Differential roles of ARID1B in excitatory and inhibitory neural progenitors in the developing cortex

Genetic evidence indicates that haploinsufficiency of ARID1B causes intellectual disability (ID) and autism spectrum disorder (ASD), but the neural function of ARID1B is largely unknown. Using both conditional and global Arid1b knockout mouse strains, we examined the role of ARID1B in neural progenitors. We detected an overall decrease in the proliferation of cortical and ventral neural progenitors following homozygous deletion of Arid1b, as well as altered cell cycle regulation and increased cell death. Each of these phenotypes was more pronounced in ventral neural progenitors. Furthermore, we observed decreased nuclear localization of β-catenin in Arid1b-deficient neurons. Conditional homozygous deletion of Arid1b in ventral neural progenitors led to pronounced ID- and ASD-like behaviors in mice, whereas the deletion in cortical neural progenitors resulted in minor cognitive deficits. This study suggests an essential role for ARID1B in forebrain neurogenesis and clarifies its more pronounced role in inhibitory neural progenitors. Our findings also provide insights into the pathogenesis of ID and ASD.

Intellectual disability (ID) and autism spectrum disorder (ASD) affect between one and three percent of the global population 1,2 . These and other related neurodevelopmental disorders represent a significant emotional and financial burden for affected individuals and their families 1 . Unfortunately, treatment for these conditions remains limited because many of the key molecular factors and their associated pathogenic mechanisms are still poorly understood. ARID1B is a sequence-specific, DNA-binding subunit in mammalian SWI/SNF or Brg1associated factors (BAF) chromatin-remodeling complexes [3][4][5] . Neurogenetic studies have shown that ARID1B haploinsufficiency causes ID and ASD [6][7][8] . Because of the clear genetic linkage, defining the function of ARID1B in brain development is a crucial step toward understanding the neurological and developmental mechanisms responsible for these pathogenic phenotypes.
During brain development, forebrain excitatory and inhibitory neurons are generated in distinct brain regions and migrate along separate pathways before converging in the cerebral cortex. Excitatory neurons are born in the ventricular zone (VZ) of the developing cerebral cortex and migrate radially into the cortical plate, usually along radial glial processes [9][10][11][12][13] . Most inhibitory interneurons (GABAergic neurons) originate from a population of neural progenitors within the medial ganglionic eminence (MGE) of the ventral telencephalon and migrate tangentially into the dorsal telencephalon [14][15][16] . Cortical and ventral neural progenitors both need to be tightly regulated to ensure proper brain development as they have distinct and complementary roles in the mature brain and are each under the control of different pathways 17,18 . The balanced and coordinated function of pyramidal neurons and interneurons regulates excitatory and inhibitory tones in the brain. An imbalance of neuronal excitation and inhibition (E/I imbalance) in the developing brain underlies the neurological dysfunctions observed in ASD and ID [17][18][19][20] . Importantly, the numbers of these excitatory and inhibitory neurons are determined by the proliferation of cortical and ventral neural progenitors, respectively, in the developing brain 21 . We previously reported a significant decrease in the total number of GABAergic interneurons in the cerebral cortex of Arid1b Scientific Reports | (2021) 11:3856 | https://doi.org/10.1038/s41598-021-82974-y www.nature.com/scientificreports/ haploinsufficient mice, suggesting that E/I imbalance may play a role in the pathology of ARID1B-related neurodevelopment disorders. In order to separately identify the neurobiological function of ARID1B in excitatory and inhibitory neurons and their progenitors, we used conditional knockout mouse models to explicitly delete the Arid1b gene in either cortical or ventral neural progenitors. In this study, we utilize an Emx1-Cre driver line to conditionally delete Arid1b in cortical neural progenitors and a Dlx5/6-Cre driver to knockout Arid1b in ventral neural progenitors [22][23][24][25] . We report impaired proliferation in the cortical neural progenitor population and, to a greater extent, in ventral neural progenitors. This may be due to altered cell cycle regulation, as we observe decreased cell cycle speed in ventral neural progenitors with homozygous Arid1b deletion and a decreased rate of cell cycle re-entry in both cortical and ventral neural progenitors. In both progenitor populations we also report an increased number of apoptotic cells. Homozygous deletion of Arid1b in ventral inhibitory progenitors leads to ID-and ASD-like behavioral phenotypes, similar to those seen in Arid1b haploinsufficient mice [26][27][28] . Knockout of Arid1b in cortical excitatory progenitors, in contrast, has little effect on the mouse behaviors we measured. Taken together, Arid1b conditional homozygous deletion has an outsize effect on ventral progenitor proliferation, which is intricately linked to animal behavior, whereas homozygous loss of Arid1b in cortical progenitors gives rise to comparatively moderate neural and behavioral phenotypes.

