Rapamycin improves social and stereotypic behavior abnormalities induced by pre‐mitotic neuronal subset specific Pten deletion

Abstract The mechanistic target of rapamycin (mTOR) pathway is a signaling system integral to neural growth and migration. In both patients and rodent models, mutations to the phosphatase and tensin homolog gene (PTEN) on chromosome 10 results in hyperactivation of the mTOR pathway, as well as seizures, intellectual disabilities and autistic behaviors. Rapamycin, an inhibitor of mTOR, can reverse the epileptic phenotype of neural subset specific Pten knockout (NS‐Pten KO) mice, but its impact on behavior is not known. To determine the behavioral effects of rapamycin, male and female NS‐Pten KO and wildtype (WT) mice were assigned as controls or administered 10 mg/kg of rapamycin for 2 weeks followed by behavioral testing. Rapamycin improved social behavior in both genotypes and stereotypic behaviors in NS‐Pten KO mice. Rapamycin treatment resulted in a reduction of several measures of activity in the open field test in both genotypes. Rapamycin did not reverse the reduced anxiety behavior in KO mice. These data show the potential clinical use of mTOR inhibitors by showing its administration can reduce the production of autistic‐like behaviors in NS‐Pten KO mice.

diagnosed with ASD. 5 Many of the pathways involved in monogenic ASD have been identified and are considered promising targets of drug development. [6][7][8][9] One such pathway is the mechanistic target of rapamycin (mTOR), which is responsible for regulating cell growth and migration in the context of available cellular resources. Mutations to genes of proteins that regulate the mTOR pathway, such as Tsc1, Tsc2 and the phosphatase and tensin homolog (PTEN) on chromosome 10, have been linked to the development of ASD. [10][11][12] Although the discussion of PTEN mutations in those with ASD were previously only found in case studies, PTEN mutations are now being recognized as a significant contributor in the development of ASD. 11,13 PTEN is a suppressor of the mTOR pathway, important in mediating cell growth and migration during neurodevelopment. 14 Loss of function mutations to PTEN in both humans and rodent models have been associated with ASD and neurological features commonly found comorbid in individuals with ASD, such as macrocephaly and epilepsy. 11,13,[15][16][17] A recent meta-analysis found that 25% of individuals with PTEN mutations were diagnosed with ASD. 11 ASD-like behavior, as well as macrocephaly and epilepsy, have also been recapitulated in a variety of Pten knockout (KO) models. [15][16][17] These models, which vary in the spatial, temporal and quantity of Pten deletions in the brain, exhibit an assortment of ASD-like behaviors similar to the heterogeneity of behavioral phenotypes found in humans. In some rodent models, the suppression of hyperactive mTOR signaling induced by PTEN deletion has shown promise in preventing the development of ASD-like symptoms and commonly found comorbid conditions. [18][19][20][21] However, because of the heterogeneity of ASD, it is important that these results be repeated in all Pten KO models to ensure effective treatments can be identified.
Rapamycin, an allosteric inhibitor of mTOR, can suppress hyperactive mTOR signaling and prevent the development of the autisticlike phenotype in models where Pten deletion occurs after neuronal differentiation. 18 The neuronal subset specific (NS) Pten KO mouse model, however, exhibits Pten deletion prior to neuronal differentiation in primarily granule cells of the hippocampus and cerebellum, as well as some pyramidal neurons throughout the cortex. 22,23 The NS-Pten KO model has been repeatedly found to exhibit deficits in social, repetitive and cognitive behaviors which may represent a more severe phenotype of ASD when compared with other models. 15,[24][25][26][27][28] However, it has not been determined if rapamycin can also prevent the development of ASD-like behaviors in this model. Here, we determine if a 2-week course of rapamycin previously shown to suppress seizures in NS-Pten KO mice, can also prevent the development of ASDlike behavioral deficits.

