Mutation of the histone demethylase Gasc1 causes ASD-like symptoms in mice

Abstract

Background

Genomic analyses of psychiatric disorders, including autism spectrum disorder (ASD), have revealed many susceptibility genes, suggesting that such disorders may be caused by multiple factors. In this sense, it has long been a question whether there is an abnormal genetic status that comprehensively explains the pathogenesis of neuropsychiatric disorders or a"promising upstream treatment target"that normalizes symptoms.

Methods

To address this question, we provide important clues with respect to GASC1 (JMJD2 C/KDM4 C), which is a histone demethylase that prominently targets trimethylated histone H3 at lysine 9 (H3 K9 me3). Gasc1 hypomorphic mutant mice were analyzed using molecular biological, biochemical, behavioral battery tests, histological, and electrophysiological techniques.

Results

Mice homozygous for a hypomorphic mutation in Gasc1 exhibited abnormal behaviors, including hyperactivity, stereotyped behaviors, and impaired learning and memory, which are reminiscent of those of human psychiatric disorders. Electrophysiological studies of hippocampal slices revealed decreased paired-pulse facilitation and enhanced long-term potentiation, suggesting synaptic dysfunction in the mutants. Increased dendritic spine density in CA1 neurons was also detected in the mutants. Intriguingly, genetic linkage studies of human ASD have mapped a susceptibility locus on chromosome 9p24.1, which contains 78 genes, including the GASC1 gene.

Conclusion

Taken together, our data suggest that histone demethylation plays a pivotal role in normal brain development and higher-order brain functions in both mice and humans.

Supplementary Information

The online version contains supplementary material available at 10.1186/s41232-025-00374-5.

Introduction

Normal structural and functional development of the brain largely involves time- and cell-specific gene expression, which is regulated by genetic and epigenetic mechanisms. There are two types of epigenetic regulation, DNA methylation and histone modification, both of which control chromatin folding and the accessibility of transcription factors. Tri-methylation of histone H3 lysine- 9 (H3 K9 me3) is an epigenetic memory of silenced chromatin states and is stably transmitted throughout many cell generations. Thus, the demethylation of histone H3 K9 me3 may trigger the expression of a wide range of genes. GASC1 (gene amplified in squamous cell carcinoma 1, also known as JMJD2 C and KDM4 C) encodes histone H3 K9 me3, H3 K9 me2 and H3 K36 me3 demethylase. GASC1 is a member of the JMJD2 class of JmjC domain-containing Jumonji family proteins. Other JMJD2 proteins, including JMJD2 A (KDM4 A), JMJD2B (KDM4B), and JMJD2D (KDM4D), commonly exhibit histone H3 K9 me3 demethylase activity and show overlapping expression profiles among tissues and cells, suggesting that each JMJD2 protein has specific target genes or redundant functions [1].

The GASC1 gene was originally identified as a gene frequently amplified in human esophageal squamous cell carcinomas [2], and the oncogenic roles of GASC1 in cancer development and progression have been well studied (see discussion in detail). In addition, Oct4 was reported to induce Gasc1 expression, and GASC1 promoted Nanog expression in mouse embryonic stem (ES) cells [3]. The expression of other pluripotency genes has also been reported to be downregulated in Gasc1-depleted embryos [4] and ES cells [1, 5]. The function of GASC1 was further implicated in lineage-specific differentiation processes, as Gasc1 knockdown affects the differentiation of neural stem cells and other cells [6–9].

Thus, accumulating evidence indicates the importance of the histone demethylase GASC1 in ES cells and normal development, and its expression has been detected in various tissues, including the brain and eye, at high levels (tissue expression of KDM4 C—summary—The Human Protein Atlas, https://www.proteinatlas.org/ENSG00000107077-KDM4C/tissue). However, its physiological and pathological consequences remain elusive. Therefore, this study investigated its role at the whole-body level. For this purpose, we analyzed hypomorphic mutant mice for Gasc1, since the embryonic lethality of conventional Gasc1 knockout mice was predicted from its function in early embryos [4]. We first demonstrated that a hypomorphic mutation in the mouse Gasc1 gene increases histone H3 K9 me3 and H3 K9 me2 in the cerebrum and H3 K9 me3 in the brainstem, indicating the histone demethylase activity of GASC1 under physiological conditions in vivo. The Gasc1 hypomorphic mutants presented no gross phenotypes; however, they exhibited abnormal behaviors (such as hyperactivity and perseveration), synaptic dysfunction and elevated spine density in the hippocampal CA1 region. These phenotypes are reminiscent of those reported in human psychiatric disorders, including autism spectrum disorders (ASD) [10, 11]. Intriguingly, recent genome-wide association studies (GWASs) of human ASD identified copy number variations (CNVs) [12–14], single nucleotide polymorphisms (SNPs) [15, 16] and single nucleotide variations (SNVs) [16] at the 9p24.1 locus, which included the human GASC1 gene. Especially, Kato et al. [14] presented a case of a human GASC1 (KDM4 C) CNV patient who has 18 rare missense variants including the JmjC domain and expected loss of function. They further reported decreased GASC1 protein expression and altered histone methylation patterns by immunoblotting analysis using lymphoblastoid cell lines established from the patients. Their results suggest that GASC1 protein function and expression is decreased, and histone methylation patterns are also altered in the brains of human GASC1 CNV patients. Our current study provides direct evidence that a single gene mutation in Gasc1 gene causes ASD-like behavior in a mouse model. Taken together, our data demonstrate a novel role of the histone demethylase GASC1 in higher-order brain functions, and that abnormal Gasc1 expression is a cause of psychiatric disorder-like phenotypes, including ASD.

Interestingly, a comparison of gene expression levels between Gasc1 hypomorphic mutant and wild-type mouse brains revealed that Gasc1 hypomorphic mutant brains presented markedly increased expression levels of genes related to inflammation. In this context, we previously reported a marked increase in GFAP-positive reactive astrocytes in Gasc1 hypomorphic mutant brains [17], which may be related to such changes in inflammatory gene expression.

Results

Gasc1 gene activation in the mouse forebrain: Expression patterns and brain localization

To identify Gasc1-expressing cells in the developing mouse brain, we employed a transgenic mouse strain established by insertion of the gene trap vector pU- 17 into intron 1 of the mouse Gasc1 gene. The pU- 17 vector includes a splice acceptor signal, β-geo, and poly A signal from the mouse phosphoglycerate kinase- 1 (Pgk1) gene; therefore, the activation of the Gasc1 gene promoter can be monitored by the expression and enzymatic activity of β-galactosidase (Fig. 1A). We prepared cryosections of developing mouse brains heterozygous for the mutant Gasc1 gene and stained them with X-gal to visualize Gasc1-expressing cells. X-gal-positive cells were detected in the gray matter of the adult brain, particularly in forebrain regions, such as the cerebral cortex, hippocampus, striatum, olfactory bulb, and hypothalamus. Only a few X-gal-positive cells were detected in the white matter, suggesting that the Gasc1 gene is mainly activated in neurons but is less activated in oligodendrocytes (Fig. 1B-D). Double immunostaining of the brain with anti- β-galactosidase and anti-NeuN antibodies supported this observation, and as shown in Fig. 1E-K, approximately 87.5% of Gasc1-positive cells were NeuN double-positive neurons. Similarly, in the embryonic and neonatal brains, X-gal-positive cells were detected in neuron-rich regions (Suppl. Figure 1 and 2). Only a few X-gal-positive cells were detected in the ventricular and subventricular zones, suggesting that neural stem and precursor cells rarely express the Gasc1 gene. To confirm Gasc1 gene expression during mouse brain development, we performed quantitative RT‒PCR. A significant level of Gasc1 gene expression was detected in embryonic and adult brains (Suppl. Figure 3).

