed 23 . Thus, decreased activity of afferents from the amygdala would reduce the HPA axis response to stressful environments. Of particular interest, emotional reactivity to negative stimuli early in life is predominantly influenced by the amygdala and is regulated at later stages of development by the prefrontal cortex, which can exert top-down control on the amygdala, leading to improvements in emotional regulation and anxiety 52,53 .
In addition to lower reactivity to stressful stimuli, young Scn2a +/− mice display cognitive deficits as evidenced by altered mnesic functions. The mildly reduced spatial working memory (borderline significance) and the impaired object recognition memory suggest dysfunctions in hippocampal and prefrontal or perirhinal cortex circuits, respectively 54,55 . These memory tests are based on the spontaneous tendency of rodents to explore novelty. The deficits observed are unlikely due to neophobia since young mutant mice did not show any decrease in overall exploration of the maze or objects in these tasks. On the contrary, lack of interest for novelty, which may resemble to restricted interests described in ASD patients, might contribute to memory deficits. Alternatively, alteration in attentional processes in young Scn2a +/− mice may interfere with memory performance. Indeed, attention-deficit hyperactivity disorder (ADHD) is a frequent comorbid trait in ASD patients 34,37 and has been reported in one autistic adolescent carrier of loss-of-function SCN2A mutation 41 .
Unexpectedly, in adulthood, much milder behavioral abnormalities were observed in Scn2a +/− mice, since only social communication was significantly reduced. Mutant mice also showed a tendency for slower spatial learning in the Barnes maze, consistent with the delayed spatial learning reported recently by Middleton et al. 25 in the same test. Importantly, adult heterozygous mice do not display impairments in memory performance in this task and the other memory tasks used (i.e., spontaneous alternation and object recognition). A tendency to decreased immobility time in the tail suspension test was also detected in adult mutant mice in agreement with the very recent study of Tatsukawa et al. 26 . The absence of social behavior deficits that we observed in the direct social interaction task is further consistent with the results reported in that study. However, Tatsukawa et al. 26 showed that adult heterozygous mice exhibited a mild hyperactivity in the open-field test, a lower level of anxiety in the plus maze test (but the contrary in the light-dark box test) and a reduced performance in the rotarod task, abnormalities that we did not detect using the same tests. The differences between our results and those of Tatsukawa et al. 26 may be explained by different housing conditions (single versus group-housed mice with mixed genotypes in the present study), experimental conditions (luminosity, duration of tests) and protocols used (one day with 6 trials versus 2 days tests preceded by a training session for the rotarod test in our study) 56 . Another known source of variability in the outcome of behavioral tests is the laboratory environment, especially for the open-field and plus maze tests 57,58 .
The distinct phenotype observed between young and adult Scn2a +/− mice suggest that the abnormal behaviors observed in young mutant mice can be rescued to some extent in adulthood by compensatory mechanisms that might emerge at the end of the juvenile/adolescent period of development (after P44). Notably, this compensation should not be directly linked to the physiological switch between Nav1.2 and Nav1.6, because it occurs at around P20 in mouse brain 13 , and could instead be linked to other more complex homeostatic mechanisms triggered by Nav1.2 haploinsufficiency. Indeed, when activity is perturbed in neuronal networks, different homeostatic mechanisms that attempt to restore the initial condition can take place 59 . Decreased expression of Na v 1.2 channels may result in alterations in the excitatory/inhibitory (E/I) balance, a central concept in ASD etiology 60,61 . Nelson and Valakh 62 have proposed that homeostatic mechanisms which take place in ASD during development in response to E/I imbalance in some circuits may be insufficient and can become maladaptive amplifying the imbalance in other circuits. In Scn2a +/− mice, compensatory mechanisms might be relatively efficient to restore behavioral dysfunctions at adulthood. Of note, compensatory processes in ASD patients during development are thought to underlie the reduced severity of autistic symptoms reported in some individuals at adulthood 63 . Alternatively, adult Scn2a +/− mice may display a mixed phenotype relevant to both autism and schizophrenia, as proposed by Tatsukawa et al. 26 . Indeed, both disorders share some clinical manifestations, such as deficits in social communication, but schizophrenia typically emerges in late adolescence or early adulthood. Moreover, schizophrenia and autism can overlap in some adult patients that tend to display an atypical phenotype [64][65][66][67] , and, more importantly, several studies have reported loss-of-function SCN2A mutations/variants in schizophrenic patients 10,38,68 . Unfortunately, no detailed clinical information is available about these patients.
In summary, our data demonstrate that Na V 1.2 haploinsufficiency leads, in young male mice, to social, communicative, motor, cognitive and emotional dysfunctions that are relevant to ASD and comorbid phenotypes. By contrast, adult mice show only communicative deficits. There are similarities between the phenotype of young Scn2a +/− mice and that of some young autistic patients with SCN2A mutations, indicating that these mice can be an ASD model with construct and face validity during the juvenile/adolescent period of development. However, it is not possible to correlate the phenotype of adult Scn2a +/− mice with that of adult ASD patients that carry SCN2A mutations because, in our knowledge, information about clinical features of these patients is not available. Further investigations in Scn2a +/− knockout mice at the cellular and circuit level may provide important new insights into mechanisms contributing to these phenotypes and possible homeostatic responses that can attenuate their severity.