| Animals
Mice were neuron subset-specific Pten (NS-Pten) KO mice generated using Cre-LoxP (GFAP-Cre; Pten loxP/loxP ) as previously described. 22,23 NS-Pten loxP/+ heterozygote mice were bred to produce NS-Pten +/+ wildtype (WT), NS-Pten loxP/+ heterozygous (HT) and NS-Pten  29 To examine the effects of suppressing hyperactive mTOR activity in NS-Pten KO mice, subjects were administered either rapamycin dissolved in the vehicle solution, the vehicle alone, or were naïve to both the vehicle and treatment. 21 Rapamycin was administered as a 2-week loading dose prior to behavioral testing at 10 mg/kg intraperitoneally for 10 days (Monday through Friday for 2 weeks) beginning at $4 weeks old to male and female, WT and KO mice ( Figure 1). Vehicle and naïve treated mice were collapsed to form the control group after an initial analysis determined a lack of differences between these groups across each test.

| Behavioral testing
Behavioral testing was conducted during the light phase between 8:00 a.m. and 5:00 p.m. Testing began at $6 weeks of age and occurred over 3 weeks (Figure 1). No more than two behavioral tests occurred per week. All testing areas were cleaned with 30% isopropyl alcohol between each testing session.

| Open field
To determine changes to locomotive, stereotyped and anxiety-like behaviors, all subjects underwent the open field test. 29

| Elevated plus maze
The elevated plus maze was used as an additional measure of anxietylike behavior. 30,31 A plus-shaped maze was suspended 40 cm above the ground. Two arms were enclosed with acrylic walls and two arms were unenclosed. Each arm was 30 cm Â 5 cm and stemmed from a 5 cm Â 5 cm central platform. The maze was isolated in a temperature, light and noise-controlled room. Each subject was placed in the center of the maze and allowed to freely explore the maze for 10 min without an experimenter present. Ethovision XT video tracking software (Noldus, Netherlands) was used to record the time each subject spent in the open and closed arms, as well as track the number of entries into the arms.

| Marble burying
As a measure of stereotypic and compulsive behaviors, mice underwent a marble burying task used previously. 15 Mice were placed in a plastic cage containing a grid shaped pattern of 20 black marbles on 3 cm of bedding. The number of marbles at least 75% buried in the bedding after 30 min were compared between groups.

| Social preference
Mice were measured for their preference for social interaction using the three-chambered social preference task modified to include a novel object. 15 During each phase, test mice were placed in the central chamber with the partitions removed and allowed to explore the entire testing apparatus freely for 10 min. The novel mouse and object were placed on different sides for each test subject as a counter-balance measure to prevent side-bias. The time spent in each chamber was recorded in each phase, as well as the time spent at the cups housing the novel mouse and novel object.

| Trace fear conditioning
To determine differences in fear memory, contextual and cued fearbased learning was analyzed using a trace fear conditioning task and Freeze Frame software (Actimetrics, IL, United States). 33   Mauchly's test of sphericity ( p > 0.05). A value of p < 0.05 was considered significant for each statistical test. All data are graphed, and only statistically significant interactions and main effects not within an interaction are reported below. The data that support the findings of this study are available from the corresponding author upon reasonable request.

