In utero Exposure to Valproic Acid throughout Pregnancy Causes Phenotypes of Autism in Offspring Mice

Valproic acid (VPA) is a widely used antiepileptic drug that decreases epileptic activity in the cerebral neocortex mainly by inhibiting sodium/T-type calcium channels and GABA transaminase [1]. VPA also inhibits histone deacetylases and thus epigenetically modifies cell proliferation/differentiation in developing tissues, including the fetal nervous system [2‒6]. Specifically, VPA exposure has been reported to alter the expression patterns of cell cycle-related genes in developing mammalian tissues and to dysregulate the cell proliferation/differentiation characteristics of stem/progenitor cells [7, 8].

A series of recent clinical studies in humans have indicated that VPA exposure in utero increases the risks of central nervous system anomalies [9], autism spectrum disorders [10, 11], and low intelligence quotient scores [12] in offspring. We previously reported that in utero VPA exposure in mice, using a plasma concentration lower than that of the treatment level in humans, increased the production of projection neurons in the superficial neocortical layers [8]. VPA altered the cell cycle kinetics of the neuronal progenitor cells, with an increased amount of acetylated histone proteins in the nuclear fraction [2, 3, 8]. The increased level of acetylation resulted in altered expression levels of G1-phase regulatory proteins, namely, cyclin-dependent kinases and cyclin-dependent kinase inhibitors, and resulted in the abnormally low differentiation probability (the value of Q or the Q fraction) of the neuronal progenitor cells during the early stage of neocortical histogenesis [8]. According to our previous analyses, these alterations in the value of Q have been shown to result in a robust alteration of the number of projection neurons in the neocortices [8, 13, 14, 15, 16, 17, 18].

Taken together, we hypothesized that neocortical dysgenesis caused by VPA exposure in utero would lead to abnormal behaviors in mouse offspring that would resemble the symptoms of autism in human subjects. Previous reports aimed at demonstrating autistic phenotypes in rodents used a single injection of high-dose VPA to pregnant dams; the dose used was likely to have caused VPA intoxication in both the dams and the embryos [19]. Thus, in the present study, we exposed pregnant mice throughout their pregnancies to therapeutic levels of VPA dissolved in drinking water.

Pregnant CD-1 mice (Japan SLC, Inc., Shizuoka, Japan) were maintained under a 12-h light/dark schedule. Since the half-life of VPA is reported to be less than 1 h in mice [20], we administered 0.4% VPA (Sigma-Aldrich, St. Louis, MO) or distilled water as a control in drinking water from embryonic day (E) 1 until birth as previously described [8]. We limited the number of pups reared by a single mother to eight in order to standardize the nutritive conditions of the pups born to these prolific CD-1 dams. After weaning on postnatal day 21, two to three male littermates from each mother were randomly selected for behavioral tests, and different groups of mice were housed in three to four per cage with ad libitum access to food and water. The general health check and neurological screens were conducted as previously described [21]. Behavioral testing was performed using male mice between 9:00 a.m. and 6:00 p.m., and we conducted the experiments with a single batch (control group, n = 19; VPA group, n = 19). To avoid the influence of time of day on behavioral performance, the order of trials for each test was counterbalanced between experimental groups. The battery of behavioral tests was conducted in the following order, starting with the least stressful test: general health check and neurological screens, light/dark transition test, open field test, elevated plus maze test, hot plate test, social interaction test in a novel environment, rotarod test, 3-chamber social approach test, startle response/prepulse inhibition (PPI) test, Porsolt forced swim test, Barnes maze test, T-maze test, tail suspension test. To minimize the carryover effects of the previous test, each test was performed at least 1 day apart. After each trial in the tests, the apparatus was cleaned with hypochlorite solution or 70% ethanol to prevent a bias due to olfactory cue. The age of the male subjects when they were tested is summarized in online supplementary File (for all online suppl. material, see www.karger.com/doi/10.1159/000530452). All behavioral testing procedures were approved by the Animal Care and Use Committee of Fujita Health University.

The hot plate test at 55.0°C (Columbus Instruments, Columbus, USA) was applied to evaluate sensitivity to a painful stimulus. The latency to the first paw response was recorded, with a cutoff time of 15 s [22, 23].

The light/dark transition test was performed as previously described [24]. Briefly, the apparatus consisted of a cage (21 × 41.5 × 25 cm) that was divided into two sections of equal size by a partition with a door (O’HARA & Co. Ltd., Tokyo, Japan). One chamber was brightly illuminated (approximately 390 lux), whereas the other was dark (less than 5 lux). Behavioral indices shown in figures were recorded automatically by ImageLD software.