Fig. 1.

Neuronal expression of β-geo inserted in intron 1 of the mouse Gasc1 gene. A, The gene trap cassette containing the β-geo gene with a splice-acceptor signal and a polyA site was inserted into intron 1 of the mouse Gasc1 gene. This insertion disrupted GASC1 protein production, in theory, and enabled us to monitor Gasc1 promoter activity. B, β-geo expression detected by X-gal staining was widely observed in the brains of adult mice heterozygous for the mutant Gasc1 gene. C and D, Fivefold magnified images of the square regions indicated in B demonstrate that β-geo was intensely expressed in the cortical plate, the dentate gyrus and Ammon’s horn in the hippocampus. E-K, Double immunostaining with anti-β-galactosidase (E and I) and anti-Neu-N (F and G) antibodies. G and K are merged images. H, Nuclear staining with Hoechst 33,258. I-K, Magnified images indicated by white boxes in E–G. Scale bar in B = 200 μm. Scale bar for E–H = 100 μm

Low-level expression of Gasc1 detected in the homozygous mutant brain: Impact on mRNA variants and protein levels

To investigate the function of GASC1 in vivo, mice homozygous for the trap vector inserted in the Gasc1 gene were generated. Because the trap vector was inserted into the first intron but did not disrupt any exon sequences, leaky expression of the Gasc1 gene was detected in the homozygotes. In addition, according to the gene information obtained from Ensembl (http://asia.ensembl.org/Mus_musculus/Gene/Summary?g=ENSMUSG00000028397;r=4:74242497-744058600), the mouse Gasc1 gene encodes several alternative splice variants transcribed from multiple promoters. Thus, we carefully examined the amount of Gasc1 mRNA in the brains of wild-type and homozygous young adult individuals (7–10 weeks old) via several RT‒PCR primer sets. The RT‒PCR primers designed to amplify the regions from exons 1 to 4 of Gasc1 mRNA detected clear bands in the cerebra, brain stems and cerebelli of wild-type mice. However, no band was detected in the mutant brain samples, suggesting that the full-length canonical Gasc1 mRNA is not expressed in the homozygote (Fig. 2A). Compared with those of the wild-type control, approximately half the amount of the mRNAs containing exons 5 and 6 were detected in the mutant brains (Fig. 2B). These genes might correspond to mRNAs encoding GASC1 variants that are listed in the Ensembl database and transcribed from exons 0 or 2. More importantly, Gasc1 mRNA, which was detected by primer sets at exons 11 and 13, encoded functional GASC1 proteins containing JmjN, JmjC, two PHD and two Tudor domains referred to in the Ensembl database. Compared with that in the wild-type control, the expression of this mRNA in the mutant was reduced by 80% (Fig. 2C). On the basis of these results, we estimated that the expression of functional GASC1, a histone demethylase, in the homozygous brain was reduced to approximately 20% of that in the WT brain. The reduction in GASC1 at the protein level could not be determined because of the lack of proper anti-GASC1 antibody, even though we tried to generate it via the use of 4 independent peptide sequences. While the reduction in GASC1 in the mutants did not significantly affect the expression of other KDM4 family members in the cerebra (Fig. 2D), the histone H3 K9 me3 level in the cerebra and brainstem and the H3 K9 me2 level in the cerebra (Fig. 2E, Suppl. Figure 4) were increased.

Fig. 2.

Reduced expression of Gasc1 in Homozygote brains leads to an increase in highly methylated histone H3 K9. A, Gasc1 mRNA expression in the cerebrum, brain stem, and cerebellum of wild-type and Gasc1-homozygous mice (7–10 weeks old) was detected via PCR with primers specific for exons 1 and 4 (n = 3 each). B, Relative Gasc1 expression in the cerebrum, brain stem, and cerebellum in wild-type and Gasc1 homozygous mice detected by real-time PCR; the PCR primers were designed for exons 5 and 6. The p-values for Cbr, BS, and Cbl were 0.0153, 0.0267, and 0.0013, respectively (n = 3 each). C, Relative Gasc1 expression in the cerebrum, brain stem, and cerebellum in wild-type and Gasc1 homozygous mice detected by real-time PCR; the PCR primers used were exons 11 and 13. The p-values for Cbr, BS, and Cbl were 0.0051, 0.00022, and 0.0622, respectively (n = 3 each). D, No changes in the expression levels of the Gasc1 families Kdm4a, Kdm4b and Kdm4d in the cerebra of wild-type and Gasc1 homozygous mice were detected by real-time PCR. The p-values for Kdm4a, Kdm4b, and Kdm4 d were 0.692, 0.120, and 0.964, respectively (n = 3 each). E, Increased levels of trimethylated histone H3 K9 in homozygous cerebrum and brain stem regions compared with those in the wild type. Similarly, there was an increase in the level of the di-methylated histone H3 K9 in homozygous cerebra compared with that in wild-type cerebra. Each bar shows the statistically calculated relative values of the homozygous mutants when the value of the wild type is set to 1. The mean ± standard deviation for Cbr were 1.08294 ± 0.13427 for me1, 1.36149 ± 0.24992 for me2, and 1.83289 ± 0.61606 for me3. For BS, they were 0.86441 ± 0.19269 for me1, 0.79991 ± 0.23256 for me2, and 1.4227 ± 0.08659 for me3. For Cbl, they were 0.92284 ± 0.06456 for me1, 0.79527 ± 0.15923 for me2, and 1.01143 ± 0.1191 for me3 (n = 3 each)

No gross abnormalities in the brain structure of Gasc1 homozygous mice

The Gasc1 homozygous pups were born alive, but 70% of them died within 2 days after birth (P2) for unknown reasons (Suppl. Figure 5 A, B). The remaining survivors presented a reduced body weight, especially females, during early postweaning (Suppl. Figure 5 C, D). However, no gross abnormalities were detected in the brain structure of Gasc1 homozygotes at postnatal day 0 (P0), as determined by immunostaining with anti-NeuN, anti-Ki67, and anti-GFAP antibodies (Suppl. Figure 6). Layer formation of the cortical plate was also maintained in the mutants, as examined by anti-Tbr1, anti-Ctip2, and anti-Cux1 antibody staining of neonatal brains (Suppl. Figure 7). Adult brain sections from homozygotes and heterozygotes were stained with X-gal to visualize the cells in which the Gasc1 gene promoter was activated. No remarkable abnormalities in the distribution of X-gal-positive cells were detected in the adult brains of homozygotes compared with those of heterozygotes (Suppl. Figure 8). The number of NeuN-positive cells in the homozygous hippocampi and cerebral cortices was comparable to that in the wild-type mice. Thus, no gross abnormalities were detected in the developing or adult brains of Gasc1 homozygous mice.

Behavioral test battery: Hyperactivity and cognitive deficits in Gasc1 homozygotes

The Gasc1 homozygotes were fertile, but the mothers failed to nurture their offspring. Young adult homozygotes, though only occasionally, presented with epileptic seizures after 3 weeks of age. Since Gasc1 homozygotes presented no gross abnormalities in brain structure, reduced Gasc1 expression was suspected to contribute to executive function deficits. Thus, we subjected the homozygous males to a comprehensive battery of behavioral tests (Suppl. Table 1). Twenty homozygous and age-matched wild-type males at or older than 11 weeks of age were used for the tests. While the homozygotes were apparently normal, they presented many abnormal behaviors compared with the wild-type controls, as described below.

