Loss of BAF (mSWI/SNF) chromatin-remodeling ATPase Brg1 causes multiple malformations of cortical development in mice

Abstract

Mutations in genes encoding subunits of the BAF (BRG1/BRM-associated factor) complex cause various neurodevelopmental diseases. However, the underlying pathophysiology remains largely unknown. Here, we analyzed the function of Brahma-related gene 1 (Brg1), a core ATPase of BAF complexes, in the developing cerebral cortex. Loss of Brg1 causes several morphological defects resembling human malformations of cortical developments (MCDs), including microcephaly, cortical dysplasia, cobblestone lissencephaly and periventricular heterotopia. We demonstrated that neural progenitor cell renewal, neuronal differentiation, neuronal migration, apoptotic cell death, pial basement membrane and apical junctional complexes, which are associated with MCD formation, were impaired after Brg1 deletion. Furthermore, transcriptome profiling indicated that a large number of genes were deregulated. The deregulated genes were closely related to MCD formation, and most of these genes were bound by Brg1. Cumulatively, our study indicates an essential role of Brg1 in cortical development and provides a new possible pathogenesis underlying Brg1-based BAF complex-related neurodevelopmental disorders.

Introduction

The mammalian cerebral cortex is a complex structure that contains different classes of neuronal and glial subtypes organized radially in six layers and tangentially in several functional domains. Inside-out complex cortical layer-specific neurons are produced in a specific temporal order by neural progenitor cells (NPCs) through self-renewal and differentiation (1). NPCs comprise apical progenitors (APs) located in the ventricular zone (VZ) and basal progenitors (BPs) located in the subventricular zone. The NPCs undergo series key events, such as the proliferation and differentiation of neural progenitors, specification of neuronal subtypes, programmed cell death, as well as the migration and maturation of neurons to functionally integrate into various cortical circuits in the brain (2). These developmental processes of the cerebral cortex occur in discrete steps that are subtly regulated and orchestrated to ensure the normal histogenesis and function of the cortex. Malformations of cortical developments (MCDs) are characterized by abnormal cortical structures or atypical brain sizes (3). MCDs are the most common problems in the cortex because of abnormalities in normal neurogenesis, and usually appear to cause cognitive defects, intellectual disability and other neurodevelopmental diseases (4,5).

Chromatin remodelers influence gene expression via chromatin structure modulation to alter the accessibility of specific DNA regions by transcriptional machinery and other DNA-binding molecules. The mammalian SWItch/sucrose nonfermenting (mSWI/SNF), also known as the BAF (BRG1/BRM-associated factor) complex, is a class of adenosine triphosphate (ATP)-dependent chromatin remodelers that contains at least 15 different subunits, including the interchangeable core ATPase Brg1 (Smarca4) or Brm (Brahma homolog, Smarca2), and utilizes energy from ATP hydrolysis to alter nucleosomal units in the chromatin structure (6). Mutations in the BAF complex are detected in nearly all types of cancer (7). Whole-exome sequencing and genome-wide association studies have revealed that mutations in BAF complex subunits are widespread in various syndromic and non-syndromic cognitive dysfunction phenotypes (8,9). Emerging studies suggest that BAF complexes play central roles in the epigenetic regulatory mechanisms that impact neural developmental processes (8,10). As the enzymatic subunit of BAF complexes, Brg1 is critical for the function of the BAF complex. BRG1 mutations are associated with many cancer types and are related to developmental disorders (11,12). Recent human genome analyses have revealed that BRG1 mutations are highly associated with neurodevelopmental diseases such as Coffin-Siris syndrome (CSS, MIM 135900) and autism spectrum disorders (13–17). CSS is a rare congenital disorder characterized by growth deficiency, intellectual disability, coarse facial appearance, feeding difficulties in infancy and hypoplasia of the fifth distal phalanges, fingernails and toenails (18). In addition to BRG1, several genes encoding the BAF complex subunits have been identified to be causative, including AT-rich interaction domain 1A (ARID1A), AT-rich interaction domain 1B (ARID1B), AT rich interactive domain 2 (ARID2), Double PHD fingers 2 (DPF2), SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily b, member 1 (SMARCB1), SWI/SNF related, matrix associated, actin dependent regulator of chromatin subfamily c member 2 (SMARCC2) and SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily e, member 1 (SMARCE1) (14,19,20). Other than BAF complex genes, PHD finger protein 6 (PHF6), SRY-box transcription factor 11 (SOX11) and SRY-box transcription factor 4 (SOX4) are additional causative genes of CSS (21–23). CSS patients with BRG1 mutations usually exhibit abnormal brain structures (16,17). Collectively, these data raise an interesting question about the role of BRG1 in cerebral development. Indeed, several studies have investigated the role of Brg1 in cerebral development based on different conditional knockout mouse models (24–27). However, the function of Brg1 in cerebral development remains largely unknown, and the underlying mechanism of neurodevelopmental diseases caused by abnormal BAF complexes remains unelucidated.

Loss of Brg1 results in early embryonic lethality in mice (28). To address the function of Brg1 in cortical development, we induced the cerebrum-specific deletion of Brg1 by crossing Brg1^flox/flox mice with Emx1-Cre and Bhlhb5-Cre mice. The Emx1-Brg1^f/f mice, in which Brg1 was deleted during the developmental stage of early cortical neurogenesis, exhibited several morphological defects resembling human MCDs, including microcephaly, cortical dysplasia, cobblestone lissencephaly and periventricular heterotopia. The hippocampus was completely absent in Emx1-Brg1^f/f mice. We further explored the mechanism of the multiple MCDs formation upon Brg1 deletion on the cellular and molecular levels. The Bhlhb5-Brg1^f/f cerebral cortex, in which Brg1 was deleted during the postnatal developmental stage, was slightly smaller than control, without other obvious MCDs. Overall, our study indicates an essential role of Brg1 in cortical development and provides a possible pathogenesis of intellectual disability and other neurodevelopmental disorders caused by mutant Brg1-based SWI/SNF complexes.

Results

Expression and conditional knockout of Brg1 in the developing cerebral cortex

First, we examined the expression of Brg1 in the developing cerebral cortex. Immunostaining using an anti-Brg1 antibody indicated that Brg1 is abundantly expressed in almost all cell types in the developing cerebral cortex at the embryonic stage (Fig. 1A; Supplementary Material, Fig. S1A). In the postnatal cerebral cortex, Brg1 immunostaining signal was decreased in most cell types as development progressed (Supplementary Material, Fig. S1B). Notably, Brg1 immunoreactivity was maintained at a relatively high intensity in the cornu ammonis field of the hippocampus in the mature brain (Supplementary Material, Fig. S1B). Western blot analysis confirmed the decreased expression of Brg1 along with postnatal cortical development (Fig. 1B). The expression data suggest that Brg1 may play an important role in early cortical development.