The open field test was used to evaluate locomotor activity and emotional response [22, 23]. Each male mouse was placed in the corner of an open-field apparatus (40 × 40 × 30 cm; AccuScan Instruments). The center of the floor was illuminated at 100 lx. Total distance traveled, vertical activity (rearing), and time spent in the central area (20 × 20 cm) were recorded over 120 min.

The elevated plus maze test was conducted as previously described [25]. The apparatus consisted of two open arms (25 × 5 cm) and two enclosed arms of the same size with 15-cm-high transparent walls (O’HARA & Co.). Each male mouse was placed in the central square of the maze (5 × 5 cm) facing one of the closed arms, and its behavior was recorded over 10 min. Behavioral indices shown in figures were measured using ImageEP software. Five of the control mice fell from the open arms during the experiments, and we excluded the data for the mice.

A social interaction test was conducted as previously described [22, 23]. Two male mice that were previously housed in different cages were placed together in a box (40 × 40 × 30 cm) and allowed to explore freely for 10 min. Since pairs of mice were used for the test, the number of samples was reduced to half size that of the original (control group, n = 9; VPA group, n = 9). Behavioral indices shown in figures were analyzed by ImageEP software.

Sociability and social novelty preference tests were conducted as previously described [21]. The testing apparatus consisted of a rectangular, three-chambered box (20 × 40 × 46.5 cm) and the dividing walls were made from clear Plexiglas, with small rectangular openings (5 × 3 cm) allowing access into each chamber (O’HARA & Co.). A habituation session was performed in the apparatus for 10 min before the sociability test. In sociability test, an unfamiliar male mouse (stranger 1) that had had no prior contact with the subject mouse was placed in one of the side chambers. The stranger mouse was enclosed in a small, round wire cage that allowed nose contact between the bars but prevented fighting. The subject mouse was first placed in the middle chamber and allowed to explore the entire test box for 10 min. After the first 10 min, each mouse was tested in a second 10-min session to quantify social preference for a new stranger mouse (stranger 2). The stranger 2 mouse was enclosed in an identical small wire cage that had been empty during the first session. The test mouse thus had a choice between the first, already-investigated unfamiliar mouse (stranger 1) and the novel unfamiliar mouse (stranger 2). The amount of time spent around the cage(s) and mean duration per contact(s) were measured and calculated. Data from mice with zero contacts (VPA group, n = 2) were excluded because the mean duration per contact(s) could not be calculated. Data acquisition and analysis were performed using ImageCSI software.

A startle-reflex measurement system (O’HARA & Co.) was used to measure startle response and PPI [22, 23]. Each male mouse was placed in a Plexiglas cylinder and left undisturbed for 10 min. White noise (40 ms) was used as the startle stimulus for all trial types. The background noise level in the chamber was 70 dB. The intensity of the startle stimulus was 110 or 120 dB. The prepulse sound (74 or 78 dB, 20 ms) was presented 100 ms before the startle stimulus. A test session consisted of four combinations of prepulse and startle stimuli (74 and 110, 78 and 110, 74 and 120, and 78 and 120 dB) and two types of startle stimulus only trials. Six blocks of the six trial types were presented in a pseudorandom order such that each trial type was presented once within a block. The average inter-trial interval was 15 s.

The Porsolt forced swim test was performed as previously described [22, 23]. The apparatus consisted of four Plexiglas cylinders (22 cm in height and 11 cm in inner diameter) that were filled with diluted sodium hypochlorite solution at room temperature up to a height of 7.5 cm. Each male mouse was placed in the cylinders, and immobility time was recorded over a 10-min test period. Data acquisition and analysis were performed using ImageTS software.

The Barnes circular maze task was conducted as previously described [22, 23] on a circular surface (1.0 m in diameter, with 12 holes equally spaced around the perimeter; O’HARA & Co.). A black Plexiglas escape box (17 × 13 × 7 cm), which had paper cage bedding on its bottom, was located under one of the holes. The hole above the escape box represented the target, and the location of the target was consistent for a given mouse but randomized across mice. After the last training session, a probe test was conducted without the escape box for 3 min, to confirm that this spatial task was acquired. The time spent around each hole(s) was recorded using ImageBM software.

Tail suspension test was performed as described previously [22, 23]. Each mouse was suspended 30 cm above the floor by the tail. The immobility time was recorded for 10 min. Data acquisition and analysis were performed using ImageTS software.