There was no significant difference in general health or neurological performance, including weight, temperature, appearance, whiskers, coat, righting reflex, whisker twitch, ear twitch, reaching and key jangling, between the homozygote and wild-type control groups at the time of the battery tests. Neuromuscular strength was examined via wire-hanging tests, which revealed that the strength of homozygotes significantly decreased more rapidly than that of the wild-type controls did (Suppl. Figure 9). However, in contrast, there was no difference in forelimb grip strength between the two groups (Suppl. Figure 10). The plate test was used to evaluate sensitivity to painful stimuli. The mice were placed on a hot plate set at 55 °C, and latency to the first response, including a foot shake or a paw lick, was recorded. Compared with the wild-type controls, the homozygotes were significantly less sensitive to high temperatures (Suppl. Figure 11).

Hyperactivity and repetitive behavior: Phenotypic characteristics of Gasc1 homozygotes

A mouse was placed in a novel open-field environment with photobeam scanning arrays to automatically monitor locomotion. The total distance traveled summed every 5 min markedly increased in homozygotes during the first 0–30 min and 30–60 min, indicating that the homozygotes were hyperactive (p < 0.0001, p = 0.0033, respectively; Fig. 3A). Stereotyped repetitive movements every 5 min markedly increased in homozygotes during the first 0–30 min and 30–60 min (p = 0.0001, p = 0.0363, respectively; Fig. 3B). Notably, stereotyped behavior is a cardinal symptom of autism spectrum disorders (ASDs). There was no significant difference in vertical activity. We detected a significant increase in the time spent in the center of the field (Suppl. Figure 12); however, a part of this difference could be explained by hyperactivity of the mutants. In the light/dark transition test, homozygotes traveled significantly longer distances than did the wild-type controls did, confirming their hyperactivity (Suppl. Figure 13). There was no difference in the time spent in the light room, transitions between the light and dark rooms, or latency to first enter the light room. The elevated plus maze test was conducted to evaluate anxiety-like behavior (Suppl. Figure 14). The total distance traveled and number of entries into any arm were significantly greater in the homozygote group than in the control group, reflecting hyperactivity. Compared with that of the wild-type control, the number of entries into the open arms of the homozygotes was significantly greater. The time spent in the center significantly increased, whereas the time spent in the closed arms significantly decreased in the homozygote group. The hyperactivities of the homozygous mutant mice were further detected in the forced swimming and tail suspension tasks (Suppl. Figure 15 and 16). Taken together, these data suggested that homozygotes were hyperactive and that their anxiety-like behavior decreased.

Fig. 3.

Significant increase in locomotor activity in Gasc1 homozygotes. A, Significant increase in locomotor activity, as indicated by the total distance traveled, and B, significant increase in stereotypic behaviors in a novel open field environment. A mouse was placed in the open field apparatus with photobeam scanning arrays to automatically monitor locomotion (40 × 40x30 cm; AccuScan Instruments, Columbus, Ohio). The total distance traveled and stereotypic counts (number of repetitive interruptions of the beams) were recorded for 120 min. C, Hyperactivity of the Gasc1 homozygote was further detected in the home cage test. Two mice of the same phenotype (e.g., two wild-type mice or two homozygous mice) were placed in a home cage with 12-h light and 12-h dark cycles. Locomotor activity was monitored for 7 days, and the average values are shown in the graph. The homozygous mice were significantly hyperactive in this study

Deficits in learning and memory: Impaired cognitive function in Gasc1 homozygotes

Motor learning and motor coordination were tested with a rotarod (UGO Basile Accelerating Rotarod). The rotation speed was increased from 4 to 40 rpm over a 300-s period, and the latency to fall from the rotarod was recorded (Fig. 4A). Wild-type mice and homozygotes showed similar abilities to hold on the rotarod at the first trial; however, wild-type mice had longer durations than homozygotes did (genotype effect p = 0.0035, genotype × trial interaction p < 0.0001), suggesting that homozygotes impaired motor learning/coordination abilities.

Fig. 4.

Abnormal behaviors of Gasc1 homozygotes. A, Gasc1 homozygotes had impaired motor learning/coordination abilities. The mice were placed on rotating drums. The rotation speed was increased from 4 to 40 rpm for 5 min. The latency to fall from the rotarod was recorded. B, Spatial learning and reference memory abilities were examined via the Barnes maze. The Barnes task was conducted on a white circular table with 12 holes around the perimeter. A black Plexiglas escape box was placed under the hole at 0 degrees as a target. The trials were conducted for 6 consecutive days. On day 7, probe tests were carried out without the escape box, and activities were recorded via video tracking software. C, Methylphenidate decreased Gasc1 homozygosity. Mice housed in home cages were given an intraperitoneal injection of MPH or saline at a ratio of approximately 17:45 (between 17:20 and 18:05 due to the technical constraints of sequentially injecting many mice), approximately 1 h before the beginning of the dark cycle (the end of the light cycle), and their activities were measured from 23:45 to 1:45, when the mutant showed the highest activity. MPH administration decreased the activity of homozygotes to the control level but had no effect on the activity of wild-type mice

To examine spatial learning and reference memory abilities, a Barnes spatial navigation task was carried out on a white circular table with 12 holes around the perimeter. A black escape box was placed under the hole at 0° to be a target. The mice were allowed to learn the target position hinted at by some distal visual room cues. The latency to reach the target hole was significantly greater in the homozygote group (p = 0.0370; Suppl. Figure 17 A). Compared with the wild-type controls, the homozygotes made more errors to stop over the incorrect holes and traveled longer distances (p = 0.0002 and p < 0.0001, respectively; Suppl. Figure 17B, C). A probe test was conducted after the escape box was removed from the table. Compared with homozygotes, wild-type mice remembered the former target well and spent more time around the hole (p = 0.0057; Fig. 4B). The total number of visits to holes was equivalent between homozygotes and the wild-type control (Suppl. Figure 17D). The total distance traveled by homozygotes was greater than that traveled by the wild-type control, as inferred from hyperactivity (Suppl. Figure 17E).

Working and reference memory were assessed by forced alternation and left‒right discrimination tasks via a T-maze. In the forced alternation task, a trial consisted of a forced-choice run followed by a free-choice run. In the free-choice run, the mouse was allowed to choose either the right or left area. When the mouse chose the correct area that was not visited in the forced choice run and received a reward pellet, it was considered a correct response (Suppl. Figure 18). One session consisted of 10 trials. Compared with those in session 1, the accuracy rate in session 8 was significantly greater in wild-type control mice (p = 0.0104; Suppl. Figure 18D). In contrast, homozygotes showed no difference in the accuracy rate between sessions 1 and 8. Thus, homozygotes presented significantly impaired working memory. In the left‒right discrimination tasks, until session 8, the correct arm (on which the pellet was placed) was always fixed either on the left or right side, and the mouse was free to choose the area either on the left or right side (Suppl. Figure 19 A and B). After the 8 th session, the correct arm was changed to the opposing arm to access reversal learning, memory erasure/rewrite, and perseveration. The homozygotes performed better in the simple task until session 8, with significant differences in the third, fourth, and fifth sessions; however, they delayed finding position changes in sessions 9, 10 and 11 immediately after the change (Suppl. Figure 19 C). These data suggested that Gasc1 homozygotes presented significantly impaired reversal learning and memory erasure/rewrite and a perseverative tendency.