Results
Decreased cortical progenitor proliferation in Emx-Arid1b mice. Conditional deletion of Arid1b in cortical progenitors was accomplished by crossing mice heterozygous for Emx1-Cre, which drives the expression of Cre recombinase in progenitors that give rise to cortical excitatory neurons 23 , with Arid1b LoxP/LoxP mice 26 . Using Western blotting, we confirmed conditional knockout of ARID1B in mutant samples ( Supplementary  Fig. 1A).
We first examined the proliferation of cortical neural progenitors in the VZ of the dorsal telencephalon from Arid1b LoxP/LoxP ;Emx1-Cre + (Emx-Arid1b) mice by immunostaining consecutive cortical sections from E14.5-E15.5 mouse embryos. Background cells were marked with a DAPI nuclear stain (Fig. 1A). DAPI labeling did not reveal any significant decrease in cells in the VZ of the developing cerebral cortex of Emx-Arid1b mice (263.8 cells), compared with Arid1b LoxP/LoxP ;Emx1-Cre − (Emx-WT) control littermates (274.0 cells) (Supplementary Fig. 2A). Staining with an anti-phosphorylated Histone-H3 (PH3) antibody, which is used to determine cells undergoing mitosis 29 , revealed a significant reduction (p < 0.05) in the percentage of DAPI-labeled mitotic cortical neural progenitors in Emx-Arid1b mice (4.162%), compared with controls (7.487%) (Fig. 1B,C). Staining for Ki67, which is present during all stages of the cell cycle and absent in quiescent (G 0 ) cells 30,31 , showed no significant difference in the percentage of cells undergoing active proliferation in the VZ of Emx-Arid1b mice (12.67%), compared with controls (11.88%) (Fig. 1B,C). We also peritoneally injected all pregnant dams with bromodeoxyuridine (BrdU), a thymidine analog that is incorporated into dividing cells during DNA replication [32][33][34] , and found no significant decrease in the percentage of BrdU-positive cortical neural progenitors in Emx-Arid1b mice harvested 1 h postinjection (20.04%), compared with controls (29.67%) (Fig. 1B,C). To further examine the cortical progenitor population, we immunostained sections of the developing cerebral cortex with an antibody against a marker for intermediate progenitors, Tbr2, and discovered a significant decrease in this population in Emx-Arid1b mice (126.9 cells), compared with controls (150.4 cells) (Fig. 1D,E). Impaired ventral progenitor proliferation in Dlx-Arid1b mice. We utilized Arid1b LoxP/LoxP ;Dlx5/6-Cre + (Dlx-Arid1b) mice to examine the effects of homozygous deletion of Arid1b in ventral progenitors that give rise to cortical GABAergic inhibitory interneurons. Conditional knockout was confirmed by Western blotting (Supplementary Fig. 1B). In E14.5-E15.5 Dlx-Arid1b mice, we first showed that Dlx-Arid1b mice do not display a reduction in the number of likely progenitor cells in the VZ of the MGE (344.3 cells), compared with Arid-1b LoxP/LoxP ;Dlx5/6-Cre − (Dlx-WT) control littermates (330.7 cells) ( Supplementary Fig. 2B). We next examined ventral neural progenitor proliferation by immunostaining for PH3, Ki67 and BrdU in the VZ of the MGE. Background cells were marked by DAPI ( Fig. 2A). We observed a significant decrease (p < 0.05) in the percentage of cells positive for PH3 (3.852%) (Fig. 2B,C), as well as a significant decrease (p < 0.001) in the percentage of cells expressing Ki67 (7.643%) (Fig. 2B,C), both compared with controls (8.829% and 26.090%, respectively). These indicate large reductions in the proportion of mitotic and actively proliferating ventral neural progenitors in these mice. Following BrdU injection into all Arid1b LoxP/LoxP ;Dlx5/6-Cre + (Dlx-Arid1b) pregnant dams, displayed a significant decrease (p < 0.05) in the percentage of cells positive for BrdU in the VZ of the MGE (23.57%), compared with controls (66.76%) (Fig. 2B,C). Consequently, we also observed a significant decrease (p < 0.001) in parvalbumin (PV)-positive interneurons in the cerebral cortex of P30 Dlx-Arid1b mice, compared with controls (Fig. 2D,E). We did not, however, detect any change in the total number of neurons in the cerebral cortex at P30, as shown by NeuN immunostaining (Supplemental Fig. 3A,B). Together, these results show a reduction in the number of proliferating ventral neural progenitors in the MGE, leading to a lower number of PV-positive interneurons in the adult cerebral cortex.