The forced-alternation task without food deprivation was performed as previously reported [21, 26] with a T-maze apparatus (O’HARA & Co.). A mouse was subjected to 10 consecutive trials in a session (cutoff time, 50 min). On the first run of each trial, a mouse was forced to choose one of the arms of the T-maze. Following the forced-choice run, the free-choice run automatically begins. If the mouse enters the opposite arm that it was forced to choose in the forced-choice run, its response is judged as “correct.” After 4 consecutive sessions, a delay (3, 10, 30, 60 s) was applied after forced-choice run (sessions 5–8). Extending delay time could increase working memory load and task difficulty. Behavioral indices shown in figures were measured using ImageTM software.

The application software used for the behavioral experiments was based on ImageJ program (http://rsb.info.nih.gov/ij/), which was modified for each test by T. Miyakawa. Statistical analysis was conducted using StatView (SAS Institute, Cary, NC) as previously reported [22, 23]. If data were excluded for reasons other than death, it was noted in the Materials and Methods for each experiment. Data were analyzed using a one-way ANOVA, two-way repeated measures ANOVA, or paired T test with offspring as “n = 1.” Furthermore, we re-analyzed our findings including 1) physical characteristics, 2) open field test, 3) hot plate test, 4) three-chamber social approach test, 5) Barnes maze test, and 6) T-maze test (with delayed alternation) with data from offspring from the same dams averaged together and the dam as “n = 1” (shown in online suppl. File). We excluded the social interaction test from this additional analysis because the data were obtained from pairs of mice derived from different mothers and we do not have data for each mouse. Values in graphs are expressed as mean ± SEM.

We did not detect any external congenital anomalies, including macroscopic spina bifida, in the in utero VPA-exposed group of mice. Comparisons between the VPA group and a control group showed no significant difference in body weight (p = 0.830), body temperature (p = 0.549), or grip strength (p = 0.974). Moreover, there was no significant difference between the VPA group and the control group in the results of the rotarod test (p = 0.125) (Fig. 1a). These results are consistent with re-analysis by ANOVA using the dam as “n = 1” (shown in online suppl. File).

The VPA group exhibited a shorter total distance traveled (p = 0.0004, n = 19 for each group), a lower vertical activity (p = 0.0039, n = 19 for each group), and lower stereotypic counts in an open field (p = 0.0057, n = 19 for each group) than the control group (Fig. 1b–d). These results are consistent with re-analysis by ANOVA using the dam as “n = 1” (shown in online suppl. File). In addition, the VPA group exhibited decreased physical activity in a dark environment, compared with the control group (1,500 ± 32.2 cm vs. 1,637 ± 57.2 cm; p = 0.0437, n = 19 for each group) (Fig. 1e). This decreased physical activity was observed during the social interaction test (4,557 ± 340 cm vs. 2,887 ± 223 cm; p = 0.0008, n = 9 for each group) (Fig. 1f). The VPA group also had a longer latency to fall in the wire hang test (9.68 ± 1.56 s vs. 5.68 ± 0.865 s; p = 0.0313, n = 19 for each group) (Fig. 1g). Considering the above, we concluded that VPA exposure in utero decreased locomotor activity in a manner that was not due to abnormal motor function or decreased muscle strength.

To investigate whether VPA exposure in utero causes abnormal social interaction activity, which is a characteristic feature of autistic patients, we analyzed the number and total duration of contacts and the total duration of active contact (moving more than 10 cm in distance to reach another mouse) between in utero VPA-exposed mice and stranger mice using a social interaction test. The in utero VPA-exposed mice made significantly fewer contacts with the stranger mice than the controls (69.0 ± 7.45 vs. 96.7 ± 8.65; p = 0.0276, n = 9 for each group) (Fig. 2a), and the total duration of active contact with the stranger mice was also decreased (21.5 ± 2.75 vs. 36.7 ± 4.48; p = 0.0103, n = 9 for each group) (Fig. 2b). The total duration of contact was unaltered in the in utero VPA-exposed group (p = 0.386) (Fig. 2c). The mean duration per contact increased, though the difference was not statistically significant, in the VPA group (p = 0.0615) (Fig. 2d). We also conducted a three-chamber social approach test. In the social novelty preference test, there is no significant difference in time spent around the cage between groups (p = 0.196) (Fig. 2e). However, mean duration per contact tended to be significantly longer in in utero VPA-exposed mice compared to that of the control (treatment effect, p = 0.0836; treatment × location interaction, p = 0.0437) (shown in online suppl. File). The mean duration per contact with the stranger mice was significantly longer in the in utero VPA-exposed mice, compared with the control mice (p = 0.0434) (Fig. 2f). In contrast, the mean duration per contact with the familiar mice was unaltered in the in utero VPA-exposed mice compared with that for the control group (p = 0.791) (Fig. 2f). These results are consistent with re-analysis by ANOVA using the dam as “n = 1” (shown in online suppl. File). Taken together, these results indicated that VPA exposure in utero induced abnormal social interaction activity.