Materials and Methods
Animals. Generation and genotyping of Scn2a +/− and Scn2a −/− mice have been described by Planells-Cases et al. 69 . Original heterozygous breeders were provided by Dr. E. Glasscok of Louisiana State University Health Sciences Center, Shreveport 70 . The heterozygous Scn2a +/− and control Scn2a +/+ littermate mice that we used in the experiments were generated from crosses between heterozygous Scn2a +/− males and wild-type females on a C57Bl6/J background (Charles Rivers, France). All mice were housed in a controlled environment (22-24°C, 40-50% humidity) under a 12 h dark/light cycle with access to food and water ad libitum. After weaning (post-natal days 21-23), mice were group-housed with same sex littermates of mixed genotype (n = 3-5/cage) in standard polycarbonate cages containing plastic houses and nesting material. Only male mice were used for the behavioral characterization of Scn2a knock-out mice, since ASD affects males more frequently than females. Behavioral tests were performed sequentially in young (between P22 and P44, corresponding to the juvenile (P21-28) and adolescent (P29-45) periods 71 ) and adult (between P60 and P95) male mice. For all experiments, the animals were habituated to the testing room at least 30 min before the beginning of tests. Heterozygous Scn2a +/− mice (n = 14) and their control littermates (n = 17) (from a total of 7 litters) were submitted to the behavioral tasks in the following order: anxiety tests, reciprocal social interaction, marble burying test, ultrasonic vocalizations, memory tests, rotarod and tail suspension test. All the tests were conducted in young and adult mice except one memory task, the Barnes maze test, which was not evaluated in young mice in order to perform all the behavioral experiments before the end of the adolescent period. An additional group of male mice (from a total of 5 litters) was used to assess olfactory function in young then adult mice (n = 14 for each genotype). All procedures were performed in accordance with the European Council Directive

Ultrasonic vocalizations (USV).
To investigate mice social communication, USV were recorded in a sound-attenuating chamber using an ultrasound microphone (CM16/CMPA, Avisoft bioacoustics, Glienicke, Germany) positioned above the cage. In juvenile mice, pairs of same age, sex and genotype from different litters were simultaneously introduced into the cage after 1 h of social isolation and vocalizations were recorded for 5 min 72 . In adult mice, courtship USV were emitted by male after placing a C57BL6/J estrus female in the recording chamber during 3 min 73 . USV spectrograms were generated using Avisoft SASLab Pro software (Avisoft bioacoustics). The number, duration and peak frequency of ultrasonic calls were analyzed.
Olfaction test. Since olfactory impairment could interfere with the social behaviors measured in male mice, olfactory discrimination between social and non-social scents was assessed as previously described with minor modifications 74 . Squares of filter paper (5 × 5 cm) were soaked with water or coffee (non-social scent) (approximately 0.5 ml of solution). Social odors were generated after placing filter paper squares during 3 days in cages housing non familiar juvenile males or adult females and used to test young or adult male mice, respectively. Mice were placed in clean plexiglas cages without bedding for 10 min of habituation. Then, scents were introduced sequentially (water, coffee, social odor) for 3 min each, spaced by 1 min interval. Olfactory preference corresponds to the difference between the time spent sniffing social or non-social scents and that spent sniffing water. Social odors are attractive whereas coffee is aversive.

Reciprocal social interaction test and self-grooming behavior. Social interaction was evaluated in
freely moving pairs of mice as previously described with slight modifications 75 . The test mouse was introduced in the open-field arena and allowed to explore for 20 min. The time spent in spontaneous repetitive grooming behavior was measured during the last 10 min. Then, an unfamiliar, age-matched, C57Bl6/J male wild-type stimulus mouse was placed in the arena and the behavior of the pair of mice was videotaped for 5 (adult) or 10 (juvenile) min. Total time spent in social interactions (such as nose-to-nose or anogenital sniffing, following or mounting) initiated by the test mouse toward the stimulus mouse was measured manually by the same experimenter, who was blind to the genotype of test mice. The latency for the first social contact was also analyzed.