ARID1B regulates cell cycle progression in both cortical and ventral neural progenitors. Next, we explored the effects of conditional Arid1b knockout on cell cycle progression. Cell cycle speed and cell cycle re-entry were examined in both Emx-Arid1b and Dlx-Arid1b mice. Cell cycle speed was assessed by first injecting pregnant dams with BrdU 30 min prior to removal of E14.5 embryos, followed by immunostaining of brain sections using both anti-BrdU and anti-Ki67 antibodies. The ratio of Ki67-BrdU-double-positive cells to the total number of Ki67-positive cells, essentially the proportion of actively proliferating cells that entered S-phase in a 30 min time span, provides a quantifiable estimate of cell cycle speed 35,36 . Homozygous knockout of Arid1b  www.nature.com/scientificreports/ in cortical neural progenitors had no significant impact on cell cycle speed in the VZ of the developing cerebral cortex (Emx-Arid1b: 0.2526 vs. Emx-WT: 0.1867) (Fig. 3A,B). In the ventral neural progenitor population, however, Arid1b deletion led to a 50.33% reduction (p < 0.01) in this measurement of cell cycle speed, compared with control littermates (Dlx-Arid1b: 0.1599 vs. Dlx-WT: 0.3213) (Fig. 3E,F).
To quantify cell cycle re-entry, we injected pregnant dams with BrdU 24 h prior to euthanasia and then co-immunostained E15.5 embryonic brains with anti-Ki67 and anti-BrdU antibodies. The ratio of Ki67-BrdUdouble-positive cells to the total number of BrdU-positive cells provides a measure of the cells that re-enter the cell cycle 35,36 . There was a 46.06% (p < 0.05) reduction in BrdU-positive cortical neural progenitors that express measurable levels of Ki67 in Emx-Arid1b mice (0.1637), compared with controls (0.3029), which indicates impaired cell cycle re-entry (Fig. 3C,D). We further examined the effects of this reduction in cell cycle re-entry amongst Emx-Arid1b mice by immunostaining the cerebral cortex for a marker of superficial pyramidal neurons, Cux1, in P30 animals, and found that there is a significant decrease (p < 0.001) in Cux1-positive neurons in Emx-Arid1b mice (427.1), compared with controls (559.7) (Fig. 4A,B). Ventral neural progenitors also showed an impairment in cell cycle re-entry, as they had fewer BrdU-positive ventral neural progenitors positive to Ki67 (− 71.48%) in Dlx-Arid1b mice (0.1405), compared with Dlx-WT littermates (0.4925) (Fig. 3G,H).

Conditional homozygous deletion of Arid1b leads to increased apoptosis in cortical and ventral neural progenitors.
In addition to examining the proliferation of cortical and ventral neural progenitors, we assessed apoptotic progenitors in the E14.5-E15.5 developing cerebral cortex and ventral telencephalon in conditional Arid1b mutants. We used an anti-cleaved-Caspase 3 (anti-cl.-Cas 3) antibody to determine the relative percentage of cells undergoing apoptosis. In Emx-Arid1b mice, the percentage of DAPI-labeled cells in the VZ of the dorsal telencephalon co-expressing cl.-Cas 3 (26.72%) was significantly higher (p < 0.01) than in wild type mice (11.20%) (Fig. 5A,B). In Dlx-Arid1b mice, the percentage of cells expressing cl.-Cas 3 in the VZ of the MGE (30.13%) was likewise significantly higher than in controls (11.36%) (Fig. 5C,D). From these results we conclude that homozygous deletion of Arid1b leads to a comparable increase in apoptosis among both cortical and ventral neural progenitor populations.