Figure 1

Brg1 expression and specific Brg1 deletion in the developing cerebral cortex. (A) Coronal sections of E12.5, E13.5 and E14.5 control and Emx1-Brg1^f/f forebrain stained for Brg1. Sections were counterstained with DAPI to label the nuclei. The right panel represents higher magnification views of the boxed areas. Arrows indicate the cortical hem still expresses Brg1. (B) Western Blot analysis of Brg1 expression in the developing cerebral cortex. (C) Western Blot analysis of Brg1 expression in E14.5 control and Emx1-Brg1^f/f cerebral cortex. (D) Quantification of Brg1 protein level after normalization to β-Actin. n = 3. ^**P < 0.01. Cx, cortex; GE, ganglionic eminence. Scale bars: 200 μm.

To determine the function of Brg1 in early cerebral development, we generated cortex-specific knockout mice by crossing Emx1-Cre mouse line expressing the Cre recombinase specifically in the developing cortex as early as embryonic day 10.5 (E10.5) (29,30) with mice containing exon2 and exon3 of the Brg1 gene flanked by loxP sites (31). The Emx1-Brg1^f/+ mice survived and reproduced normally and showed no obvious abnormalities in their gross morphology compared to wild-type mice. Most Emx1-Brg1^f/f mice died during the perinatal period, but a small subset survived up to approximately two weeks. The surviving Emx1-Brg1^f/f mice were significantly smaller than those of the control mice (Fig. 2A). We then examined whether Brg1 was efficiently and specifically deleted in Emx1-Brg1^f/f cortex. Immunostaining images revealed that Brg1 was normally expressed in Emx1-Brg1^f/f cerebral cortex at E10.5 (Supplementary Material, Fig. S1A). By E12.5, Brg1 immunoreactivity was dramatically downregulated and exhibited a faint immunostaining signal in the majority of cells in Emx1-Brg1^f/f cerebral cortex, except in the cortical hem wherein Brg1 was normally expressed at this developmental stage (Fig. 1A), suggesting Emx1-Cre mediated recombination occurred and Brg1 protein was degraded. By E13.5, Brg1 immunoreactivity was not detected in the majority of cells in Emx1-Brg1^f/f cerebral cortex, except in the cortical hem (Fig. 1A). Regarding the cortical hem, Brg1 was still expressed, but downregulated at E13.5 (Fig. 1A). By E14.5, Brg1 immunoreactivity was lost in most cells of the cortical hem (Fig. 1A). A small number of cells remained positive for Brg1 in Emx1-Brg1^f/f cerebral cortex (Fig. 1A). These cells may be interneurons or microglia in which Emx1-Cre is not expressed (29,30). The deletion of Brg1 was further validated by Western blot analysis (Fig. 1C, D). Thus, Emx1-Cre mediated efficient deletion of Brg1 occurred between E12.5 and E13.5 in Emx1-Brg1^f/f cerebral cortex, except in the cortical hem where Brg1 deletion occurred about two days later.

Figure 2

Loss of Brg1 leads to microcephaly and complete absence of the hippocampus. (A) Gross morphology of the P14 control and Emx1-Brg1^f/f mice. Emx1-Brg1^f/f mice were much smaller than controls. (B) Brains of P3 control and Emx1-Brg1^f/f mice. The Emx1-Brg1^f/f brain was smaller and more transparent than the control. Arrows indicate separated cerebral hemispheres by a midline gap. (C) H&E-stained coronal sections of E12.5, E14.5, E16.5 and P1 control and Emx1-Brg1^f/f forebrain. (D) Coronal sections of E14.5, E16.5 and P3 control and Emx1-Brg1^f/f forebrain stained for Prox1. (E) Coronal sections of E14.5, E16.5 and P3 control and Emx1-Brg1^f/f forebrain stained for Neurod1. Sections were counterstained with DAPI to label the nuclei. Cx, cortex; GE, ganglionic eminence; CA, cornu ammonis; DG, dentate gyrus. Scale bars: 500 μm.

Loss of Brg1 leads to microcephaly and complete absence of the hippocampus

The appearance of Emx1-Brg1^f/f cerebrum was smaller and more transparent than control (Fig. 2B), and was caudally connected but rostrally separated in the midline (Fig. 2B), suggesting that deletion of Brg1 led to brain hypoplasia and microcephaly. Coronal hematoxylin and eosin (H&E)-stained brain sections revealed that Emx1-Brg1^f/f cerebral cortex was slightly thinner than control at E12.5 (Fig. 2C). By E14.5, and thereafter, Emx1-Brg1^f/f cerebral cortex was dramatically thinner than control, and the lateral ventricles were larger in Emx1-Brg1^f/f forebrain (Fig. 2C). Moreover, we did not observe the typical structures of the hippocampus in Emx1-Brg1^f/f mice at different developmental stages (Fig. 2C). To examine whether the morphological absence of the hippocampus in Brg1 mutants results from a failure to generate hippocampal cells or from the failure of hippocampal cells to develop into the hippocampus once they are generated, we stained the sections for Prospero homeobox 1 (Prox1) and Neuronal differentiation 1 (Neurod1), which are two specific molecular markers of the hippocampus (32,33). In the control cerebral cortex, Prox1 immunoreactivity was detected in the developing dentate gyrus (Fig. 2D), while Neurod1 immunoreactivity was detected in the developing cornu ammonis and the dentate gyrus (Fig. 2E). Neither Prox1 nor Neurod1 were detected where they were expected, which is in the hippocampus in Emx1-Brg1^f/f cerebral cortex at any stage tested (Fig. 2D, E), suggesting that the hippocampus was absent in Emx1-Brg1^f/f mice.

Loss of Brg1 impairs cortical progenitor self-renewal, neuronal differentiation and apoptotic cell death

Because the cerebral cortex was much thinner in Emx1-Brg1^f/f mice, we examined whether the loss of Brg1 influenced the maintenance of cortical progenitors. Immunostaining experiments revealed that the number of Paired box 6 (Pax6)+ or SRY-box transcription factor 2 (Sox2)+ APs was decreased, but the proportion was dramatically increased in Emx1-Brg1^f/f cerebral cortex at E14.5 and E16.5 (Fig. 3A–F). The proportion of T-box brain 2 (Tbr2) labeled BPs was dramatically decreased in Emx1-Brg1^f/f cerebral cortex (Fig. 3G, H). To test cell proliferation upon Brg1 deletion, we labeled M-phase cells by phospho-Histone H3 (PH3) immunoreactivity and S-phase cells via a 1 h 5-ethynyl-2'-deoxyuridine (EdU) pulse. The number of PH3 positive cells was decreased in Emx1-Brg1^f/f cerebral cortex at E12.5 and E13.5 (Fig. 3I, J). Similarly, a reduced proportion of EdU-positive cells was observed in Emx1-Brg1^f/f cerebral cortex at E12.5 and thereafter (Fig. 3K, L). To examine whether loss of Brg1 causes changes in cell-cycle withdrawal, we labeled progenitor cells with EdU at E12.5–E14.5, and labeled those progenitors still cycling 24 h later by immunostaining with Marker of proliferation Ki-67 (Ki67), as previously described (34). Surprisingly, this analysis revealed a significant reduction in the cell-cycle exit index in Emx1-Brg1^f/f cerebral cortex compared to the control (Fig. 3M, N). These results indicated that although Brg1 deletion inhibited the NPCs’ exit from the cell cycle, the proliferation activity of NPCs was reduced, and cortical progenitor self-renewal was impaired in Emx1-Brg1^f/f cerebral cortex.