Contextual and cued fear conditioning: Reduced fear responses in Gasc1 homozygotes

Fear memory was measured by contextual and cued fear conditioning tests (Suppl. Figure 20). For conditioning, each mouse was placed in a test chamber (Suppl. Figure 20 A) and exposed to 3 sets of 55 dB white noise for 30 s (conditioned stimulus), followed by a 0.3 mA foot shock (unconditioned stimulus) for 2 s and a 90 s interval. When the mouse stopped moving for more than 2 s, it was judged as “freezing”. Throughout the conditioning phase, the freezing behavior of the homozygotes also increased, but the freezing behavior of the homozygotes always decreased compared with that of the controls (Suppl. Figure 20D). We compared the distance traveled before and after electrical stimulation in the mutants and controls. There was no significant difference between the first, second, and third stimulations (Suppl. Figure 20E). In the contextual test, the conditioned mouse was placed in the same chamber, and its behavior was recorded for 5 min. The homozygotes showed fewer freezing reactions (Suppl. Figure 20G. In the cued test, the conditioned mouse was placed in a triangular box (a different shape of box, Suppl. Figure 20B), and its behavior was recorded. After 3 min of habituation, it was exposed to 55 dB noise for 3 min without foot shock. Compared with the wild-type controls, the homozygotes presented less freezing behavior in the no-sound, no-stimulation phase, as did the conditioning phase. The homozygotes remembered the conditioned stimulus, as they showed increased freezing but were less abundant than the wild-type controls were (Suppl. Figure 20I).

Prepulse inhibition test: Enhanced startle response in Gasc1 homozygotes

The amplitude of the acoustic startle reflex was measured. Gasc1 homozygotes presented a significantly increased startle response to an acoustic stimulus at 110 dB. Sensorimotor gating was measured by prepulse inhibition of the acoustic startle response (PPI). A mouse placed in the Plexiglas cylinder received a 74 or 78 dB prepulse for 20 ms, and the pulse stimulus was 110 or 120 dB for 40 ms. Homozygotes presented significantly elevated PPIs with 78 dB prepulses and 110 dB pulse stimuli (Suppl. Figure 21).

Hyperactivity in the home cage test: Persistent activity levels in Gasc1 homozygotes

The hyperactivity of homozygotes was further detected via home cage social interaction tests. Two mice of the same genotype (e.g., two wild-type or two homozygous mice) were placed in a home cage with 12-h light and 12-h dark cycles. Locomotor activity was monitored by an infrared video camera and quantified via Image HA software. The activity of wild-type mice increased during the dark period, with 2 peaks per day at 19H and 6H around the “light to dark” and “dark to light” transitions. Homozygotes exhibited hyperactivity with 2 peaks a day during their nocturnal active phase (Fig. 3C and D), although their first peak shifted to the mid-dark period (the second peak at the end of the dark period (6H) did not shift). There was no difference in locomotor activity between the wild-type and homozygous mice during the light period. There was no significant difference in the interaction time between wild-type and homozygous mice (Fig. 3C and Suppl. Figure 22). When wild-type mice were kept in black boxes, the circadian rhythm of behavior gradually shortened, but conversely, it tended to lengthen in homozygotes (Suppl. Figure 23).

Methylphenidate decreased hyperactivity: Reduction of hyperactivity in Gasc1 homozygotes

De novo mutation and linkage analyses of human ASD have been reported. Interestingly, they included a chromosomal deletion of 78 genes encompassing the GASC1 gene on chromosome 9p24.1 and SNPs/CNVs near or within the GASC1 locus, suggesting that GASC1 is a novel candidate ASD susceptibility gene. Methylphenidate (MPH) is a dopamine transporter and norepinephrine transporter blocker that is commonly prescribed for attention-defective hyperactive disorders and for hyperactivity in children with autism and pervasive developmental disorders. Since the major phenotype of the Gasc1 homozygous mutants was hyperactivity, MPH was administered to assess its efficacy against the mouse phenotype (Fig. 4C). The mice housed in home cages were given an intraperitoneal injection of MPH or saline (0.1 ml/10 g body weight, which is equivalent to 1 mg of MPH/kg) at approximately 17:45 (between 17:20 and 18:05 due to the technical constraints of sequentially injecting many mice), approximately 1 h before the beginning of the dark cycle (the end of the light cycle). Locomotor activity from 23:45–01:45 accumulated and plotted when the mutants presented the highest activity (Suppl. Figure 24). MPH decreased homozygote activity to the level of the control mouse, whereas the same amount of methylphenidate had no effect on the activity of the wild-type mice, suggesting that the cause of mouse hyperactivity is related to the hyperactivity of the human attention-defective hyperactive disorder and hyperactivity observed in children with autism and pervasive developmental disorders.

Increased dendritic spine density: Synaptic alterations in Gasc1 mutants

Increased spine densities have been reported in human postmortem studies of ASD and psychiatric patients. This motivated us to analyze spine density in Gasc1 mutant mice. The brains of 10- to 12-week-old Gasc1 mutant and wild-type male mice were subjected to Golgi staining (Fig. 5). The number of dendritic spines in the CA1 stratum radiatum was manually counted on Z-stack images captured by a BIOREVO microscope (BZ- 9000, Keyence, Japan). The spine density was slightly but significantly greater in Gasc1 mutants than in wild-type controls.

Fig. 5.

Increased spine density of CA1 pyramidal neurons in the stratum radiatum of Gasc1 homozygous hippocampi. A, A drawing of a hippocampal and CA1 pyramidal neuron. The number of spines on secondary dendrites in the stratum radiatum of 10- to 12-week-old male mice was counted. B, Representative images of CA1 neurons and magnified images of the red boxed area showing spines on dendrites. C, Spine number per 1 microm on each dendrite. A total of 148 dendrites from 6 wild-type mice and 149 dendrites from 6 mutant mice were measured, and the density of each spine is indicated by a dot in the graph. There was a significant increase in spine density in the mutant hippocampi. D, The average spine density of 6 wild-type and 6 mutant mice was plotted. A significant increase in mutant spine density was observed

Enhanced long-term potentiation and attenuated paired-pulse facilitation: Synaptic plasticity changes in Gasc1 mutants

Gasc1-mutant mice displayed deficits in spatial learning and memory, which was often discussed along with hippocampal functions. They further showed increase in the dendritic spine density in the stratum radiatum of the hippocampal CA1 region (Fig. 5). These findings encouraged us to study the long-term potentiation (LTP) of synaptic transmission between Schaffer collaterals and CA1 pyramidal neurons in the mutants. Field excitatory postsynaptic potentials (fEPSPs) elicited by Schaffer collateral stimulation were recorded in the stratum radiatum of the CA1 region. High-frequency tetanic stimulation (100 Hz, 1 s) resulted in significantly greater potentiation of synaptic transmission in the mutant slices than in the wild-type control slices (Fig. 6A). The potentiation of the fEPSP in the wild-type slices was attenuated immediately after tetanic stimulation; however, the fEPSP amplitude decay was slower in the mutant slices, resulting in enhanced LTP in the mutant slices.

Fig. 6.

Increased LTP and attenuated PPF were detected in Gasc1 homozygous hippocampal sections. Stimulating and recording electrodes were placed in the section to record the field excitatory postsynaptic potential (fEPSP) of Schafer collaterals to CA1 pyramidal dendrites. A After tetanic stimulation, increased LTP was detected in the mutant slices. For the measurement of PPF, we used interstimulus intervals of 25, 50, 100, 200, 400, 800, and 1000 ms. B Reduced PPF was detected in the Gasc1 mutant slices. Twelve slices from 6 wild-type and 16 slices from 7 mutant mice were used for LTP recordings, and 12 slices from 4 wild-type and 17 slices from 5 mutant mice were used for PPF recordings

LTP is an index of postsynaptic function at Schaffer collateral-CA1 pyramidal synapses, whereas paired-pulse facilitation (PPF) is an index of presynaptic release probability. Activation with a pair of stimuli evokes synaptic currents, with the second response being greater than the first response. The relationship between the magnitude of the PPF ratio (second/first fEPSP amplitudes) in the Schaffer collateral-CA1 glutamatergic synapses and the interpulse interval indicates a reduced PPF in the Gasc1 mutant slices (Fig. 6B), suggesting a decrease in the number of readily releasable synaptic vesicles, inactivation of release sites, and/or abnormalities in calcium regulation at the presynapse side.