| Open field
Utilizing a three-way ANOVA, analysis of the distance traveled in the center of the arena showed a main effect of sex, F(1, 109) = 4.72, p < 0.05, (Figure 2A), with females traveling further than males. There was also a two-way interaction between treatment and genotype, F (1, 109) = 5.69, p < 0.05. In the control group, KO mice traveled less distance in the center compared with WT mice, p < 0.001. After treatment with rapamycin, there was no difference between WT mice and KO mice in center distance. The effect appears to be primarily on WT mice, as those treated with rapamycin were found to travel less in the center than controls, p < 0.001. In KO mice, there was no difference between those treated with or without rapamycin. Analysis of the time in the center showed an interaction between sex and genotype, F(1, 109) = 6.90, p < 0.05 ( Figure 2B). KO females spent more time in the center compared with KO males, p < 0.01 and KO males spent less time in the center compared with WT males, p < 0.001. There was also an interaction between treatment and genotype, F(1, 109) = 6.09, p < 0.05 ( Figure 2B). WT mice given rapamycin spent less time in the center compared with WT control mice, p < 0.05, and WT control mice spent more time in the center than KO control mice, p < 0.001. The distance traveled in the surround was investigated and showed a main effect treatment, F(1, 109) = 5.47, p < 0.05. Control mice were found to travel further than treated mice ( Figure 2C). There was also a main effect of sex, F(1, 109) = 8.0, p < 0.01. Females traveled further than males ( Figure 2C). Analysis of the time spent in the surround showed an interaction between sex and genotype, F(1, 109) = 6.28, p < 0.05 ( Figure 2D). KO males spent more time in the surround compared with KO females, p < 0.01. Male KO mice also spent more time in the surround compared with male WT mice, p < 0.001.
There was also an interaction between treatment and genotype, F Rapamycin reduced circling behavior compared with control mice, and females had higher counts of circling than males. There was no main effect of genotype, and no interactions were found.   There was also a main effect of treatment, F(1, 98) = 6.08, p < 0.05, with mice administered rapamycin spending more time in the chamber with the novel mouse than control mice ( Figure 5C). We found no differences in the time spent in the chamber housing the novel object ( Figure 5D).

| Social preference
As a further measure of novel mouse interaction, we also assessed the time spent at the cup housing the novel mouse to determine if mice interacted with the novel mouse or merely preferred this chamber. Our analysis showed a main effect of genotype, F(1, 98) = 10.85, p < 0.01. KO mice spent less time at the cup with the novel mouse than WT mice ( Figure 5E). Analysis of the time spent at the cup housing the novel object showed no differences between any of the groups ( Figure 5F).

| Trace fear conditioning
Three mice were removed from analysis because of death, seizures during training, or equipment malfunction. After removing these mice, the groups sized for the control groups were 14 female WT and 15 female KO mice, 17 male WT and 13 male KO mice. For the rapamycin treated mice there were 13 female WT and 8 female KO mice, 21 male WT and 13 male KO mice. During habituation, mice were placed in the testing chamber without the CS (white noise) or US (foot shock) to determine baseline activity levels and habituate the mice to the chamber. A three-way ANOVA was used to compare the percent of time spent freezing in the chamber. While there was a two-way interaction between genotype and treatment, F(1, 101) = 4.38, p < 0.05, however, after post hoc analyses there were no significant differences ( Figure 6A).
Mice were assessed for cued memory performance using four presentations of the CS in an altered testing chamber 24 h after training. The within-subjects factor, "trial," consisted of the average percent freezing behavior during the baseline, tone and ITIs for each level of the between-subjects factors (sex, genotype and treatment). We analyzed these factors using a repeated measures ANOVA, and analysis of the within-subjects effects showed an interaction between genotype and trial, F(2, 198) = 3.77, p < 0.05. Pairwise comparisons of percent freezing at each trial across genotype showed that KO mice froze more than WT mice at baseline, p < 0.01 ( Figure 6B), but not during the tone ( Figure 6C), or ITI ( Figure 6D). There were no other significant findings. There were also no overall differences in freezing as calculated in the between-subjects effects.
Twenty-four hours after the cued memory test, we placed mice in the original testing chamber to evaluate contextual fear memory. A three-way ANOVA analyzing the differences in percent freezing dur-   18 However, these previous studies used different doses, treatment regimens and background strains. As we did not find a general suppression of activity in any of the subsequent tasks, it is possible that the hypoactivity observed in the open field was only an initial behavioral response to rapamycin treatment. Although the regimen used here suppresses neuronal hypertrophy and seizures in this model, [19][20][21] it may also initially reduce general activity. As in our previous studies, we found KO mice exhibit a small but significant increase F I G U R E 5 Rapamycin induced a general increase in social preference. During the habituation trial, females spent more time in the right chamber than in the left chamber (A, B). When compared with males, females also spent more time in the right chamber (B). Placement of novel mouse and novel object were thus randomized in subsequent trials to mitigate side bias. During the testing phase, more time was spent in the chamber with the novel mouse than in the chamber containing the novel object by both control and rapamycin treated mice (C, D). Rapamycin treated mice spent more time in the novel mouse chamber than control mice (C). While wildtype (WT) mice preferred the chamber with the novel mouse over the chamber with the novel object, (C, D) there were no differences in knockout (KO) mice (C, D). KO mice also spent less time in the chamber with the novel mouse compared with WT mice (C). Analysis of the time spent at the cups shows the duration of time spent at the novel mouse cup was greater than that of the novel object cup in both KO and WT mice (E, F). However, KO mice spent less time with the novel mouse than WT mice (E). Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001. in time spent in the open arms of the elevated plus maze, indicative of reduced anxiety-like behavior. 15,26 Although we found rapamycin had no effect on anxiety-like behavior in the elevated plus maze in either WT or KO mice, we observed a reduction in exploratory behavior in the open field of treated WT but not treated KO mice. As the rapamycin induced reduction in activity is associated with less exploratory behavior in treated WT but not treated KO mice, this side effect may only be found in those with typical, rather than hyperactive mTOR activity.