Figure 3

Loss of Brg1 impairs cortical progenitor self-renewal. (A) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for AP marker Pax6. (B) Quantification of Pax6+ cell numbers in E14.5 and E16.5 control and Emx1-Brg1^f/f neocortex. n = 3. (C) Quantification of Pax6+ cell ratios in E14.5 and E16.5 control and Emx1-Brg1^f/f neocortex. n = 3. (D) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for the AP marker Sox2. (E) Quantification of Sox2+ cell numbers in E14.5 and E16.5 control and Emx1-Brg1^f/f neocortex. n = 3. (F) Quantification of Sox2+ cell ratios in E14.5 and E16.5 control and Emx1-Brg1^f/f neocortex. n = 3. (G) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for the BP marker Tbr2. (H) Quantification of Tbr2+ cell ratios in E14.5 and E16.5 control and Emx1-Brg1^f/f neocortex. n = 3. (I) Coronal sections of E12.5 and E13.5 control and Emx1-Brg1^f/f forebrain stained for PH3. (J) Quantification of PH3+ cell numbers in E12.5 and E13.5 control and Emx1-Brg1^f/f neocortex. n = 3. (K) EdU was injected into pregnant mice at E12.5, E13.5 and E16.5 for embryo retrieval 1 h later. Coronal sections of EdU-labeled E12.5, E13.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for EdU. (L) Quantification of EdU+ cell ratio in (K). n = 3. (M) EdU was injected into pregnant mice at E12.5 and E14.5 for embryo retrieval 24 h later. Coronal sections of EdU-labeled E13.5 and E15.5 control and Emx1-Brg1^f/f forebrain stained for EdU and Ki67. (N) Quantification of EdU + Ki67- to total EdU+ cell ratio in (M). n = 3. Sections were counterstained with DAPI to label the nuclei. The error bars indicate the SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, Student’s t-tests. Cx, cortex; GE, ganglionic eminence. Scale bars: 100 μm.

For neurogenesis, NPCs exit the cell cycle for neuronal differentiation, so we tested whether delayed cell-cycle exit influenced neuronal differentiation. We stained the sections for T-box brain 1 (Tbr1), a transcription factor expressed by earlyborn neurons soon after differentiation and strongly expressed by layer 6 neurons (35,36), as well as Tuj1, a marker of differentiated neurons. Corresponding to the inhibited cell-cycle exit, the Tbr1+ and Tuj1+ cells were dramatically diminished in Emx1-Brg1^f/f cerebral cortex (Fig. 4A, B), suggesting that neuronal differentiation was inhibited. The increased proportion of Pax6+ or Sox2+ APs in Emx1-Brg1^f/f cerebral cortex (Fig. 3A–F) was consistent with the delayed cell-cycle exit, and inhibited neuronal differentiation.

Figure 4

Loss of Brg1 leads to disrupted neuronal migration and cortical dysplasia. (A) Coronal sections of E14.5 control and Emx1-Brg1^f/f forebrain stained for Sox2, Tbr2, Tbr1 and Tuj1. (B) Coronal sections of E16.5 control and Emx1-Brg1^f/f forebrain stained for Sox2, Tbr2, Tbr1 and Tuj1. (C) EdU was injected into pregnant mice at E12.5 for embryo retrieval 1, 24 and 72 h later. Coronal sections of EdU-labeled E12.5, E13.5 and E15.5 control and Emx1-Brg1^f/f forebrain stained for EdU and Sox2. (D) Quantification of ratio of EdU+ cells in Sox2+ VZ to total EdU+ cells. The error bars indicate the SD. ^**P < 0.01, Student’s t-tests. n = 3. (E) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for Nestin. Sections were counterstained with DAPI to label the nuclei. Scale bars: 100 μm.

We also examined cell apoptosis after Brg1 deletion by the TdT-mediateddUTP-Xnickendlabeling (TUNEL) assay and cleaved-Caspase3 immunostaining. TUNEL assays and cleaved-Caspase3 immunostaining revealed that apoptosis increased in Emx1-Brg1^f/f cerebral cortex at both E14.5 and E16.5 (Supplementary Material, Fig. S2).

Taken together, these data suggest that inactivation of Brg1 impaired cortical progenitor self-renewal, neuronal differentiation and apoptotic cell death, and the microcephaly in Emx1-Brg1^f/f mice was caused by a combination of reduced cell proliferation and increased apoptotic cell death.

Loss of Brg1 leads to disrupted neuronal migration and cortical dysplasia

We observed no obvious lamination in Emx1-Brg1^f/f H&E-stained brain sections (Fig. 2C). Immunoreactivity further revealed disorganized distribution of Sox2+ APs, Tbr2+ BPs, Tbr1+ neurons and Tuj1+ neurons in Brg1-deficient brain sections (Fig. 4A, B), suggesting that deletion of Brg1 leads to cortical dysplasia. The abnormal lamination prompted us to test whether neuronal migration was affected in Emx1-Brg1^f/f cerebral cortex. To evaluate this directly, we injected EdU into pregnant mice at E12.5, and harvested embryos 1, 24 and 72 h later to trace cell migration. A 1-h EdU pulse revealed that almost all EdU-labeled newborn cells were located in Sox2 positive VZ in both Emx1-Brg1^f/f and control cerebral cortex (Fig. 4C). After 24 h, we observed a smaller subset of EdU-labeled cells migrating out of Sox2 positive VZ in Emx1-Brg1^f/f cerebral cortex (Fig. 4C). After 72 h, a majority of EdU-labeled cells were located in the cortical plate and intermediate zone in the control cerebral cortex, while in Emx1-Brg1^f/f cerebral cortex, a majority of EdU-labeled cells were observed in Sox2 positive VZ (Fig. 4C). Although located in Sox2 positive VZ, many EdU-labeled cells were Sox2 negative in Emx1-Brg1^f/f cerebral cortex, which was rarely observed in the control cerebral cortex (Supplementary Material, Fig. S3), suggesting that these cells were differentiated neurons that failed to migrate. These results indicate that neuronal migration was compromised in the Emx1-Brg1^f/f cerebral cortex.

Radial glial cells guide the migration and placement of newborn neurons in the developing cerebral cortex (37), and radial glial basal processes promote neural migration to form the laminar organization and cortical columns of neurons (38–41). To examine the characteristic radial morphology of radial glial cells, we stained for Nestin to reveal the radial glial processes. In the control neocortex, the apical and basal processes of RGCs spanned across the entire structure radially from the VZ to the pial surface (Fig. 4E). Conversely, the cell somata were scattered, and radial glial processes appeared shortened, no longer aligned and radially oriented in Emx1-Brg1^f/f neocortex (Fig. 4E). These results demonstrated that the radial glial scaffold was abnormal in the Brg1-deficient cerebral cortex, which likely contributed to disrupted neuronal migration.