Marble burying task.
Repetitive behavior related to digging was assessed using the marble burying test as described by Thomas et al. 28 . Juvenile or adult mice were placed into a transparent Plexiglas box (30 × 30 × 30 cm) or a polycarbonate cage (57 × 20 × 40 cm) respectively, containing a 5 cm layer of fresh bedding (poplar wood chips, average particle size 3.5-4.5 mm; Lignocel Select, SAFE, France). Twenty marbles were disposed in five (2019)  www.nature.com/scientificreports www.nature.com/scientificreports/ spaced rows of four marbles on the bedding surface. The number of marbles with more than 50% of their surface covered were scored as buried during a 20 min session. The number of buried marbles was continuously scored during the test by the same investigator blind to the animal genotype.
Rotarod. Motor coordination, balance and motor learning were assessed using an accelerating rotarod (Bioseb-In vivo Research Instruments,Vitrolles, France). Mice were habituated to the procedure before the test phase: each mouse trained until they were able to stay 1 min on the circular rod at a constant speed of 4 rpm. Mice were then placed on the rod that accelerated from 4 to 40 rpm over 5 min. Mice were given three trials a day for 2 consecutive days with an inter-trial interval of 30 min 49 . Latency to fall from the rotating rod was automatically recorded.
Anxiety tests: Open-field and Elevated-plus maze tests. The open-field test is commonly used to assess exploratory behavior and anxiety in rodents. The white and opaque arena (40 × 40 × 30 cm) was brightly illuminated at 300 lux. Mice were allowed to explore the arena for 30 min. The distance traveled and the time spent in the periphery (6 cm from the walls) and the center (20 × 20 cm) of the open-field as well as the total distance in the entire arena were automatically measured using ANY-maze video-tracking system (Stoelting Europe, Dublin, Ireland).
Anxiety behavior was also evaluated using the elevated-plus maze test. The maze, slightly illuminated (30 lux), consisted of two open arms (25 × 5 × 2.5 cm) and two enclosed arms (25 × 5 × 25 cm) elevated 40 cm above the floor. Mice were placed in the center of the maze and allowed to explore the maze for 5 min. An arm entry was defined as all 4 paws of the mouse in an arm. The number of entries and the time spent in the open and the closed arms was determined using ANY-maze video-tracking system.

Tail suspension test (TST).
Responsiveness to an inescapable stressful situation was measured using the TST test. Mice were suspended by the tail to a hook using adhesive tape. The time spent immobile, thought to reflect a resignation state, was recorded during a 6-min session through an automated device (Bioseb). Mice which climbed their tail were removed from analysis. Spontaneous alternation behavior. Spatial working memory was assessed in a symmetrical grey plastic Y maze (30 × 6 × 15 cm). Mice were placed at the end of one arm and allowed to freely explore the maze for 8 min. The sequence and the number of arm entries were manually recorded. An alternation was defined as consecutive entries into the three different arms (A, B or C) on overlapping triplet sets (ex: ABCABBACAB = 5). The alternation score (%) was calculated as the ratio of actual (total alternations) to possible (total arm entries-2) alternations x100.
Novel object recognition task. This memory test based on the spontaneous tendency of rodents to explore novelty was performed as previously described 76 in an open-field apparatus illuminated at 30 lux. During the phase of habituation (Day 1), mice were allowed to freely explore the arena during 5 min and then to explore two identical objects for 10 min. On experimental day (Day 2), mice were submitted to two trials with an intertrial interval of 1 hour. During the first trial (acquisition phase), mice were placed in the open-field in the presence of two identical objects (different from Day 1) and stayed the time necessary to explore for 20 s the two objects. Mice that did not explore the objects within a 12 min period were excluded. During the second trial (retention test), one of the objects was replaced by a novel object and mice were allowed to explore both objects (novel and familiar) for 5 min. Mice that did not explore the objects for a minimum of 15 s (young) or 20 s (adult) were excluded. The pairs of objects used were different in shape, size (height between 12 and 19 cm) and texture (glass, plastic or metal). The two objects were placed symmetrically in the center of the arena about 6 cm away from the wall. The location of the novel object was counterbalanced across animals (right or left side). Recognition memory was evaluated using a discrimination index (DI). This index corresponds to the difference between the time spent exploring the novel object and the familiar object divided by the total time of exploration of both objects and is expressed in percentage.

Barnes maze test.
Spatial learning and memory were tested using the Barnes maze. The maze consisted of a white circular platform (90 cm diameter) with 20 holes (5 cm diameter) along the perimeter. Mice can access to an escape box hidden under one of the hole (target hole). Extra-maze visual cues were placed on the walls of the testing room in order to help the mice to learn the position of the target hole. During the phase of acquisition (day1-4), mice received three trials per day separated by 30 min. At the beginning of each trial, mice were placed in the center of the maze under a rectangular opaque box. The box was removed 10 s after the onset of light (700 lux) and mice were allowed to explore the maze. The trial ended when mice entered the escape box with all four paws or after 3 min have elapsed. In the latter case, mice were gently guided to the escape box by the experimenter and remained there for 1 min. For each trial, the latency to enter the escape box was measured. The day after the acquisition phase (day 5) and one week after (day 12), mice were submitted to a probe trial for 90 s without the escape box. The latency and the distance traveled before finding the target hole as well as the number of visits to each hole were automatically recorded using ANY-maze video-tracking system. The Barnes maze were divided into 4 quadrants: the correct (where the target hole was located), opposite, left and right quadrants. Measurements of time spent in each quadrant were also made on days 5 and 12.
Statistical analysis. Data were analyzed for normality (Shapiro-Wilk test) and equal variances using Origin 8.0 (OriginLab Corporation, Northampton, USA). When data were normally distributed, comparisons between genotypes in young or adult mice were evaluated using unpaired Student's t-test or Welch's t-test (when variances were different), otherwise non parametric tests were applied. When appropriate, a two-way repeated measures