Arid1b knockout decreases β-catenin nuclear localization in vitro and in vivo.
We previously showed that Arid1b haploinsufficiency leads to a significant reduction in β-catenin protein expression and the expression of several Wnt/β-catenin signaling target genes in the ventral telencephalon 26 . Two other groups have also shown that ARID1B is involved in the regulation of β-catenin signaling 37,38 . Here we assessed the localization pattern of the transcription factor β-catenin because it requires nuclear translocation in order to regulate downstream gene expression. We cultured primary neural progenitors from the MGE of E12.5 Arid1b +/+ and Arid1b −/− brains, and immunostained unstimulated cells using an anti-β-catenin antibody. Qualitative analysis revealed decreased β-catenin in the DAPI-stained nuclei of Arid1b −/− cells, compared with Arid1b +/+ controls. We determined the object-corrected fluorescence colocalization of β-catenin and DAPI 39 . The mean Pearson's coefficient for Arid1b +/+ cultures was 0.61 and 0.20 for Arid1b −/− cultures, indicating significantly reduced nuclear β-catenin in the absence of ARID1B (p < 0.01) (Fig. 6A,B). Next, we examined nuclear β-catenin localization in the mouse brain by immunostaining MGEs of E12.5 Dlx-Arid1b mice and comparing DAPI-β-catenin colocalization with control MGEs. Similar to what we observed in vitro, we detected significantly less (p < 0.01) overlapping β-catenin and DAPI staining in Dlx-Arid1b MGEs than in controls (Fig. 6C,D). Pearson's coefficient of colocalization in Dlx-Arid1b MGEs was lower (0.03) than that in controls (0.27). Furthermore, we examined the expression of β-catenin downstream genes including c-Myc, n-Myc, Cyclin D1, Axin2, Lef, and Tcf4. The levels of c-Myc, n-Myc, Axin2, and Tcf4 were deceased in Emx-Arid1b samples compared with controls ( Fig. 6E). Similarly, the c-Myc, Axin2, and Lef levels were reduced in Dlx-Arid1b mutants (Fig. 6F). Taken together, we conclude that loss of Arid1b leads to aberrant β-catenin signaling in the brain.
Cortical progenitor-specific deletion of Arid1b in neural behavior. We next examined the behavioral outcomes of homozygous deletion of Arid1b in excitatory and inhibitory neurons and their progenitors in 8-10-week-old mice. Anxiety is a frequent comorbid condition with ASD in humans [40][41][42] , therefore, we first examined anxiety-like behaviors in Emx-Arid1b mice by performing the elevated plus maze and open field assays. Emx-Arid1b mice spent the same amount of time in the open arms of the elevated plus maze (13.24 s) as controls (10.16 s) (Fig. 7A). In the open field test, Emx-Arid1b mice entered, and spent the same amount of time in, the center portion of the open field apparatus (62.52 s), compared with wild type littermates (63.69 s) (Fig. 7B,C). From these results we concluded that Emx-Arid1b mice display no appreciable anxiety-like behaviors. We next assessed depression-like behaviors using the tail suspension and forced swim tests. Emx-Arid1b mice did not spend a discernably different amount of time immobile in either of these assays, compared with control mice (Fig. 7D,E). These results show that homozygous deletion of Arid1b in cortical neural progenitors does not appear to engender emotional symptoms such as anxiety-like and depression-like behaviors in mice.
We also investigated the effects of homozygous Arid1b deletion in cortical neural progenitors on cognitive performance in the novel object recognition task. Arid1b +/+ ;Emx1-Cre (Emx-WT) littermates displayed a recognition index of 70.88%, indicative of a stronger preference for interaction with the novel object (Fig. 7F). Emx-Arid1b mice, on the other hand, exhibited a recognition index of just 48.05% in this task, which indicates no real preference for a novel object versus a familiar one in an otherwise empty open field. Thus, complete Arid1b deletion in cortical progenitors leads to cognitive impairments in mice. Additionally, the measurement of grooming time showed no significant difference between Emx-Arid1b (29.40 s) and Emx-WT mice (19.98 s), suggesting that stereotypic behavior is not associated with Arid1b deletion in cortical progenitors (Fig. 7G). Furthermore, Emx-Arid1b demonstrated social novelty deficits compared with their Emx-WT littermates (Fig. 7H

Conditional knockout of Arid1b in ventral neural progenitors results in emotional and social impairments.