Loss of Brg1 causes pial basement membrane integrity defects and ectopic deposition of neurons at the pial surface

Interestingly, we often observed ectopias protruding beyond the pial surface of the brain in the H&E-stained sections of the embryonic Emx1-Brg1^f/f cerebral cortex older than E14.5 (Fig. 5A). To determine the cellular composition of ectopias, we immunostained E14.5 and E16.5 cerebral sections with the neuronal markers Tuj1 and Tbr1. Immunostaining images revealed that ectopias exhibited strong Tuj1 and Tbr1 expression (Fig. 5B, C). Next, we immunostained cerebral sections with the neuronal marker NeuN at postnatal stages, when most of the neurons reached their correct final position. NeuN-positive neuron clusters that exceeded the normal cell sparse layer 1 were often observed in the postnatal Emx1-Brg1^f/f cerebral sections (Fig. 5D). These data indicate that the ectopias on the pial surface mainly comprise neurons, and this malformation resembles human cobblestone lissencephaly.

Figure 5

Loss of Brg1 causes pial basement membrane integrity defects and ectopic deposition of neurons at the pial surface. (A) H&E-stained coronal sections of E14.5 and P3 control and Emx1-Brg1^f/f forebrain. (B) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for Tuj1. (C) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for Tbr1. (D) Coronal sections of P3 control and Emx1-Brg1^f/f forebrain stained for NeuN. (E) Coronal sections of E14.5 control and Emx1-Brg1^f/f forebrain stained for Laminin. Sections were counterstained with DAPI to label the nuclei. Arrows indicate ectopias protruding beyond the pial surface. Arrowheads indicate ectopias in the ventricular surface labeled by neuronal markers. Scale bars: 100 μm.

Pial basement membrane integrity is critical for normal cortical development and acts as a barrier for neurons to cross the pial surface (42–44). We then examined basement membrane integrity by checking for Laminin immunoreactivity. Laminin labeling revealed discontinuous pial basement membranes in Emx1-Brg1^f/f cerebral cortex, notably in areas adjacent to ectopias (Fig. 5E). Collectively, these findings demonstrate that Emx1-Brg1^f/f mice have abnormal neuronal localization typical of cobblestone lissencephaly, likely caused by breaches in the pial basement membrane.

Loss of Brg1 causes abnormal apical junctional complexes and ectopic deposition of neurons in the ventricular margin

In addition to ectopias at the pial surface, we also observed protrusions along the lateral ventricles in the H&E-stained sections of embryonic Emx1-Brg1^f/f cerebral cortex (Fig. 6A). To identify cell formation and analyze the dynamics of protrusion formation, we first performed immunostaining with the neuronal markers Tuj1 and NPC marker Nestin. At E14.5, similar to the control, Nestin-positive NPCs were localized at the surface of the lateral ventricles in Emx1-Brg1^f/f cerebral cortex (Fig. 6B). Although the lamination was abnormal, protrusions were rarely observed in the ventricular margin at this developmental stage (Fig. 6B). By E16.5, when protrusions were easily observed at the ventricular surface in Emx1-Brg1^f/f cerebral cortex, the Nestin-positive NPCs layer was discontinuous (Fig. 6B). Most protrusions exhibited Nestin immunoreactivity, and small Tuj1+ neuronal nodules formed in some protrusions at the apical surface (Figs 5B and6B). Similarly, Tbr1+ neurons were also found in some protrusions at the apical surface (Fig. 5C). Notably, at the ventricular surface of Emx1-Brg1^f/f hippocampal primordium, most cells were Tuj1 positive, and weak Nestin-positive cells were scattered along the hippocampal primordium (Fig. 6B). In postnatal day 3 (P3) Emx1-Brg1^f/f ‘undifferentiated hippocampus,’ we also observed massive NeuN-positive neurons at the ventricular surface (Figs 5D and6C), similar to the E16.5 hippocampal primordium stained for Tuj1. These data indicate that APs protruded into the ventricular space between E14.5 and E16.5, followed by the ectopic position of differentiated neurons in the ventricular surface during cortical development, indicating that this malformation was a form of periventricular heterotopia.

Figure 6

Loss of Brg1 causes abnormal AJCs and ectopic deposition of neurons in the ventricular margin. (A) H&E-stained coronal sections of E14.5 and P3 control and Emx1-Brg1^f/f forebrain. (B) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for Tuj1 and Nestin. (C) Coronal sections of E14.5 and E16.5 control and Emx1-Brg1^f/f forebrain stained for NeuN. (D) Coronal sections of E14.5 control and Emx1-Brg1^f/f forebrain stained for ZO1, β-catenin, Par3 and aPKC. (E) Coronal sections of E16.5 control and Emx1-Brg1^f/f forebrain stained for ZO1, β-catenin, Par3 and aPKC. Asterisk indicates ectopia labeled by Tuj1. Arrows indicate ectopias in the ventricular margin in (A), (B) and (C), as well as disrupted sites of AJCs in (E). Arrowheads indicate the AJCs localize in the middle of the hippocampal primordium in (E). Brackets indicate the ventricular surface stained by neuronal markers in the hippocampal primordium in (B) and (C), and areas almost complete absent of AJCs in the hippocampal primordium in (E). Scale bars: 100 μm.

Apical junctional complexes (AJCs) are cell–cell adhesion structures. The NPCs are connected by AJCs to each other at the ventricle surface to form a strong network that maintains NPCs inside the cortex, thus preventing their emergence on the ventricular surface (45). To investigate the integrity of AJCs in Emx1-Brg1^f/f cerebral cortex, we analyzed several members of AJCs in the developing cerebral cortex. β-catenin, atypical protein kinase C (aPKC), Partitioning defective 3 homolog (Par3) and Zona occludens 1 (ZO1) immunostaining revealed that AJCs were largely intact in E14.5 Emx1-Brg1^f/f cerebral cortex (Fig. 6D). By E16.5, when obvious protrusions along the lateral ventricles appeared, focal disruption of AJCs was observed in Emx1-Brg1^f/f neocortex (Fig. 6E). Cortical cells protruded into the ventricular space from the ‘holes’ formed by disrupted AJCs (Fig. 6E). Corresponding to the more severe phenotype in the hippocampal primordium, AJCs were almost completely lost in this area of Emx1-Brg1^f/f cerebral cortex (Fig. 6E). We observed that some AJCs localized in the middle of the hippocampal primordium, which appeared to be completely engulfed by their nonpolarized neighbors and internalized into the cortex (Fig. 6E). Together, these results suggest that the deletion of Brg1 leads to defects in the apical surface integrity of the VZ that promotes the formation of periventricular heterotopia.