We investigated working memory, which is generally considered to be affected in autistic subjects [11], using automated T-maze test [26]. Whereas the correct responses, latency, and distance during the first 4 sessions were unaltered by in utero VPA exposure (Fig. 3a, b, c), the in utero VPA-exposed mice made significantly fewer correct responses in the delayed alternation experiment compared with the control group (p = 0.0081) (Fig. 3d). This result is consistent with re-analysis by ANOVA using the dam as “n = 1” (shown in online suppl. File). Taken together, these results suggested that VPA exposure in utero affected working memory only for more difficult tasks (Fig. 3d).

Next, we used a Barnes maze test to investigate the effect of in utero VPA exposure on spatial memory. There was no significant effect of VPA exposure on the number of search errors made, suggesting normal spatial memory in the VPA-exposed group (data not shown). When probe trials were conducted 1 day and 1 month after the completion of the 16 training sessions, there were no significant differences between the groups in the time spent around each hole (1 day, p = 0.556; 1 month, p = 0.971) (Fig. 3e). In the 1-day probe trial, both the VPA exposure group (n = 19) and the control group (n = 18) spent more time around the target hole than its adjacent holes (VPA group: target hole, 25.7 ± 7.45 s vs. adjacent holes, 14.0 ± 0.826 s [p < 0.0001]; control group: target hole, 22.7 ± 2.92 s vs. adjacent holes, 13.1 ± 1.75 s [p = 0.0068]) (Fig. 3f). One month after the training sessions, however, the VPA group spent significantly more time around the target hole than its adjacent holes (VPA group: target hole, 20.9 ± 2.93 s vs. adjacent holes, 10.7 ± 1.11 s [p = 0.0006]; control group: target hole, 20.5 ± 5.23 s vs. adjacent holes, 12.4 ± 3.95 s [p = 0.2433]) (Fig. 3f). These results are consistent with re-analysis by ANOVA using the dam as “n = 1” (shown in online suppl. File). Therefore, we concluded that in utero VPA exposure improved the accuracy of long-term spatial memory.

A decreased heat sensation threshold is known to indicate sensory hypersensitivity, as in autistic subjects [27]. Thus, we evaluated the effect of VPA exposure on heat sensitivity by measuring the latency of response to sensory stimuli applied to the feet using a hot plate set at 55°C. In fact, the results showed that the VPA group had a 27.8% shorter latency than that of the control group (2.36 ± 0.267 s vs. 3.27 ± 0.314 s; p = 0.0347, n = 19 for each group) (Fig. 3g). Although the value of p is not significant in ANOVA using the dam as “n = 1” (p = 0.0532, shown in online suppl. File), we speculate that this result might be influenced by confounding factors that include the development and housing environments of the mice during the investigation. Taken together, we concluded that in utero exposure to VPA induced hypersensitivity to pain/heat sensations.

Anxiety disorders, schizophrenia, and depressive symptoms are common comorbidities in autistic young human adults [27]. To investigate whether in utero VPA exposure directly causes those disorders, we first conducted an elevated plus maze test and a light/dark transition test to measure the level of anxiety-like behavior. No significant difference between the VPA group and the control group was seen in the number of entries into each arm (p = 0.096). Neither a significant difference was seen in the number of entries into the open arms between the VPA group and the control group in the number of entries into the open arms (p = 0.385) nor in the time spent in the open arms (p = 0.110). Consistent observations were made for the light/dark transition test; no significant difference was seen in the duration of the stay in the light between the VPA group and the control group (p = 0.887), the number of transitions (p = 0.174), and the latency to light (p = 0.459). Next, we tested the PPI of acoustic startle, which is impaired in patients with schizophrenia and is used as a measure of the schizophrenia-relevant phenotype in rodents; no significant difference was found between the VPA group and the control group (110 dB, p = 0.0505; 120 dB, p = 0.574) (Fig. 4a).