We also examined ASD-and ID-related behaviors in mice in which Arid1b is deleted in ventral neural progenitors. In the elevated plus maze test, Dlx-Arid1b mice spent significantly less time in the open arm (6.529 s) than their wild type littermates (15.440 s) (Fig. 8A). In the open field assay, Dlx-Arid1b mice entered and spent significantly less time in the center area of the apparatus (9.347 s), compared to controls (33.88 s) (Fig. 8B,C), which is indicative of anxiety-like behavior. Dlx-Arid1b mice also spent significantly more time immobile in the tail suspension assay (150.10 s), compared with control littermates (88.49 s) (Fig. 8D). The forced swim test also showed a similar trend of immobility in Dlx-Arid1b mice, but did not reach statistical significance (Fig. 8E). These results suggest that deletion of Arid1b in ventral neural progenitors may be sufficient to cause anxiety-like and depression-like behaviors in mice. In the novel object recognition test, Dlx-Arid1b mice also exhibited a significant difference in recognition index (49.75%), compared with controls (63.87%) (Fig. 8F). This suggests that, like Emx-Arid1b mice, Dlx-Arid1b mice exhibit an appreciable deficit in recognition memory, though, ostensibly, not to the same degree. Dlx-Arid1b mice exhibit a strong tendency toward repetitive behaviors, as evidenced by them spending more than 2.5 times as much time grooming themselves during a ten-minute window (53.61 s) than their wild type littermates (21.23 s) (Fig. 8G). In the sociability test using the three-chamber paradigm, control mice spent more time in the chamber containing a stranger mouse than in the empty chamber (p < 0.05). Dlx-Arid1b mice, however, showed no preference for either the empty or stranger chamber (Fig. 8H). In the social novelty test, Dlx-Arid1b mice did not display a significant difference in their time spent between chambers containing familiar and stranger mice, while control mice exhibited a clear preference for stranger (p < 0.05). Together with the grooming test results, these data show that homozygous deletion of Arid1b in the ventral neural progenitors contributes to social and stereotypic behaviors, key hallmarks of ASD-like behavior in mice.

Discussion
As we and other groups have recently reported, deletion of one Arid1b allele in mice is sufficient to cause significant behavioral deficiencies, recapitulating ID and ASD behavioral phenotypes [26][27][28] . These mouse models of Arid1b haploinsufficiency are useful because they mirror the gene dosage effects seen in human patients, but are less useful for deciphering the neurobiological function of the ARID1B protein, whose levels remain above fifty percent in Arid1b mutant heterozygotes, compared to their wild-type littermates [26][27][28] . Our group showed that adult heterozygous Arid1b knockout mice display a significant reduction in the overall number of their cortical interneurons, with no major changes in cortical pyramidal neurons 26 . It is unclear, however, what causes the decrease in cortical inhibitory neurons in the Arid1b haploinsufficient condition. We surmised that this is most likely due to defective proliferation and/or impaired cell survival in the ventral neural progenitor population.