Loss of Brg1 led to impairment of gene expression programs tightly associated with cortical development

To gain better insight into the effect of the molecular mechanism of Brg1 inactivation on cortical development, we examined the gene expression profile of the Brg1 null cortex during the stages of cortical neurogenesis (E14.5) using RNA sequencing (RNA-seq). Given that Brg1 was completely deleted in most Emx1-Brg1^f/f cortical cells by E14.5, we chose to analyze E14.5 cortex to minimize the interference of residual Brg1 protein. With a Log2FC (fold change) value of 1 as the cutoff, this analysis revealed that Brg1 loss resulted in the differential expression of 1664 genes compared with controls, among which 435 genes were downregulated, whereas 1229 genes were upregulated (Fig. 7A, B). Genes involved in cerebral cortex development were conspicuously downregulated, including Forkhead box G1 (Foxg1), FEZ family zinc finger 2 (Fezf2), Neuronal differentiation 6 (Neurod6), Zinc finger and BTB domain containing 18 (Zbtb18), Potassium voltage-gated channel subfamily A member 1 (Kcna1), Nuclear receptor subfamily 4 group A member 3 (Nr4a3), Eomesodermin (Eomes), DAB adaptor protein 1 (Dab1), POU class 3 homeobox 2 (Pou3f2), POU class 3 homeobox 3 (Pou3f3), Neuropilin 1 (Nrp1), Neurod1, LDL receptor related protein 8 (Lrp8), Dopamine receptor D1 (Drd1), Neuregulin 1 (Nrg1), Empty spiracles homeobox 2 (Emx2), SATB homeobox 2 (Satb2), Brevican (Bcan) and Cell adhesion associated, oncogene regulated (Cdon) (Fig. 7C). This was further validated by real-time quantitative polymerase chain reaction (qRT-PCR) analysis (Supplementary Material, Fig. S4A). We then performed gene ontology (GO) term enrichment analysis (46,47), which could be classified into three categories: biological process, cellular component and molecular function, to gain further insight into the signaling pathways regulated by Brg1 in the developing cerebral cortex. GO biological process category showed that all the top 10 terms of downregulated genes were related to cortical development and neuronal development (Fig. 7D), while upregulated genes were mainly enriched in pathways related to renal system development, pattern specification process, ossification, cilium movement, epithelial tube morphogenesis and cell fate commitment (Fig. 7E). In the GO cellular component category, several of the top 10 terms of genes that were downregulated and upregulated upon Brg1 loss were related to neuronal morphogenesis, including axon part, main axon, dendritic shaft, axolemma, plasma membrane-bounded cell projection cytoplasm, axoneme part (Fig. 7D, E). We noticed that 4 of the top 10 GO cellular component terms of upregulated genes were associated with the extracellular matrix (Fig. 7E), which are important in the development of the cerebral cortex through non-cell-autonomous mechanisms (48). Consistently, extracellular matrix deregulation upon Brg1 deletion in the developing brain analyzed by RNA-seq was also reported in a recently published paper, in which Brg1 deletion was mediated by an inducible Sox2-CreER^T2 mouse line (49). Among the GO molecular function category, 8 of the top 10 downregulated terms were related to channel activity and transmembrane transporter activity (Fig. 7D), which participate in the formation of synaptic cell–cell communications. Five of the top 10 upregulated terms were related to interactions with the extracellular matrix, including glycosaminoglycan binding, heparin binding, extracellular matrix binding, sulfur compound binding and integrin binding (Fig. 7E), emphasizing the role of Brg1 in extracellular matrix regulation.

Figure 7

RNA-sequencing analysis reveals that the loss of Brg1 leads to impairment of gene expression programs tightly associated with cerebral cortex development. (A) Volcano plot of RNA-seq analysis illustrating the differentially expressed genes responding to depletion of Brg1 in E14.5 cerebral cortex. (B) Relative expression heatmap of 1664 significantly differentially expressed genes. (C) Heatmap of selected differentially expressed genes known to be important for cerebral cortex development. (D) GO term enrichment analysis for downregulated genes in E14.5 Emx1-Brg1^f/f cerebral cortex. (E) GO term enrichment analysis for upregulated genes in E14.5 Emx1-Brg1^f/f cerebral cortex. (F) KEGG pathway enrichment analysis for differentially expressed genes. (G) GSEA plots of GO terms involved in ‘neurogenesis,’ ‘cell proliferation,’ ‘neuron differentiation,’ ‘neuron migration,’ ‘basement membrance’ and ‘tight junction’ for RNA-seq expression data.

Next, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) to analyze the enrichment of biological pathways. The results largely matched the GO term enrichment analysis, because terms such as the calcium signaling pathway, axon guidance, cell adhesion molecules, signaling pathways regulating pluripotency of stem cells, extracellular matrix (ECM)-receptor interaction and neuroactive ligand-receptor interaction were enriched in the top 20 categories of differentially expressed genes (Fig. 7F; Supplementary Material, Fig. S4B, C). Moreover, upregulated genes fell into cancer-related categories, such as basal cell carcinoma, proteoglycans in cancer and hepatocellular carcinoma (Supplementary Material, Fig. S4B), suggesting the involvement of Brg1 in tumor inhibition.

Neuronal progenitor self-renewal, neuronal differentiation, neuronal migration, radial glial cell morphology, pial basement membranes and AJCs were abnormal in Emx1-Brg1^f/f cerebral cortex. Thus, we used the transcriptome data to analyze differentially expressed genes associated with these biological processes or cellular structures. As a result, many genes related to neuronal differentiation (Supplementary Material, Fig. S4D), neuronal migration (Supplementary Material, Fig. S4E) and basement membrane formation (Supplementary Material, Fig. S4F) were deregulated in Emx1-Brg1^f/f cerebral cortex. We also noticed that many genes participating in the regulation of the actin and microtubule cytoskeleton, which play important roles in neuronal morphology, were abnormal in Emx1-Brg1^f/f cerebral cortex (Supplementary Material, Fig. S4G, H). Similarly, gene set enrichment analysis (GSEA) suggested significant enrichment for gene expression profiling after Brg1 deletion among gene set neurogenesis, cell proliferation, neuron differentiation, neuron migration, basement membranes and tight junctions (Fig. 7G).

To identify genes likely to be direct targets of Brg1, we analyzed previously published Brg1-FLAG chromatin immunoprecipitation sequencing (ChIP-seq) data in the developing forebrain of a knock-in mouse strain, in which 3 × FLAG tags were inserted into Brg1 (50). ChIP-seq analysis identified 51 512 Brg1 peaks, and 57.55% of Brg1 peaks were localized to promoters (Supplementary Material, Fig. S5). ChIPseeker analysis identified 19 658 peak-related genes by assigning 51 512 peaks to retrieve the nearest genes around the peak. By comparing our RNA-seq results with the ChIP-seq data, we found that 1216 differentially expressed genes were targeted by Brg1 (Supplementary Material, Fig. S5B). Among them, 888 were upregulated, accounting for 72% of all upregulated genes, while 328 were downregulated, accounting for 75% of all downregulated genes. GO and KEGG enrichment analysis of these genes revealed that the enriched terms were closely associated with cortical development, which were similar to GO and KEGG enrichment analysis of all differentially expressed genes (Supplementary Material, Fig. S5C, D, E).

Taken together, the RNA-seq and ChIP-seq data demonstrated that Brg1 directly regulates the transcriptional activity of a large number of genes closely related to cortical development, suggesting that the multiple MCDs of Brg1-deficient cerebral cortex were caused by deregulated gene expression profiling upon Brg1 inactivation, but was not a secondary effect of general cortical degeneration.