Microarray analyses: Gene expression profiles and ASD-relevant pathways

Disruption of GASC1 histone demethylase activity affects histone methylation status (Suppl Fig. 4). The abnormal gene expression caused by Gasc1 mutation appears to be directly related to the abnormal behaviors of the mutant mice and several symptoms of human ASD patients. Thus, we compared the gene expression profiles of the brains of the mutant and wild-type control mice used for the behavior test battery. We first chose the hippocampi, where abnormal synaptic functions and spine numbers were detected. Total RNA prepared from 3 Gasc1 mutant hippocampi and 3 wild-type hippocampi at 12 months of age was subjected to an Agilent gene expression array and analyzed via GeneSpring GX 10.0, which detected 1258 probes (842 genes) with 1.5-fold higher or lower levels of expression in the mutant than in the wild type (Suppl. Figure 25 A, Suppl. Table 2). Among them, 609 genes were identified via the Ingenuity Pathway Analysis (IPA) ™ database, and 585 genes with functional information were identified. IPA functional analysis revealed significant changes in “psychological disorders”, “genetic disorders”, “nervous system development and function”, and “behavior”, which are relevant to ASD symptoms (Suppl. Figure 26; Suppl. Table 3). Total RNA was also extracted from the cerebra, amygdala, and hypothalami of 3 Gasc1 mutants and 3 wild-type strains at 12 months of age and subjected to microarray analysis. We detected a total of 2949 genes (4145 probes) that were differentially expressed (1.5-fold higher or less) in any region of the mutant brain compared with the wild-type control (Suppl. Figure 25B, C, D). Forty-four genes (47 probes) were commonly overexpressed in the 4 regions of the mutant brain (hippocampus, cerebrum, amygdala and hypothalamus) by more than 1.5-fold, and 31 genes (38 probes), including Gasc1 itself, were commonly underexpressed in the mutant 4 regions by less than 0.67-fold (Suppl. Table S2). The microarray data have been deposited with DDBJ GEA (https://www.ddbj.nig.ac.jp/index-e.html), and the accession number is E-GEAD- 879.

Discussion

This study demonstrated that the reduced expression of the histone demethylase Gasc1 gene led to abnormal behaviors with increased spines and synaptic dysfunction, suggesting its important roles in increased brain function. These data suggest that histone methylation, in addition to DNA methylation (e.g., Rett syndrome), plays important roles in normal brain development and increased brain function in both mice and humans. GWAS of human ASD identified copy number variations (CNVs) [12–14] and single nucleotide polymorphisms (SNPs) [15, 16] at the 9p24.1 locus, which includes the human GASC1 gene. Thus, the Gasc1 mutant mouse may be a useful animal model for investigating ASD pathogenesis and developing novel therapeutics. The short-term effect of methylphenidate on the mutant phenotype supported this idea. The comparative analysis of the GASC1-regulated downstream genes and the human ASD susceptibility genes listed by the Simons Foundation Autism Research Initiative (SFARI) provides valuable information for assessing genes that are directly associated with ASD symptoms (Suppl. Figure 27).

GASC1 has been reported to catalyze the demethylation of histone H3 K9 me3 and H3 K9 me2. H3 K9 me3 and H3 K9 me2 are widely recognized as gene silencing marks. We detected increased levels of histone H3 K9 me3 in the cerebra and brainstem and H3 K9 me2 in the cerebra at the Western blot level (Fig. 2E, Suppl. Figure 4), indicating that the demethylation activity of GASC1 might have broad effects on chromatin. In contrast, the expression of a limited number of genes was changed in the mutant brains, as detected by microarray analysis, suggesting that GASC1 may have specific target sites. Since GASC1 by itself has no DNA binding motif, it can access DNA through some transcription factors that bind to specific nucleotide sequences. It was previously demonstrated that GASC1 assembled with ligand-bound androgen receptor and that LSD1 on androgen receptor target genes induced the removal of methyl groups from mono-, di-, and trimethylated H3 K9 and thus stimulated androgen receptor-dependent transcription [18]. As described above, GASC1 was also demonstrated to bind to the Nanog gene and regulate its expression in ES cells [3]. These reports suggest that GASC1 can bind and regulate specific sets of genes engaged by its binding partners. Indeed, the expression of Sfrp1 (secreted frizzled-related protein 1), which was reported to be involved in the maintenance and differentiation of embryonic stem cells in association with Nanog, was increased in the Gasc1-mutnat brains. In addition, several other GASC1-binding proteins have been demonstrated to also bind to HIF1 and MyoD; however, the binding partners and/or target sequences of GASC1 are not yet fully understood. Our microarray analysis demonstrated the increased expression of Slc25a37 (Solute Carrier Family 25 Member 37, also known as Mitoferrin- 1) and the decreased expression of Hpgd (15-hydroxyprostaglandin dehydrogenase), which are downstream genes of HIF1 transcription factor. The increased expressions of the MyoD downstream genes Spag9 (Sperm Associated Antigen 9) and Frg1 (FSHD Region Gene 1) were detected in Gasc1 mutant brains. However, none of them have been reported to be directly involved in ASD. Future studies are expected to clarify all of these molecular interactions.

Our microarray analysis revealed both gene silencing and activation in Gasc1 mutant brains. As discussed above, the accumulation of histone H3 K9 me3 caused by GASC1 dysfunction leads to gene silencing, and its secondary effects may promote gene activation. It is also possible that the accumulation of histone H3 K36 me3, another target of GASC1 to a lesser extent, promotes the transcriptional activities of various genes. Interestingly, the Gasc1 hypomorphic mutant mouse brains presented markedly increased expression levels of genes related to inflammation. For example, Ccl21a (C–C motif chemokine ligand 21), which showed the greatest increase in expression, is known to play an important role in immune and inflammatory responses, especially in lymphocyte and T-cell recruitment. On the other hand, the expression of Hpgd was markedly decreased in the brains of Gasc1 hypomorphic mutants. Hpgd is involved in prostaglandin metabolism and plays an important role in inflammatory responses. In addition, we previously reported that a marked increase in GFAP-positive reactive astrocytes was observed in the cortex of Gasc1 hypomorphic mutant mice [17], which may be related to the changes in inflammatory gene expression mentioned above. Indeed, in Gasc1 mutant brains, abnormal expressions of genes thought to be expressed in glial cells, including Sfrp1 (secreted frizzled-related protein 1), ptprg (protein tyrosine phosphatase, receptor type G), cd59a (CD59 antigen), and enpp6 (ectonucleotide pyrophosphatase/phosphodiesterase 6). More careful analysis would be needed to address whether these genes are involved in ASC or ADHD symptoms.

Gasc1 mutant mice exhibit deficits in functions that span multiple domains, including memory, spatial reasoning, and social interaction. All of these higher brain functions are supported by the cerebral cortex and hippocampus. In particular, the involvement of hippocampus in memory and spatial reasoning has been widely known and the relationship between hippocampal function and social behavior has also attracted attention, recently. In other words, it is very likely that impaired hippocampal function contributes to many of the symptoms seen in ASD.

Altered dendritic spine morphology and density have been reported in human ASD and ADHD. Such synaptic abnormalities may contribute to cognitive and social behavioral disorders such as working memory deficits. In particular, increased spine density in the hippocampal CA1 region is closely related to cognitive functions such as learning, memory, and social behavior. Increased spine density in CA1 neurons often correlates with increased excitatory signaling, leading to excitatory/inhibitory imbalance, which may explain the behavioral abnormalities observed in Gasc1 mutant mice. Further careful analysis would be required to explain the phenotype of Gasc1 mutants along this direction.