Several BAF subunits have been shown to influence cell cycle progression and cell division [43][44][45][46][47][48] . In this study, we show that conditional homozygous knockout of Arid1b leads to an overall decrease in the number of actively proliferating and/or mitotic cells, as well as fewer newborn neurons or neural progenitors in the MGE. We also report that loss of Arid1b in ventral neural progenitors leads to increased apoptosis in the MGE. The defects in proliferation are due, at least in part, to improper cell cycle regulation, as we observe reduced cell cycle speed and a lower rate of cell cycle re-entry in Arid1b-deleted ventral neural progenitors. Somewhat unexpectedly, homozygous deletion of Arid1b in cortical neural progenitors that produce cortical pyramidal neurons also leads to significant defects in cell proliferation and survival, although these impairments are not as severe as what we www.nature.com/scientificreports/ observe with Arid1b deletion in the ventral telencephalon. It appears that Arid1b deletion in cortical progenitors leads to pre-mitotic cell cycle arrest, as the number of progenitors apparently active in the cell cycle is not altered. These findings suggest that Arid1b plays a more substantial role in regulating proliferation of ventral neural progenitors than it does in cortical neural progenitors. Homozygous deletion of Arid1b leads to similar increases in apoptosis in both cortical and ventral neural progenitor populations. Thus, the role of ARID1B in cell survival is likely similar in both progenitor subgroups and potentially elsewhere in the brain or periphery. In the current study we examine the effects of ARID1B during a relatively brief developmental window (E14.5-E15.5), but during this time frame many changes take place. Although we did not observe any striking differences in the metrics we used to evaluate proliferation and survival between E14.5 and E15.5 brains, these are distinct www.nature.com/scientificreports/ developmental phases, and a more careful examination of these timepoints, individually, may be warranted. Overall, it remains to be determined whether ARID1B's functions in regulating cell proliferation or survival is likely to play a larger role in pathology of ARID1B-related neurodevelopmental disorders. We show that homozygous deletion of Arid1b leads to reduced nuclear localization of β-catenin in ventral neural progenitors. Brahma-related gene-1 (BRG1), encoded by SMARCA4, is the core subunit of the mammalian BAF chromatin remodeling complex 49,50 and interacts directly with ARID1B in the nucleus 51,52 . BRG1 has also been shown to physically interact with nuclear β-catenin at target gene promoters and facilitate transcriptional activation, most likely via chromatin remodeling 53 . A recent report confirms that ARID1B regulates Wnt/β-catenin signaling in HEK293T cells in a BRG1-dependent manner 54 . Vasileiou et al. also reported that expression of β-catenin target genes is increased in peripheral blood lymphocytes taken from human patients with ARID1B loss-of-function mutations 37 . Furthermore, ARID1B interacts with β-catenin in a human osteosarcoma cell line, and the ARID1B-β-catenin interaction in these cells is mediated by BRG1 37 . This suggests a negative role for ARID1B and BRG1 in regulating β-catenin activity. Yet another group's findings support this conclusion, as they demonstrated that shRNA-mediated knockdown of Arid1b leads to increased expression of β-catenin target genes in Stat3 knockout, sciatic-nerve-derived neurofibroma spheres 38 . One further group did report a contradictory finding, however, showing that BRG1 enhances β-catenin-mediated gene expression by binding to β-catenin at target gene promoters and facilitating chromatin remodeling 53 . Our recent study involving global heterozygous knockout of Arid1b agrees more with this finding, as we see significant decreases in the mRNA expression levels of several β-catenin target genes in the ventral telencephalon 26 . Multiple studies have shown broad diversity in the configuration of BAF complexes, depending on cell types 43,46,48,[55][56][57][58][59][60][61] . Different switches in subunit composition play a large role in determining which genes are targeted by BAF complexes and how they mediate transcription. For this reason, the effects of Arid1b knockdown in non-neural progenitor or non-neuronal populations may be entirely different than what is observed in the brain. It is entirely possible that ARID1B acts to repress β-catenin in a BRG1-dependent manner in peripheral blood lymphocytes and tumor cell lines and enhance β-catenin transcriptional activation in the developing brain. This is important because proper control of β-catenin signaling is necessary for the proliferation of neural progenitors in the ventral telencephalon 62 . Therefore, the reduction in ventral progenitor proliferation in Dlx-Arid1b mice may be, at least in part, due to dysregulation of β-catenin signaling.
In addition to regulating Wnt/β-catenin signaling, we previously reported that Arid1b haploinsufficiency leads to disruptions in the epigenetic regulation of the Pvalb gene, which encodes PV 26 . This may explain the reduction in cortical PV-positive interneurons we observed here in Dlx-Arid1b mice, and in the aforementioned study when we performed heterozygous deletion of Arid1b in ventral progenitors 26 . As a member of the BAF chromatin remodeling complex, it should come as no surprise that deletion of Arid1b would have a far-reaching impact on the expression of many genes. Indeed, two groups performed RNA-sequencing on brains taken from Arid1b haploinsufficient mice and reported differential regulation of a large number of genes, including many ASD-related genes 27,28 . Therefore, the underlying causes of ARID1B-related behavioral phenotypes are likely varied and overlapping.