| Rapamycin corrects stereotypic behavior
Stereotypies are compulsive behaviors that are important to the survival and reproductive success of an animal. Mice bury foreign objects that are potentially harmful, investigate small holes that may contain food, and groom themselves to keep clean. In cases of disease or stress, an aberrant enhancement or suppression of these behaviors can be observed. 42  It is important to consider that rapamycin may increase anxiety like behavior, 43,44 and that marble burying is enhanced when mice are under duress. 42 While untreated KO mice were less active and showed increased grooming behavior in the open field in the current study, rapamycin did not result in anxiety-like behavior in the elevated plus maze or suppress exploratory behavior in the open field task in KO mice. Therefore, it is unlikely that rapamycin-induced anxiety is the cause of the increases in marble burying observed here. KO mice F I G U R E 6 Trace fear conditioning was unaffected by rapamycin. There were no differences in activity levels during habituation. Y axis scale adjusted for visualization of data (A). Knockout (KO) mice spent more time freezing to the novel context compared with wildtype (WT) mice (B) but not during the tone (C) or inter-trial interval (D). In the trained context, mice administered rapamycin froze less than control mice and KO mice spent marginally more time freezing compared with WT mice (E). Data are presented as mean ± SEM. *p < 0.05; **p < 0.01.
have been previously shown to show impaired motor capabilities. 22,25 While marble burying may generally be an effective measure of stereotypic behavior, mice with altered motor activity can confound the outcome of this task. 42 As the stereotypic behavior produced during marble burying in the treated KO mice is improved by rapamycin, it is possible that motor capabilities may be improved and allow for enhanced performance during the marble burying test. Future studies should further examine the possibility that rapamycin treatment may improve the motor capabilities of KO mice.

| Rapamycin increases sociability
Previous studies have showed NS-Pten KO mice exhibit reduced social behavior in both the three-chamber social task and the social partition task, wherein KO mice spent less time than WT mice investigating and interacting with a novel mouse. 15 We were able to reproduce this social impairment also using the three-chamber social interaction task and show that KO mice spend less time in the chamber and at the cup of a novel conspecific. Although the effect of rapamycin on social behavior was not genotype specific, treatment may have improved the social deficits in KO mice. While treated mice did not spend more time at the cup housing the novel mouse, they did spend more time in the novel mouse chamber compared with untreated mice. It is possible that the effects we observed here may have been more robust if social behavior was tested immediately after treatment. In other KO animal models of hyperactive mTOR signaling, such as the NSe-Pten, 18 Tsc, 45 and Cntnap2 40 KO strains, impairments in social behavior can be corrected by inhibiting mTOR signaling with rapamycin. Our data supports mounting evidence that suppression of mTOR hyperactivity via rapamycin may improve social behavior deficits that occur secondary to enhanced mTOR signaling. We add to this body of literature by demonstrating that the social deficits that arise from pre-mitotic deletion of Pten may also be reversed by rapamycin.