In terms of histone demethylation and gene regulation, we would like to mention that the hippocampus of Gasc1 mutant mice has altered expression of the genes encoding proteins involved in synaptic plasticity and neuronal structure, as follows. Eps8 l1 (Epidermal Growth Factor Receptor Pathway Substrate 8-Like Protein 1) is involved in the reorganization of the actin cytoskeleton and may affect spine morphology and density. Apol7c (Apolipoprotein L7 C) is involved in lipoprotein metabolism and may affect brain function and synaptic plasticity. Wdfy1 (WD Repeat and FYVE Domain Containing 1) is involved in autophagy and may affect synaptic function and plasticity. It is also necessary to consider abnormalities in synaptic pruning in addition to abnormalities in spine formation. Further experiments, including the use of animal models of disorders, are needed to fully elucidate their molecular mechanisms and to develop effective treatments.

We analyzed hypomorphic mutant mice with reduced expression of the Gasc1 gene because Gasc1 null mutants are thought to be embryonic lethal due to functional defects in early embryo development [3–5]. Interestingly, a recent report demonstrated that conditionally targeted Jmjd2c (Gasc1)-deficient ES cells and mice, in which Jmjd2c/Gasc1 exon 9 was deleted, were practically viable and fertile, suggesting that Gasc1 deficiency was compatible with embryonic and postnatal development [19]. The phenotype of these mutant mice appears to be similar to that of our Gasc1 hypomorphic mutant mice, from the point that homozygous mice were born with approximately the expected frequency in both cases. However, our mice differ significantly in that a comprehensive behavioral test battery revealed a variety of behavioral abnormalities. In mammals, four KDM4 family members, KDM4 A, KDM4B, KDM4 C (GASC1), and KDM4D, exhibit overlapping expression in various tissues and organs and may play redundant roles. Therefore, multiple gene knockouts could lead to more severe phenotypes even in the early developmental stages.

Many reports suggest the oncogenic roles of GASC1 in cancer development and progression. The GASC1 gene was originally identified as a gene that is frequently amplified in human esophageal squamous cell carcinomas [2]. Recently, several reports have demonstrated GASC1 gene amplification in medulloblastomas [20, 21], breast cancers [22], B-cell lymphomas and Hodgkin lymphomas [23]. The expression levels of GASC1 are significantly high in aggressive breast cancers and are involved in cell growth [22]. Ectopic expression of GASC1 in immortalized, nontransformed mammary epithelial MCF10 A cells induces growth factor- and anchorage-independent proliferation and growth [22]. One of the GASC1 targets in MCF10 A cells is NOTCH1, which is responsible for stem cell self-renewing proliferation; thus, the role of GASC1 may be linked to stem cell phenotypes. The molecular interaction of GASC1 with the androgen receptor has been reported in prostate cancer [18], where GASC1 promotes the transcription of androgen receptor-dependent genes and cell proliferation. Thus, amplification and increased expression of GASC1 gene have been reported in cancer; however, the hypomorphic mutant mice analyzed in this study showed reduced expression of Gasc1, and no cancer formation was detected. Therefore, it seems that the behavioral abnormalities of Gasc1 mutant mice are not related to cancer formation. Although it is not directly related to carcinogenesis, low expression of the Gasc1 gene may have a negative effect on cell proliferation. Supplementary Figure S5. C and D show the developmental profile of body weight. Gasc1 homozygotes show a decrease in body weight 3 weeks after weaning.

In our study, abnormal gene expression caused by Gasc1 mutation was detected in the brains of Gasc1 hypomorphic mutants. Further analysis of the upregulated and downregulated genes and their targets will increase our understanding of the function of this histone demethylase in higher-order brain functions.

Materials and methods

Animals

The Gasc1 mutant mouse strain was obtained from Transgenic, Inc. (Kobe, Japan). It was established by insertion of the gene trap vector pU- 17 into intron 1 of the Gasc1 gene in TT2 ES cells. pU- 17 contains a splice acceptor signal, β-geo, and poly A signal of mouse phosphoglycerate kinase- 1 (PGK), which are floxed by modified loxP sites [24]. Gasc1 mutant mice were maintained by backcrossing into the C57BL/6 N strain. All procedures described here were approved by the Animal Care and Use Committee of the National Institute for Physiological Science (approval number: 09 A090, 10 A112) and the relevant committee at each participating institution, which is based on national laws and guidelines, such as the Act on Welfare and Management of Animals; Standards for Care and Keeping; mitigation of the suffering of laboratory animals; basic guidelines for performing animal experiments at research institutions; and guidelines for the proper execution of animal experiments in Japan.

Immunohistochemistry

The mice were deeply anesthetized via the intraperitoneal injection of pentobarbital. The mice were perfused with 4% paraformaldehyde. The brains were dissected and postfixed in 4% paraformaldehyde overnight. After cryoprotection with serial concentrations of sucrose solution, 20-μm cryostat sections were made and pretreated with blocking buffer (phosphate-buffered saline (PBS) with 5% normal sheep serum) for 1 h at room temperature. Primary and secondary antibodies were diluted in PBST (PBS with 0.1% Tween 20 or Triton X- 100).

Antibodies

The following primary antibodies were used in this study: rabbit polyclonal anti-β-galactosidase, mouse monoclonal anti-NeuN (CHEMICON, CA, USA), anti-Tbr2 (Abcam, Cambridge, MA, USA), mouse monoclonal anti-MAP2ab (Sigma, St, USA), mouse monoclonal anti-neuron specific βIII-tubulin (COVANCE, New Jersey, USA), mouse monoclonal anti-Ki67 (BD Pharmingen, New Jersey, USA), rabbit polyclonal anti-GFAP (Dako, Denmark), anti-histone H3 K9 me3 (Abcam, Cambridge, MA, USA), anti-histone H3 K9 me2 (Upstate, NY, USA), anti-histone H3 K9 me (Upstate, NY, USA), anti-Tbr1 (Abcam, Cambridge, MA, USA), anti-Ctip2 (Abcam, Cambridge, MA, USA), and anti-Cux1 (Santa Cruz, CA, USA) antibodies. Alexa-conjugated secondary antibodies (Molecular Probes, OR, USA) were used to visualize the antigens.

X-gal staining

The mice were deeply anesthetized via the intraperitoneal injection of pentobarbital and perfused with 0.2% paraformaldehyde. The brains were dissected and postfixed in 0.2% paraformaldehyde overnight. After cryoprotection with 10%, 20% and 30% sucrose in PBS, 20-μm cryosections were made and immersed for 10 min in the detergent rinse solution: 0.1 M phosphate buffer (pH 7.3), 2 mM MgCl₂, 0.01% sodium deoxycholate, and 0.02% NP- 40 (Sigma N6507). The sections were immersed overnight in the X-Gal solution as follows: X-gal solution: 0.1 M phosphate buffer (pH 7.3), 2 mM MgCl_2, 0.01% sodium deoxycholate, 0.02% NP- 40 (Sigma N6507), 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, and 1 mg/ml X-gal.