Arid1b haploinsufficient mice demonstrate marked ID-and ASD-like characteristics [26][27][28] , making them a useful mouse model for studying neurodevelopmental disorders. In addition to social and cognitive deficits, Arid1b haploinsufficient mice spend considerably more time grooming than wild type mice, which is considered an appropriate measure of repetitive or stereotyped behaviors in mouse models of ASD 26,27 . Arid1b haploinsufficient mice also display significant depression-like behaviors 26 , further replicating human symptoms as depression is a common comorbid condition in individuals with ID and ASD 40 .
With the apparent cortical excitatory/inhibitory imbalance we previously observed in Arid1b haploinsufficient mice 26 , we were interested to determine whether ARID1B depletion in cortical or ventral neural progenitors and their progeny contributes to specific behavioral phenotypes individually. We report that homozygous deletion of Arid1b in cortical neural progenitors has little to no notable effect on the emotional or social behavior, compared with controls. It does, however, lead to a significant decrease in cognitive function, as evidenced by impaired recognition memory in Emx-Arid1b mice (Fig. 6). Conversely, Arid1b deletion in ventral neural progenitors leads to considerable social and emotional deficits. We have previously shown that multiple ASD-and ID-like behaviors can be rescued in Arid1b haploinsufficient mice by increasing GABA tone using the GABA positive allosteric modulator, clonazepam. This suggests the critical status of ventral neural progenitors and their inhibitory neuronal progeny in Arid1b pathology. The current study supports this idea that ARID1B plays an important role in neural behaviors, especially emotional and social activities, by regulating ventral neural progenitors. Future studies will need to examine specific neuronal subpopulations and circuits related to particular behaviors in Arid1b mutant mice to provide a better understanding of the mechanistic link between ARID1B and behavior.
Taken together, we conclude that ARID1B plays distinct, yet overlapping roles in cortical and ventral neural progenitors. As multiple BAF subunits have been linked to the regulation of cell proliferation and survival 43 , the function of BAF complexes can be both cell-type and subunit-composition dependent 37,[43][44][45][46][47][48]50,52,56,59,63 . It is likely that ARID1B indeed fulfills different roles in divergent cell populations at different times. Our study reiterates that mutations to a single gene can have a broad impact on multiple cell types and regulate diverse functions.
Stereology. For quantifying numbers of cells, images of 10 different 50 µm brain sections were taken at periodic distances along the rostrocaudal axis with a Zeiss LSM710 confocal microscope as described previously 26,35,65 . N values for each experiment are described in figure legends. Where indicated, only cells within certain regions of interest (2.5 mm 2 white boxes within the VZ of the dorsal telencephalon or MGE) were analyzed in each section. Mouse cultured neurons were also assessed with this microscope. Cell numbers are described in figure legends. Images were analyzed using ZEN (Zeiss) and ImageJ (NIH). The calculated values were averaged, and some results were given as a percentage of DAPI labeled nuclei in the analyzed region.
BrdU administration and cell cycle analysis. For proliferation assays, BrdU administration and analysis of cell cycle speed and re-entry were performed as described previously 35,36 . Intraperitoneal injection of BrdU (20 mg per kg body weight, dissolved in 0.9% saline) into pregnant mice at E14.5 was performed 30 min prior to sacrifice. Sections from these mice were also used for quantifying overall Ki67 and BrdU labeling in Figs. 1 and 2.
For the analysis of cell cycle re-entry, control and mutant mice were exposed to BrdU for 24 h. Brain slices were then immunostained with antibodies to BrdU and Ki67. These sections were also used to evaluate Ki67 expression in Figs. 1 and 2, but not for evaluating BrdU. The ratio of cells labeled with BrdU and Ki67 to total cells that incorporated BrdU was determined. For the analysis of cell cycle length, the ratio of progenitor cells positive for Ki67 and BrdU to the total Ki67 labeled cells was assessed after a 1 h BrdU pulse.
Colocalization. Object-corrected fluorescence colocalization was performed as described by Moser et al. 39 .
Briefly, 10 sequential, individual Z stack images, spaced approximately 4 μm apart, were captured using a Zeiss LSM710 confocal microscope and analyzed for colocalization between DAPI and β-catenin using ImageJ (NIH) software with the macro described in the above reference 39 . The mean corrected Pearson's coefficient for each sample was recorded and the mean of all samples for each genotype is reported in the results.