Loss of Brg1 in the postnatal cerebral cortex cause slightly marked cortical morphological defects

Brg1 expression was significantly downregulated in postnatal cortical development; therefore, we wondered if deletion of Brg1 in the postnatal developmental stages affects cortical morphology. To delete Brg1 in the postnatal developing cerebral cortex, we crossed Brg1^flox/flox mice with Bhlhb5-Cre mouse line (51) to generate Bhlhb5-Brg1^f/f mice. Bhlhb5 is a transcription factor first expressed in the nascent cortical plate of the developing cerebral cortex (51). By mating the Rosa26R-tdTomato reporter mouse line (52), we observed that Bhlhb5-Cre mediated recombinants mainly occurred in the cerebral cortex (Supplementary Material, Fig. S6A). Immunostaining images revealed that Brg1 immunoreactivity was downregulated but not completely lost in some cells in Bhlhb5-Brg1^f/f cerebral cortex at P1 (Supplementary Material, Fig. S6B). By P3, Brg1 immunoreactivity was not detected in the majority of cells in Bhlhb5-Brg1^f/f cerebral cortex (Supplementary Material, Fig. S6B). These results demonstrated that Brg1 deletion in most Bhlhb5-Brg1^f/f cortical cells occurred between P1 and P3, suggesting that Bhlhb5-Brg1^f/f mice are suitable models for studying Brg1 function in the postnatal cerebral cortex. The appearance of Bhlhb5-Brg1^f/f brain was grossly normal (Supplementary Material, Fig. S7A). Coronal H&E-stained brain sections revealed that Bhlhb5-Brg1^f/f brain was slightly smaller than the control, but did not exhibit obvious malformations as observed in Emx1-Brg1^f/f cerebral cortex (Supplementary Material, Fig. S7B). By 3 months old, Bhlhb5-Brg1^f/f brains exhibited no significant changes compared to P30 Bhlhb5-Brg1^f/f brains, and a large number of Brg1 negative cortical cells were still exist (Supplementary Material, Fig. S6B), indicating that Brg1 is dispensable for cortical cell maintenance. The phenotypes of Emx1-Brg1^f/f and Bhlhb5-Brg1^f/f cerebral cortex suggest a temporally specific and key role of Brg1 in cortical development, particularly in the early stage.

Discussion

In this study, we investigated the role of Brg1 in the developing cerebral cortex using conditional gene knockout. We showed that Brg1 was highly expressed in the early developing cerebral cortex and downregulated during the postnatal developmental stage, suggesting a crucial role for Brg1 in the early development of the cerebral cortex. To bypass early embryonic lethality caused by conventional inactivation of Brg1, we specifically deleted Brg1 in the developing cerebral cortex using the Cre/LoxP strategy. The Emx1-Brg1^f/f mice, in which the cerebral cortex Brg1 deletion occurred between E12.5 and E13.5, presented with significantly smaller bodies and brains. In addition to microcephaly, we also observed several MCDs, including cortical dysplasia, cobblestone lissencephaly, and periventricular heterotopia in Emx1-Brg1^f/f mice. The hippocampus was not formed in Emx1-Brg1^f/f mice. Further analysis revealed that the important processes of cortical development, including cortical progenitor self-renewal, neuronal differentiation, neuronal migration and cell apoptosis were impaired in Emx1-Brg1^f/f cerebral cortex. We also observed that the pial basement membrane and AJCs were disrupted in Emx1-Brg1^f/f cerebral cortex. The combination of these abnormal biological processes and cellular components led to multiple MCDs in Emx1-Brg1^f/f mice. At the molecular level, RNA-seq analysis revealed that the gene expression program closely associated with these developmental processes was deregulated after Brg1 deletion. Furthermore, by combining our RNA-seq results with previously published ChIP-seq data (50), we demonstrated that most of the deregulated genes were direct targets of Brg1 in the developing forebrain. The brains of Bhlhb5-Brg1^f/f mice, in which the cerebral cortex Brg1 deletion occurred between P1 and P3, were slightly smaller than the control mice, and did not exhibit obvious MCDs as in Emx1-Brg1^f/f cerebral cortex. These observations highlight a key role of Brg1 in early cortical development and suggest that inactivation of Brg1 can cause multiple MCDs in the cerebral cortex, in addition to microcephaly.

In this study, deletion of Brg1 mediated by Emx1-Cre led to microcephaly, which was caused by reduced proliferation and increased apoptotic cell death. This phenotype is similar to that of Nestin-Brg1^f/f (24,27) and hGFAP-Brg1^f/f (25) mice, in which deletion of Brg1 is induced in NPCs by Nestin-Cre and hGFAP-Cre. Expression of Cre recombinase was earlier in Emx1-Cre mice than in Nestin-Cre and hGFAP-Cre mice (29,53–55), suggesting that Brg1 plays an important role in different stages of early cortical neurogenesis. In Emx1-Brg1^f/f mice, Prox1 and Neurod1 immunostaining confirmed the absence of the hippocampus. While in hGFAP-Brg1^f/f mice, the hippocampus was severely hypoplastic, most hippocampal cell types have been examined by immunohistochemistry (25). This difference indicates that hippocampal development is more sensitive to stage-specific Brg1 inactivation.

Beyond microcephaly, the most striking MCDs in Emx1-Brg1^f/f cerebral cortex were cortical dysplasia, cobblestone lissencephaly and periventricular heterotopia, which have not been reported in Brg1-deficient cerebral cortices. We demonstrated that neuronal migration was disrupted after Brg1 deletion, leading to cortical dysplasia. Nestin staining revealed abnormal radial glial processes in Emx1-Brg1^f/f cerebral cortex, which is bound to cause aberrant cell migration. RNA-seq revealed that many genes related to cell migration were deregulated after Brg1 deletion. However, we could not confirm whether cell-autonomous defects in migrating Brg1-deficient neurons contribute to migration defects. Cobblestone lissencephaly is a migration disorder caused by the overmigration of neurons through gaps in the pial basement membrane (3). In Emx1-Brg1^f/f cerebral cortex, we observed that the pial basement membrane was disrupted and neuronal ectopia protruded in the broken site, which subsequently caused cobblestone lissencephaly. Periventricular heterotopia has traditionally been regarded as a disorder of abnormal migration, but recent data have argued that adherens junctions and actin cytoskeletal dynamics along the ventricular lining may be the primary pivotal factor for PH formation (56). Similar to the pial basement membrane, the AJCs were shown to be disrupted in Emx1-Brg1^f/f cerebral cortex, and that leakage of NPCs in the ventricular surface is the main cause of periventricular heterotopia. Corpus callosum agenesis is frequently observed in CSS patients with BRG1 mutations (57). Indeed, we did not observe corpus callosum in Emx1-Brg1^f/f mice. Given that the formation of corpus callosum relies on correct midline patterning, formation of telencephalic hemispheres, birth and specification of commissural neurons and axon guidance across the midline (58), we concluded that the absent of corpus callosum was likely to be a secondary effect of global cortical degeneration.