We conducted the Porsolt forced swim test and tail suspension test to assess learned helplessness, which is a feature of depression-like behavior in rodents. The duration of immobility was significantly longer in the VPA group than in the control group (day 1, p = 0.0039; day 2, p < 0.0001) (Fig. 4b), whereas the tail suspension test results did not show any difference between the groups (p = 0.736) (Fig. 4c). We speculated that the decreased locomotor activity in the VPA-exposed group, described in a previous section, might have led to the abnormal increase in immobility in the Porsolt forced swim test. Given these results, it is difficult to determine whether VPA exposure in utero is associated with depression-like behavior. Based on the above findings taken together, we conclude that in utero VPA exposure does not directly cause symptoms of anxiety, schizophrenia, and depression.

In the present study, we described the behavioral characteristics of in utero VPA-exposed mice. These characteristics include signs/symptoms commonly associated with human subjects with autism and/or intellectual disability [28]. The underlying pathophysiology of autism is considered to be multifactorial and heterogeneous. An analysis of a gene database containing information from autistic patients has indicated that the number of causative genes might be as large as 1,000; however, none of these causative genes accounts for more than 1% of the entire population of individuals with autism [29, 30].

Brain autopsies of autistic children have demonstrated increased numbers of synapses per neuron in the cerebral cortex compared with controls [31]. It has been hypothesized that, in autism, neuronal circuits involved in language and social behavior may be disrupted by an impaired excitation/inhibition balance in the cerebral cortex [32]. Taken together, these results suggest that an impaired excitation/inhibition balance of the cerebral circuits, presumably caused by a disparity between excitatory and inhibitory synapses or neurons, might be a common underlying pathology in autism.

Whereas we detected significantly fewer contacts with stranger mice in the in utero VPA-exposed mice, the mean duration of contact with the stranger mice was increased in the VPA-exposed group, compared with the control group, in two different social interaction tests (Fig. 2a, f). Autistic children often exhibit a fixated interest in the same object and/or human, and the target object is not always a static material; sometimes it is an animal, part of a plant (leaf, branch, etc.), or moving object (wheel, ripple, etc.) [27]. Thus, the increased duration of each contact with the stranger mice in the VPA-exposed mice might suggest the autistic phenotype of fixated interest that can be observed in children with autism.

We also detected decreased locomotor activity in the in utero VPA-exposed mice in various tests, including the open field and light/dark transition tests. This poor coordination of motor movements of the upper and lower extremities suggests a phenotype that is occasionally observed in autistic patients; such children tend to have poor coordination skills, especially when they are running or using their hands and fingers simultaneously [27]. In addition, this decreased locomotor activity may have affected the number of contacts with stranger mice in the social interaction test and the percentage of immobility in the forced swim test in the in utero VPA-exposed mice. Decreased activity in the in utero VPA-exposed mice should be considered as a confounding factor in behavioral tests.

Impaired working memory has been reported in autistic patients, especially in those with decreased intelligence quotient scores [12]. We detected an impairment in working memory in the delayed alternation experiment in in utero VPA-exposed mice using the automated T-maze test (Fig. 3d) [26]. However, the accuracy of long-term spatial memory as assessed using the Barnes maze test was improved in the VPA-exposed mice compared with the controls (Fig. 3f). Although additional histological or electrophysiological data are needed to draw any conclusions, we speculated that the improved long-term spatial memory among VPA-exposed mice may be a consequence of improved hippocampal function. In addition, this phenomenon may be analogous to the characteristics of some, if not all, autistic patients with improved long-term visual memory retention, described as “photographic memory” or a savant ability [33]. Since impaired “short-term” memory negatively affects the intelligence quotient score, the decreased IQ commonly observed in human autistic subjects might be, at least in part, due to the abnormal function of working memory, as observed in the present study [12].

In clinical settings, autistic children often suffer from hypersensitivity to specific sensory input [34]. The biological mechanism responsible for such hypersensitivity is under debate; some reports have indicated that the gating threshold of autistic patients for sensory stimuli is so low that they cannot obtain habituation to repetitive and/or high-intensity sensory input [35]. In the present study, the mice exposed to VPA in utero had a shorter latency to heat stimulation, despite an abnormally decreased physical activity level, suggesting that VPA exposure resulted in hypersensitivity resembling the symptoms of autistic children. This hypersensitivity may be due to the increased number of projection neurons in the neocortex [8, 36]. This possibility is further discussed in the following section.