| Rapamycin impairs contextual fear memory
We found rapamycin had no impact on freezing in a novel context or to the auditory cue but reduced freezing in both WT and KO mice in the trained context. Contextual fear memories have been shown to rely on proper mTOR signaling in the dorsal hippocampus 46-48 and a single intraperitoneal injection of rapamycin can impair contextual memory up to 3 weeks later. 46,48 Although rapamycin reduced freezing in the trained context, in the current study we found trace fear conditioning still resulted in the acquisition of contextual fear memories. Memory impairments associated with systemic rapamycin typically occurs with acute doses 2-4 times larger than that administered here. 46,48 When rapamycin is administered over repeated sessions at smaller doses, memory may be less effected.
These and our data collectively suggest the memory deficits associated with rapamycin are minor at the 10 mg/kg dose used here.

| KO mice show an enhanced generalization of fear memory
KO mice exhibited an increase in baseline freezing in the novel context after fear training, demonstrating heightened fear memory generalization. Behavioral test batteries increase handling and subject mice to multiple novel test environments, both of which can alter behavioral test performance. 49 Although we ordered the behavioral battery in a way that reduces the impact of multiple testing, it may still have affected the outcome of tests associated with fear-and anxiety-like behavior. The enolase driven NSe-Pten KO mice, which exhibit postmitotic Pten deletion, were reported to be easily agitated, aggressive to handling and sensitive to medication administration. 16,18 It is possible that our model may also exhibit a lower threshold to handling and multiple testing. In our previous analysis of fear memory in this model, KO mice were tested once, did not undergo an entire behavioral battery and were not handled for medication administration. 24 Our results support the literature suggesting Pten deletion in mice may promote an increased sensitivity to stress. 16,18,50 Additionally, anxiety disorders are a common comorbidity in ASD 51 and occur in $20% of those with ASD and intellectual disabilities. 52 Individuals with ASD can become agitated when their routines are disturbed and this may result in an increase in the severity of perseverative behavior. 51 Future analyses should determine the behavioral consequences of hyperarousal within the context of perseverative behavior in this model.

| CONCLUSION
Here we show that mTOR pathway inhibition with a 2-week course of rapamycin at 10 mg/kg starting at $4 weeks of age increases sociability and corrects stereotypic behavior in NS-Pten KO mice. As this regimen of rapamycin reverses neuronal hypertrophy and impedes the progression of seizures and subsequent mossy fiber sprouting for $3 weeks in this model 19,21 but cannot reverse the aberrant migration of Pten null neurons, 53 these data provide evidence suggesting that hypertrophy and seizures may contribute to the development of autistic-like behaviors in NS-Pten KO mice. However, mTOR also regulates protein and lipid synthesis underlying a variety of neuronal mechanisms, such as dendritic arborization, synaptic plasticity and excitability, that were not examined here. [54][55][56] PTEN mutations also lead to a loss of inhibitory interneurons 57 and increased neuroinflammation, 20,50 both of which have been associated with ASD. Thus, it is also possible that NS-Pten deletion underlies both behavioral deficits, as well as other mechanisms governing neuronal structure and function. Although ASD can be diagnosed as early as 18 months, the average age at diagnosis is $5.4-7.4 years old. 58 Data from other rodent models of hyperactive mTOR induced epilepsy and ASD-like behavior suggest that the earlier treatment can begin, the more likely ASD-like behaviors can be affected. 18 Here we also show that an acute treatment provided during the juvenile period may impact behavior in early adulthood, identifying the juvenile period as a potential developmental window of treatment. Further work, however, is needed to understand the mechanisms underlying how NS-Pten KO contributes to behavioral dysfunction and how this is impacted by rapamycin treatment. Our findings expand our understanding of the ASD-like behavioral phenotype expressed in NS-Pten KO mice, open new avenues for future behavioral research and show potential promise for the treatment of ASD in those PTEN mutations.

ACKNOWLEDGMENTS
The research was funded by NIH NINDS grant R15NS088776 (Joaquin N. Lugo).