Studies of Brg1 in different tissues revealed that the function of Brg1 relies heavily on tissue/cell-type-specific contexts (12). Deletion of Brg1 in the cerebral cortex caused an abnormal radial glial scaffold, as well as disrupted the pial basement membrane and AJCs. RNA-seq analysis revealed that the expression of many genes associated with cell junctions and extracellular matrix was significantly altered in the cerebral cortex of Emx1-Brg1^f/f. These results indicate that Brg1 is tightly associated with the cytoskeleton and epithelial integrity in the cerebral cortex. In Brg1-deficient retinae, cell lamination and cell polarity were impaired, and the AJC components ZO1, aPKC-λ and γ-tubulin were disrupted (59). In the endometrial epithelium, deletion of Brg1 leads to loss of epithelial integrity and causes adenomyosis-like phenotypes, and transcriptomic analysis revealed that the expression of genes related to cell junctions, cell adhesion and actin cytoskeletal processes was abnormal (60). In a recent study, RNA-seq revealed that many cell adhesion molecules were downregulated after the deletion of Brg1 in the otic epithelium (61). In the auditory epithelium, our previous study revealed that deletion of Brg1 in hair cells results in abnormal epithelial integrity, cell arrangement and cell polarity, with the distribution of the apical proteins Gαi, LGN and aPKC disturbed (62). Taken together, these data suggest that Brg1 may share a universal role in regulating the cytoskeleton, cell polarity and epithelial integrity across different epithelial tissues.

BAF155 and BAF170 subunits have been reported to control the stability of BAF complexes. Double knockout of scaffolding subunits BAF155 and BAF170 in the cerebral cortex results in degeneration of the entire BAF complex mediated by the ubiquitin-proteasome system (63). The morphology of Emx1-Brg1^f/f cerebral cortex in this study was similar to that of the BAF155 and BAF170 double-knockout cerebral cortices mediated by the same Emx1-Cre mouse line. The reduced proliferation, increased apoptosis and differentiation defects were similar between the two mouse models (63). These results suggest that inactivation of Brg1 phenocopies the loss of BAF complexes in the cerebral cortex, and highlight the pivotal function of Brg1 in BAF complexes. However, the phenotypic defect of Emx1-Brg1^f/f cerebral cortex seems milder than BAF155 and BAF170 double-knockout cerebral cortices in terms of the morphological features and statistical results, such as the cerebral cortex thickness, proliferation defects and differentiation defects (63). These results suggest that deletion of Brg1 leads to significant, but not complete functional loss of BAF complexes, which may be because Brm partially compensates for the function of Brg1 or other BAF subunits introducing ATPase-independent functions of BAF complexes. Further studies are needed to understand the functional interactions between Brg1 and other BAF complex subunits during development.

Genes encoding BAF complex subunits are the most frequently mutated genes in intellectual disability/neurodevelopmental disorders, and mutations in these genes are found in several cognitive dysfunction phenotypes, such as CSS, schizophrenia, Nicolaides-Baraitser syndrome and autism spectrum disorders (8,9,64). MCDs are the most common cortical problems that lead to cognitive defects, intellectual disability and other neurodevelopmental diseases (4,5). The Emx1-Brg1^f/f mice exhibited multiple phenotypic defects that assembled human MCDs, including microcephaly, cortical dysplasia, cobblestone lissencephaly and periventricular heterotopia. Microcephaly is often observed in patients with abnormal BAF complex subunits, such as those with CSS and Nicolaides-Baraitser syndrome, and is frequently associated with other clinical features, such as intellectual impairment, developmental delay and epilepsy (65). The incidence of microcephaly appears to be higher in patients with BRG1 mutations than in patients with other BAF subunit mutations (57). However, to our knowledge, the other MCDs in Emx1-Brg1^f/f mice have rarely been reported in individuals with neurodevelopmental disorders caused by abnormal BAF complex subunits. In a recent study, genetic etiology analysis revealed that the gene encoding the BAF complex subunit SMARCB1 is a candidate for periventricular heterotopia (66). Moreover, cerebral neuron heterotopia was reported in two patients with CSS and abnormal BAF complex subunits examined by magnetic resonance imaging (MRI). One harbors ARID1A mutation and exhibits gray matter heterotopia, but the detailed position of heterotopia has not been described (67). The other harbors an ARID1B mutation and exhibits periventricular heterotopia (68). In another two CSS patients with unknown mutations, histological examination of autopsy indicated neuronal heterotopias in the cerebellum (69,70). In addition, neuronal lamination defects were observed in the cerebellum (70) and diencephalon (69) of the two patients, respectively. Cobblestone lissencephaly has never been reported in CSS or other patients caused by mutations in the BAF complex subunits. One possible reason why the MCDs other than microcephaly in Emx1-Brg1^f/f mice have rarely been reported in individuals with neurodevelopmental disorders caused by abnormal BAF complex subunits may be that they are more difficult to observe than microcephaly.

In summary, our study provides new possible pathological mechanisms for neurodevelopmental diseases caused by abnormal Brg1 and other BAF complex subunits. In addition to microcephaly, multiple MCDs may be associated with BAF-related neurodevelopmental diseases. It is necessary to analyze the brain structure of patients with abnormal BAF complex subunits using high-precision imaging methods such as MRI at the time of diagnosis.

Materials and Methods

Mice

All animal experimental procedures were approved by the Medical Ethics Committee of Cheeloo College of Medicine, Shandong University. Brg1^flox/flox (31), Emx1-Cre (29), Bhlhb5-Cre (51) and Rosa26R-tdTomato (52) mouse lines were maintained on a mixed genetic background and genotyped as described previously. Brg1^flox/flox females were mated with Emx1-Brg1^flox/+ and Bhlhb5-Brg1^flox/+ males to generate Emx1-Brg1^f/f and Bhlhb5-Brg1^f/f mice. For timed pregnancies, the day of vaginal plug detection was regarded as E0.5 and the day of birth as P0.

Histology and immunofluorescence assay

Brains at different stages of development were dissected and fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C overnight. For paraffin sectioning, the samples were dehydrated by an ascending series of ethanol, cleared in xylene and embedded with paraffin in the proper orientation. Paraffin blocks were sectioned serially to a 7 μm thickness. For frozen sectioning, the samples were infiltrated with sucrose series (15%, 30%) followed by embedded in optimal cutting temperature (OCT) compound and sectioned to an 8–10 μm thickness.