VPA exposure in utero has been reported to increase the risk of autism spectrum disorder in humans [10, 11]. Of note, the administration protocol used in this report resulted in a dosage that was equivalent to the usual treatment level in clinical settings (i.e., corresponding to a serum concentration of 120–150 μg/mL, which is the level that increases the seizure threshold in mice by 50%) [8]. This result contrasted with that of a previous report, in which autistic features were reported in rodents exposed to high doses of VPA over a short period [19].

An abnormally augmented neural network excitability has been implicated in the pathogenesis of autism in both humans and mice [37]. Importantly, in utero VPA exposure increases the number of excitatory projection neurons in the superficial layers of the neocortex without altering the number of interneurons; this observation was reported in our previous study, in which the same VPA exposure protocol was adopted [8].

The abnormal hypersensitivity in mice exposed to VPA in utero may be due to the increased number of excitatory projection neurons in the superficial layers, i.e., layers II–IV, of the neocortex, with no change in the number of interneurons. Previous observations using a whole brain connectome analysis indicated that neurons in the superficial layers are connected with excitatory neurons of the superficial layers of other neocortical areas [38]. This suggests that the overproduction of projection neurons within the superficial layer may result in “self-stimulating circuitry,” thereby impairing higher brain function. In accordance with this hypothesis, the cerebral cortex in humans diagnosed with autism reportedly shows an increased density of mini-column structures compared with that in controls [39]. In addition, MRI findings for the white matter in subjects with autism show that the connectivity of adjacent areas within the cortex is increased, whereas the connectivity of the cross-sectional area for the corpus callosum is decreased [40].

Epilepsy is a major comorbidity in autistic children and adolescents; the incidence of epilepsy is more than 20 times higher than that in normal subjects [41]. Major seizure types among autistic subjects include generalized tonic-clonic seizures, absence seizures, or complex partial seizures [41]. Layer IV in the somatosensory cortex is the main target of the thalamocortical projection; in this circuity, feedforward inhibition by interneurons reportedly plays a major role in fine-tuning the excitatory/inhibitory balance of cortical function [42, 43]. Such susceptibility to hyperexcitation may be a predisposing factor to epilepsy among autistic patients.

To support the abovementioned hypotheses, a previous study by others reported that increased projection neurons in the neocortical superficial layers indeed induced an autistic phenotype in mice [36]. An increased number of neurons in the postmortem brains of autistic children has also been reported in a human study [44]. Most relevant to this issue is a case report of severe autistic phenotype in an individual with an enlarged head circumference; this case showed that the decreased expression of p27^Kip1 protein resulted in a severe autistic phenotype with enlarged head circumference [45]. Since the loss of function of p27^Kip1 in mice leads to not only an enlarged body size [46] but also an increased number of projection neurons in the superficial layers [16], an increased number of excitatory neurons in the superficial layers of the neocortex might represent a fundamental mechanism of autistic features in both humans and rodents.

In summary, we describe abnormal behavioral characteristics in in utero VPA-exposed mice that were reminiscent of the signs/symptoms of human subjects with autism spectrum disorder with a decreased intelligence quotient. Our VPA administration protocol closely reflects the clinical setting in humans. Even at this relatively low dose, continuous VPA exposure in utero was capable of inducing abnormal neocortical histogenesis, specifically the overproduction of excitatory projection neurons in the neocortical superficial layers. We conclude that such neocortical dysgenesis is likely to play a substantial role in the development of autistic phenotypes.

All behavioral testing procedures were approved by the Animal Care and Use Committee of Fujita Health University, approval number I0741.

The authors have no conflicts of interest to declare.

This work was supported by Grant-in-Aid for Scientific Research (B) (26293248, 17H04232, and 20H03649 for T.T.) and (C) (25461560 and 19K08306 for T. Mitsuhashi) of Japan Society for the Promotion of Science (JSPS), the Japan Epilepsy Research Foundation, Japan Foundation for Pediatric Research Grant (for T. Mitsuhashi), MEXT Promotion of Distinctive Joint Research Center Program (JPMXP0618217663 and JPMXP0621467949 for T. Miyakawa), and Grant-in-Aid for Scientific Research on Innovative Areas (221S0003 and 16H06276 for T. Miyakawa).

T.Mitsuhashi and S.H. designed research; S.H. and T. Miyakawa performed experiments; and T. Mitsuhashi, S.H., K.F., S.S., T. Miyakawa, and T.T. analyzed data and wrote the paper.

Takayuki Mitsuhashi and Satoko Hattori contributed equally to this work.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.