Electrophysiological studies

The mice were euthanized via rapid decapitation. Fresh brains were immersed in cutting solution containing (in mM) 120 choline Cl, 3 KCl, 1.25 NaH₂PO₄, 8 MgCl₂, 26 NaHCO₃, and 20 glucose and sectioned at a 400 μm thickness via a vibratome-type tissue slicer at 0 °C. Slices were kept and recorded in artificial cerebrospinal fluid (ACSF, in mM) at room temperature (23–25 °C): 124 NaCl, 5 KCl, 1.25 NaH₂PO₄, 2 MgCl₂, 2 CaCl₂, 26 NaHCO₃, and 20 glucose, bubbled continuously with a mixture of 95% O₂/5% CO₂. A bipolar tungsten stimulating electrode and a glass recording electrode filled with ACSF were placed in the slice to record the field excitatory postsynaptic potential (fEPSP) of Schaffer collaterals to CA1 pyramidal dendrites. In the LTP study, test stimulation was delivered every ten seconds. If the average in any 2 min period during the 20 min baseline period just before LTP induction stimuli exceeded ± 5% of the baseline average, the records were discarded. For the measurement of PPF, we used interstimulus intervals (ISIs) of 25, 50, 100, 200, 400, 800, and 1000 ms. Twelve slices from 6 wild-type and 16 slices from 7 mutant mice at 14–18 weeks of age were used for LTP recordings. Twelve slices from 4 wild-type and 17 slices from 5 mutant mice at 10–15 weeks of age were used in the PPF recordings.

Golgi staining

Golgi staining was performed via an FD Rapid GolgiStain™ Kit (FD NeuroTechnologies, Inc., Columbia MD) according to the manufacturer’s instructions.

Behavioral tests

Behavioral tests were performed according to methods described in previous papers [25–27]. The tests were conducted with male mice at or older than 11 weeks (Gasc1 homozygous mice = 20; wild-type mice = 20). Mice with comparable body weights were selected for the tests. Two mutant and two wild-type mice per cage were maintained under a 12:12 light‒12 dark cycle unless otherwise stated. The applications used for the behavioral studies were based on the ImageJ programs and modified for each test by Tsuyoshi Miyakawa. Statview (SAS Institute, Cary, NC) was used for statistical analysis. The error bars in the graphs indicate the SEMs.

General health and neurological screening

A general health and neurological screen was performed as previously described [26]. They were evaluated by the appearance of whiskers and coats, the righting reflex, the whisker-twitch, and the ear-twitch. Visual acuity was measured by holding the mice up by their tails with their feet off the floor. Normal mice stretch their forefeet toward the floor; however, blind mice do not. In some mutant mouse strains, sound stimulation with key jangling causes seizures, but it has no effect on Gasc1 mutant mice.

The neuromuscular strength was examined via wire hanging and grip strength tests according to methods described previously [25]. Briefly, in the wire hang test, the mice were placed on wire mesh, and then, the mesh was inverted. The latency to fall was recorded. A grip strength meter (O’Hara & Co., Tokyo, Japan) was used to measure forelimb grip strength. The mice that grasped a wire grid with their forelimbs were gently pulled backward by the tails until they released the grid. The peak force was recorded in Newtons [28].

Light/dark transition test

Anxiety-like behavior was examined via a light/dark transition test. The detailed method was previously reported with an online video file [29]. Briefly, a mouse box (21 × 42x25 cm) was equally separated into two areas by partitioning with a door (O’Hara & Co.). One area was brightly illuminated (390 lx), and the other was kept in the dark (< 2 lx). A mouse was placed in the dark area and allowed to freely move between the two areas for 10 min. The total number of transitions, time spent in each area, first latency to the light side, and distance traveled were recorded via Image LD software.

Open field test

Locomotor activity in a novel open field environment was measured. A mouse was placed in the open field apparatus with photobeam scanning arrays to automatically monitor locomotion (40 × 40x30 cm; AccuScan Instruments, Columbus, Ohio). The total distance traveled, vertical activity (rearing measured by beam-break counts), time spent at the center, and stereotypic behaviors (number of repetitive interruptions of the beams) were recorded for 120 min.

Elevated plus maze

The elevated plus maze test was used to measure anxiety-like behavior. The detailed procedures for conducting the test were previously reported with an online video file [29]. The apparatus comprised 2 open arms and 2 closed arms, which crossed each other in the middle. The closed arms had high (15 cm) transparent walls to enclose the arms. The apparatus was set at 50 cm above the floor. Each mouse was placed in the center area of the maze and video recorded for 10 min. The distance traveled, the percentage of entries into the open arm, and the percentage of time spent in each arm were calculated via the Image EP program.

Plate test

Pain sensitivity was evaluated using a 55 °C hot plate (Columbus Instruments, Columbus, OH). The latency to first paw response, such as a foot shake or a paw lick, was recorded.

Social interaction test

Two mice with identical genotypes (2 wild-type or 2 mutant) were placed in a novel open field environment and allowed to move freely for 10 min. The pair of mice were chosen from the different cages and first met in the open field apparatus. Social behavior was monitored through a CCD camera and analyzed via Image SI software. The number of contacts, total duration of contacts, mean duration of contact, and total distance traveled were measured.

Rotarod test

Motor coordination and balance and associative motor learning were tested with a rotarod (UGO Basile Accelerating Rotarod, Varese, Italy). Each mouse was placed on a 3 cm diameter rotating drum. The rotation speed was increased from 4 to 40 rpm over a 300 s period. The latency to fall from the rotarod was recorded.

Prepulse inhibition (PPI)

Sensorimotor gating was measured by prepulse inhibition (PPI) via an acoustic startle reflex measurement system (O’Hara & Co.). Each mouse was placed in a Plexiglas cylinder for 10 min (habituation). The background noise level was 70 dB. The pulse stimulus was 74 or 78 dB for 20 ms, and the pulse stimulus was 110 or 120 dB for 40 ms with an 80 ms interval. One test block consisted of a pseudorandom order of six trial types, i.e., two types for startle stimulus-only trials (110 and 120 dB) and four types for prepulse inhibition trials (74–110, 78–110, 74–120, and 78–120 dB). Six blocks were used to measure the PPI of the startle response.

Porsolt forced swimming test

Depression-like behavior was assessed via the Porsolt forced swimming test, also known as the behavioral despair test. Each mouse was placed in a Plexiglas cylinder (22 cm height × 12 cm diameter) filled with room temperature water to a height of 7.5 cm, and swimming behavior was recorded for 10 min. The same test was conducted the next day.

Barnes maze

Learning and memory abilities were tested via the Barnes maze. The Barnes task was conducted on a white circular table, 1 m in diameter, with 12 holes around the perimeter (O’Hara & Co.). A black Plexiglas escape box was located under the hole at 0° as a target. A mouse was placed at the center of the table and allowed to seek the target. Some distal visual room cues hint at the spatial location of the target. The table was rotated daily to prevent bias based on olfactory cues or any proximal cues on the maze. The trials were conducted for 6 consecutive days. On day 7, a probe test was carried out without the escape box, and the time spent around each well was recorded via video tracking software (Image BM).

T-maze

Working and reference memory were assessed by forced alternation and left‒right discrimination tasks via a T-maze. The experimental apparatus (40 × 54 × 25 cm) was partitioned into 6 areas (A1 to A6) connected by electric sliding doors. The start compartment named A5 was located in the bottom part of “T”. The stem area of T was named A2, and the left and right arms of T were named A1 and A3, respectively. Areas A4 and A6 comprise the connecting passages from the arms (A1 and A3) to the start compartment A5. The A1 and A3 compartments were equipped with pellet dispensers that provided 20 mg of sucrose pellets as rewards. One week before and during the tests, the mice were maintained on a restrictive diet to reduce their weight to 80% of the initial level.