For histology assay, the sections were stained by H&E. For immunofluorescence assay, samples were blocked for 30 min with 10% donkey serum, followed by incubation with primary antibodies in PBS at 4°C overnight. After three washes with PBS, samples were incubated at 37°C for 1 h in secondary antibodies. 4',6-diamidino-2-phenylindole (DAPI) was applied to stain the nuclei. Primary antibodies used were anti-Tbr1 (rabbit, 1:200, Abcam, ab183032), anti-Tbr2 (rabbit, 1:200, Abcam, ab23345), anti-Pax6 (rabbit, 1:400, Proteintech, 12 323–1-AP), anti-Sox2 (rabbit, 1:200, Abcam, ab92496), anti-Tuj1 (rabbit, 1:400, Abcam, ab52623), anti-Par3 (rabbit, 1:200, Proteintech, 11 085–1-AP), anti-Prox1 (rabbit, 1:200, Proteintech, 11 067–2-AP), anti-Neurod1 (rabbit, 1:200, Proteintech, 12 081–1-AP), anti-NeuN (rabbit, 1:200, CST, 24307), anti-Nestin (mouse, 1:200, Abcam, ab11306), anti-Ki67 (rabbit, 1:400, Abcam, ab15580), anti-pH3 (rabbit, 1:800, Bioworlde, BS4094), anti-Brg1 (rabbit, 1:400, Abcam, ab110641), anti-aPKC (rabbit, 1:200, Santa Cruz, sc216), anti-cleaved-Caspase3 (rabbit, 1:400, CST, 9664), anti-ZO1 (rabbit, 1:200, Proteintech, 21 773–1-AP), anti-β-Catenin (rabbit, 1:400, Abcam, ab32572), anti-Laminin (rabbit, 1:200, Sigma-Aldrich, L9393). Images were acquired using an OLYMPUS BX51 fluorescence microscope or a Nikon TE2000 fluorescence microscope.

Western blot analysis

The brain samples were dissected and homogenized in RIPA lysis buffer (Beyotime Biotechnology, P0013B) with 0.2 mM PMSF. The proteins were extracted on ice. The proteins from the samples (20 μg) were subjected to 10% sodium dodecy1 sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a polyvinylidene difluoride (PVDF) membrane. The PVDF membrane was blocked for 1 h with 5% nonfat dry milk, followed by incubation with primary antibodies in Tris-buffered saline containing 0.1% Tween-20 (TBST) at 4°C overnight. The PVDF membrane was washed with TBST and then incubated with secondary antibodies at room temperature for 1 h. Primary antibodies used were anti-Brg1 (rabbit, 1:20 000, Abcam, ab110641) and anti-β-actin (rabbit, 1:40 000, Proteintech, 20 536–1-AP). After washing in TBST, the protein bands were detected using enhanced chemiluminescence (ECL) detection reagents (Millipore, P90719).

TUNEL assay

For the TUNEL assay, a One Step TUNEL Apoptosis Assay Kit (Beyotime Biotechnology, C1089) was used. The experimental procedure was performed according to the manufacturer’s instructions. Briefly, the sections were digested with proteinase K and incubated with the TUNEL reaction mixture at 37°C for 1 h. The nuclei were stained with DAPI, and apoptotic cells were labeled with cyanine 3 (Cy3).

DNA-labeling analysis with 5-ethynyl-2'-deoxyuridine in vivo

EdU (RiboBio) was injected intraperitoneally into pregnant mice at a dose of 25 mg per 1 kg body weight. The mice were sacrificed, and their brains were dissected at different time points after EdU injection according to the experimental goals. For S-phase analysis, mice were sacrificed 1 h after injection. For the cell-cycle exit analysis, the mice were sacrificed 24 h after the injection. For neuronal migration analysis, the post-injection waiting times were 1, 24 and 72 h. EdU signals were stained with the Cell-Light™ EdU Apollo® 488 in vitro Imaging Kit (RiboBio, C10310–3) according to the manufacturer’s instructions.

Real-time quantitative polymerase chain reaction

The cerebral cortex at E14.5 was dissected. Then, total RNA from the cerebral cortex was extracted using Trizol reagent (Invitrogen) following the manufacturer’s protocol, and cDNA was generated by using random primers. Real-time quantitative PCR was performed with SYBR Green Real-Time PCR Master Mix (CW0957, CWBIO, China) and the appropriate primers (Supplementary Material, Table S1). The 2^−ΔΔCt method (71) was employed to estimate relative expression level.

RNA-sequencing

The cerebral cortex at E14.5 was dissected and total RNA was extracted using Trizol reagent (Ambion, 15 596) following the manufacturer’s protocol. A total of 1 μg RNA per sample was used as the input material for the RNA sample preparations. RNA-seq was performed by Novogene (Beijing, China). Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was implemented using divalent cations under elevated temperatures in the First Strand Synthesis Reaction Buffer (5X). First-strand cDNA was synthesized using random hexamer primers and M-MuLV Reverse Transcriptase (RNase H). Second-strand cDNA synthesis was subsequently performed using DNA Polymerase I and RNase H. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities. After adenylation of the 3′ ends of DNA fragments, an adaptor with a hairpin loop structure was ligated to prepare for hybridization. The library fragments were purified with AMPure XP system (Beckman Coulter, Beverly, USA) to select cDNA fragments of 370–420 bp in length. PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. PCR products were purified (AMPure XP system), and library quality was assessed using the Agilent Bioanalyzer 2100 system. Clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina) according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina Novaseq platform, and 150 bp paired-end reads were generated. RNA-seq data generated in this study were deposited into Gene Expression Omnibus (GEO) at accession series GSE201990.

RNA-seq analysis

Raw data (raw reads) in the fastq format were first processed using in-house Perl scripts. Clean reads were obtained by removing reads containing adapters, reads containing poly-N and low-quality reads from raw data. Paired-end clean reads were aligned to mm10 genome assembly using Hisat2 v2.0.5. Feature Counts v1.5.0-p3 was used to count the read numbers mapped to each gene. Subsequently, the fragments per kilobase of transcript per million mapped reads (FPKM) of each gene was calculated based on the length of the gene, and the read count was mapped to this gene. Differential expression analysis was performed using the DESeq2 R package (1.20.0). Genes with a Log2FC (fold change) value of 1 as the cutoff and a P-value<0.05 as found by DESeq2 were assigned as differentially expressed.

GO and KEGG pathway enrichment analyses of the differentially expressed genes were performed using the clusterProfiler R package.

GSEA was performed using the local version of the GSEA analysis tool accessed from http://www.broadinstitute.org/gsea/index.jsp.

Chip-seq analysis

ChIP-seq data were downloaded from the GEO database (accession number GSE37151) (50). The index of the reference genome was built using BWA (v 0.7.12), and clean reads were aligned to the reference genome using BWA mem (v 0.7.12). After mapping reads to the reference genome, we used the MACS2 (version 2.1.0) (72) peak calling software to identify regions of immunoprecipitation enrichment over the background. A q-value threshold of 0.05 was used for all data sets. ChIPseeker (73) was used to retrieve the nearest genes around the peak and annotate the genomic region of the peak.

Conflict of Interest statement

The authors declare no competing interests.

Funding

This work was supported by the National Natural Science Foundation of China (81873878, 32000573, 82171162, 82192863, 81801126), the Shandong Provincial Natural Science Foundation (ZR2020QH154), the China Postdoctoral Science Foundation (2017M622187), and the National key research and development program (2016YFC1000806).

Author contributions

Y.J., Q.L., and J.G. designed the experiments. Y.J. wrote the manuscript. Y.J., X.G., G.C., N.R., and Y.S. performed the experiments and acquired the data. Y.J. and X.G. analyzed the date. M.L. and X.Y. helped analyze the cortical phenotypes related to human MCDs. C.H. and J.L. helped modifying experiments and approaches. Q.L. and J.G. supervised the project and reviewed the manuscript.