Behavioral assays. All behavioral assays were performed as described previously 26,67,68 . Behaviors were assessed during the light cycle. Weight, activity, and feeding as health measures were considered before experiments. 3-to 4-month-old male and female mice were used for most behavioral assays. For social behavior assays, we used only male mice as sexual interactions between males and females and estrous cycle timing may interfere with accurate interpretation of social behavior. All behavioral assays were done blind to genotypes, with agematched littermates of mice.
Novel-object recognition test. The novel-object recognition test was performed as previously described 26,68 . A test mouse was first habituated to an open field arena (35 × 42 cm) for 5 min. Following habituation, the test mouse was removed from the arena and two identical objects with size (10.5 × 4.5 × 2.5 cm) were placed in opposite corners of the arena, 7 cm from the side walls. The test mouse was then reintroduced into the center of the arena and allowed to explore the field, including the two novel objects, for 10 min. After 6 h, one object was replaced with another novel object, which was of similar size but different shape and color than the previous object. The same test mouse was placed in the arena to explore the arena and the two objects. The movement of the mouse was recorded by a camera for 10 min and further analyzed by EthoVision XT 7 video tracking software (Noldus).
Three-chamber test for social interaction and novelty behavior. Social  www.nature.com/scientificreports/ wire enclosure in chamber 1. Then the openings to the flanking two chambers (1 and 3) were opened and the test mouse was allowed to explore the entire apparatus for 10 min. For the social novelty test, the stranger I mouse was randomly placed in one of the enclosures, while the test mouse had the choice of whether to investigate the stranger I mouse or a new novel mouse, designated 'stranger II' . This second novel mouse was taken from a different home cage and placed into the remaining empty wire enclosure. Time spent sniffing each partner by the test mouse was recorded for 10 min in both sociability and social novelty behavior tests. All apparatus chambers were cleaned with water and dried between trials. At the end of each test day, the apparatus was sprayed with 70% ethanol and wiped clean.
Grooming. Grooming was assessed as previously described 26,68 . A mouse was placed in a clear plastic cage (17 × 32 × 14 cm). The mouse was allowed to freely explore the cage for the entirety of the test. The first 10 min served as a habituation period. The movement of the mouse was recorded by a camera for 30 min. Recorded grooming behaviors included head washing, body grooming, genital/tail grooming, and paw and leg licking.
Elevated plus maze test. The elevated plus maze test was performed as previously described 26 Open field test. Open field activity was examined as previously described 26,68  Forced swimming test. Forced swimming test was performed as previously described 26,68 . A mouse was placed individually into a glass cylinder (20 cm height, 17 cm diameter) filled with water to a depth of 10 cm at 25 °C. After 5 min, the animals were removed from the water, dried, and returned to their home cages. They were again placed in the cylinder 24 h later, and after the initial 1 min acclimatization period, the total duration of immobility was measured for 5 min. Motionless floating was considered immobile behavior.
Tail suspension test. Tail suspension behavior was evaluated as described previously 26,68 . A mouse was suspended from the hook of a tail suspension test box, 60 cm above the surface of a table, using adhesive tape placed 1 cm from the tip of the tail. After 1 min acclimatization, immobility duration was recorded by a camera for 5 min. Mice were considered immobile only when they hung passively and were completely motionless.
Statistical analysis. Statistical analysis was performed as described previously 26,68 . Normal distribution was tested using the Kolmogorov-Smirnov test and variance was compared. Unless otherwise stated, statistical significance was determined using two-tailed, unpaired Student's t tests for two-population comparisons or one-way ANOVA followed by Bonferroni's post hoc test for multiple comparisons. Data were analyzed using GraphPad Prism and presented as means ± S.E.M. P values for each comparison are described in the results or figure legends. To determine and confirm sample sizes (n), we performed power analysis and/or surveyed the literature. Each experiment in this study was performed blind and randomized. Animals were assigned randomly to the various experimental groups, and data were collected and processed randomly. The allocation, treatment, and handling of animals was the same across study groups. Control animals were selected from the same litter as the test group. The individuals conducting the experiments were blinded to group allocation and allocation sequence. Exclusion criteria for mice were based on abnormal health conditions, including weights below 15 g at 6 weeks of age and noticeably reduced activity or feeding. Statistical data and n values for all behavioral assays are described in the figure legends. This study was carried out in compliance with the ARRIVE guidelines (http://www.nc3rs .org.uk/page. asp?id=1357). License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.