The forced alternation task was conducted after adaptation. Each trial consisted of a forced-choice run followed by a free-choice run. In the forced-choice run, a mouse was placed in the start area A5 and forced to go into either the A1 or A3 arms where a sucrose pellet was provided. The doors were subsequently opened for the mouse to return to A5. In the following free-choice run, both doors connecting to A1 and A3 were opened; thus, the mouse was allowed to choose either area. When the mouse chose the correct arm that was not visited in the forced-choice run, it received a reward pellet. When the mouse chose the incorrect arm that was the same one visited in the forced-choice run, it was confined within the area for 10 s as a penalty. The doors were subsequently opened for the mouse to return to A5. One session consisted of 10 trials, and a total of 8 sessions were conducted.

In the left‒right discrimination task, a mouse was placed in the A5 area and given a free choice to visit the A1 or A3 arms. A reward pellet was provided in the correct arm. If a mouse chose the incorrect arm, it received no reward and was confined within the area for 10 s as a penalty. The correct/incorrect arms were fixed across trials and sessions (always left or always right) and randomly assigned to each mouse. After each trial, the doors were opened for the mouse back to A5. One session consisted of 10 trials. In total, 17 sessions were conducted over 7 days. After the 8 th session, the correct arm was changed to the opposing arm to access reversal learning, memory erasure and rewrite, and perseveration.

Contextual and cued fear conditioning

For conditioning, a mouse was placed in a test chamber (26 × 34 × 33 cm). After 2 min of habituation, it was exposed to 3 sets of 55 dB white noise for 30 s (conditioned stimulus: CS), followed by a 0.3 mA foot shock (unconditioned stimulus: US) for 2 s and 90 s intervals. When the mouse stopped moving for more than 2 s, it was judged as “freezing”. In the contextual test, the conditioned mouse was placed in the same chamber, and its behavior was recorded for 5 min. In the cued test, the conditioned mouse was placed in a triangular box (a different shape of box), and its behavior was recorded. After 3 min of habituation, it was exposed to 55 dB noise for 3 min.

Tail suspension test

The tail suspension test was conducted to assess depression-like behavior. A mouse tail was taped and suspended 30 cm above the floor. The mouse developed an immobile posture in an inescapable, stressful situation. The behavior of each sample was recorded for 10 min. Data were acquired and analyzed via Image TS software.

Twenty-four-hour home cage monitoring of social interaction and locomotor activity

Two mice of the same genotype (e.g., two wild-type mice or two mutants), born to different mothers and housed in different cages, were placed in a home cage with 12-h light and 12-h dark cycles. The activity was monitored by an infrared video camera attached to the top of the cage. Images were captured once per second, and the mice were detected as particles. Social interaction was quantified as the number of particles: two particles indicated the mice with no contact, whereas one particle indicated the mice with contact. Locomotor activity was also measured via Image HA software.

Circadian rhythm monitoring

The circadian rhythm was monitored in the home cage that was used for the social interaction test. The mice were kept under 12-h light and 12-h dark (LD) cycles for the first 17 days and then kept in constant darkness (DD) for 38 days. The activity was monitored by an infrared video camera attached to the cage top and measured once per second via Image HA software.

Methylphenidate treatment in home cages

Methylphenidate (MPH/Ritalin, Sigma M- 2892) was diluted against saline (Otsuka Pharmaceutical) at 0.1 mg/ml. The mice housed in home cages were given an intraperitoneal injection of MPH or saline (0.1 ml/10 g body weight, which is equivalent to 1 mg of MPH/kg) at approximately 17:45 (between 17:20 and 18:05 due to the technical constraints of sequentially injecting many mice), approximately 1 h before the beginning of the dark cycle (the end of the light cycle). The activity was recorded via an infrared video camera attached to the cage top via image HA software.

Abbreviations

ACSF

Artificial Cerebrospinal Fluid

Apol7c

Apolipoprotein L7 C

ASD

Autism Spectrum Disorder

Ccl21a

C-C motif chemokine ligand 21

cd59a

CD59 antigen

CNVs

Copy Number Variations

DNA

Deoxyribonucleic Acid

DDBJ

DNA Data Bank of Japan

EEG

Electroencephalography

enpp6

Ectonucleotide pyrophosphatase/phosphodiesterase 6

Eps8 l1

Epidermal Growth Factor Receptor Pathway Substrate 8-Like Protein 1

ES

Embryonic Stem

Frg1

FSHD Region Gene 1

GFAP

Glial Fibrillary Acidic Protein

GASC1

Gene Associated with Squamous Cell Carcinoma 1

GWASs

Genome-Wide Association Studies

HIF1

Hypoxia-Inducible Factor 1

Hpgd

15-Hydroxyprostaglandin dehydrogenase

ISIs

Interstimulus Intervals

JMJD2 A

Jumonji Domain-Containing Protein 2 A

JMJD2B

Jumonji Domain-Containing Protein 2B

JMJD2 C

Jumonji Domain-Containing Protein 2 C

JMJD2D

Jumonji Domain-Containing Protein 2D

KDM4 A

Lysine Demethylase 4 A

KDM4B

Lysine Demethylase 4B

KDM4 C

Lysine Demethylase 4 C

KDM4D

Lysine Demethylase 4D

LTP

Long-Term Potentiation

MEXT

Ministry of Education, Culture, Sports, Science and Technology

MPH

Methylphenidate

MRI

Magnetic Resonance Imaging

PPF

Paired-Pulse Facilitation

PPI

Prepulse Inhibition

Ptprg

Protein tyrosine phosphatase, receptor type G

RNA

Ribonucleic Acid

SEM

Standard Error of the Mean

Sfrp1

Secreted frizzled-related protein 1

Slc25a37

Solute Carrier Family 25 Member 37, also known as Mitoferrin- 1

SNPs

Single Nucleotide Polymorphisms

SNVs

Single Nucleotide Variations

Spag9

Sperm Associated Antigen 9

TMDU

Tokyo Medical and Dental University

JSPS

Japan Society for the Promotion of Science

Wdfy1

WD Repeat and FYVE Domain Containing 1

Authors’ contributions

T.K. and T.T. designed the project. J.I. provided the Gasc1 hypomorphic mutant mice. N.N. increased the number of mutant mouse colonies via early embryo manipulation. Y.Y., S.H., K.T. and T.M. performed the behavioral experiments. Y.Y., Y.K., and G.S. performed the biochemistry and molecular studies. T.K., Y.Y., Y.K., S.G., A.E., K.K. and M.W. performed the histological analyses. Y.K. and T.I. performed LTP and electrophysiological experiments. T.K. and T.T. wrote the manuscript and prepared the figures, with contributions from all coauthors. All the authors reviewed all the data and edited the manuscript.

Funding

This work was supported by JSPS KAKENHI Grant Numbers JP22650077 (TK), JP22 K06234 (TT), JP15 K14346 (TT), and JP24300119 (TT) and by MEXT KAKENHI Grant Number JP20022034 (TT). This study was also supported by the Cooperative Study Program (2009 and 2010) of the National Institute for Physiological Sciences and by Nanken-Kyoten, TMDU.

Data availability

The microarray data have been deposited with DDBJ GEA (https://www.ddbj.nig.ac.jp/index-e.html), and the accession number is E-GEAD- 879. For other data, please see the supplemental file.

Declarations

Ethics approval and consent to participate

All procedures described here were approved by the Animal Care and Use Committee of the National Institute for Physiological Science (approval number: 09 A090, 10 A112) and the relevant committee at each participating institution, which is based on national laws and guidelines, such as the Act on Welfare and Management of Animals; Standards for Care and Keeping; mitigation of the suffering of laboratory animals; basic guidelines for performing animal experiments at research institutions; and guidelines for the proper execution of animal experiments in Japan.

As no human samples were used, consent to participate is not applicable to this study.

Consent for publication

All co-authors have given consent to publish.

Competing interests

The authors declare that they have no competing interests.