A Carboxy-terminal Smarcb1 Point Mutation Induces Hydrocephalus Formation and Affects AP-1 and Neuronal Signalling Pathways in Mice

The BAF (BRG1/BRM-associated factor) chromatin remodelling complex is essential for the regulation of DNA accessibility and gene expression during neuronal differentiation. Mutations of its core subunit SMARCB1 result in a broad spectrum of pathologies, including aggressive rhabdoid tumours or neurodevelopmental disorders. Other mouse models have addressed the influence of a homo- or heterozygous loss of Smarcb1, yet the impact of specific non-truncating mutations remains poorly understood. Here, we have established a new mouse model for the carboxy-terminal Smarcb1 c.1148del point mutation, which leads to the synthesis of elongated SMARCB1 proteins. We have investigated its impact on brain development in mice using magnetic resonance imaging, histology, and single-cell RNA sequencing. During adolescence, Smarcb11148del/1148del mice demonstrated rather slow weight gain and frequently developed hydrocephalus including enlarged lateral ventricles. In embryonic and neonatal stages, mutant brains did not differ anatomically and histologically from wild-type controls. Single-cell RNA sequencing of brains from newborn mutant mice revealed that a complete brain including all cell types of a physiologic mouse brain is formed despite the SMARCB1 mutation. However, neuronal signalling appeared disturbed in newborn mice, since genes of the AP-1 transcription factor family and neurite outgrowth-related transcripts were downregulated. These findings support the important role of SMARCB1 in neurodevelopment and extend the knowledge of different Smarcb1 mutations and their associated phenotypes. Supplementary Information The online version contains supplementary material available at 10.1007/s10571-023-01361-5.


Introduction
Neural development is accompanied by precisely coordinated changes in the chromatin state that determine the individual cells' fate and function. Indispensable to this process is, among others, the BAF (BRG1/BRM-associated factor) complex, which is the mammalian counterpart of the yeast SWI/SNF chromatin remodelling complex. It modulates the chromatin architecture in an ATP-dependent manner to allow DNA accessibility and gene expression (Ronan et al. 2013). The BAF complex is composed of up to 15 subunits determining its time and tissue-specific function thus enabling the complex to regulate differentiation and stemness (Alfert et al. 2019). Its critical role is reflected by the fact that mutations of BAF subunits are often detected in neurodevelopmental disorders (Sokpor et al. 2017;Bögershausen and Wollnik 2018) and in > 20% of all human cancers (Kadoch and Crabtree 2015).
During neural development, the switch from the neural progenitor BAF (npBAF) to neuronal BAF (nBAF) complexes marks the transition from neural progenitor cells to post-mitotic, differentiating neurons. Both complexes are composed of one ATPase subunit, SMARCA4 (BRG1) or SMARCA2 (BRM), the core subunits SMARCB1, SMARCC1 and SMARCC2, and various other proteins, including ARID1A and ARID1B. However, the nBAF complex differs from the npBAF complex in the expression of (I) ACTL6B instead of ACTL6A, (II) DPF1 or DPF3 instead of PHF10 or DPF2 and (III) SS18L1 instead of SS18 Alfert et al. 2019).
The SMARCB1 subunit includes an amino(N)-terminal DNA-binding domain (DBD), two repeat motifs and a short coiled-coil region. The winged helix DBD initiates the recruitment of the BAF complex to target genes (Allen et al. 2015), while the carboxy(C)-terminal coiled-coil region is believed to interact with the nucleosome acidic patch (Valencia et al. 2019).
Mutations of SMARCB1 are found in neurodevelopmental disorders as well as in a variety of malignancies (Holsten et al. 2018). Biallelic SMARCB1 inactivation, mostly due to whole gene and exon deletions and truncating variants, is characteristic of the development of malignant rhabdoid tumours at a young age (Versteege et al. 1998;Holsten et al. 2018). Rhabdoid tumours are aggressive paediatric cancers that are usually localized in the central nervous system (atypical teratoid rhabdoid tumour, ATRT), the kidney (rhabdoid tumour of the kidney, RTK) or soft tissues (malignant rhabdoid tumour, MRT) (Nesvick et al. 2020;Nemes et al. 2022). By contrast, nontruncating variants of SMARCB1, which are mostly located in the exons one, two, eight and nine, are associated with late-onset tumours, such as schwannoma and meningioma (Holsten et al. 2018), and with neurodevelopmental disorders, such as intellectual disability with plexus hyperplasia (Kleefstra et al. 2012;Diets et al. 2019) and the Coffin-Siris syndrome (Santen et al. 2013;Tsurusaki et al. 2014). The latter is a rare congenital disorder characterised by developmental delay, microcephaly, coarse facial features, and aplastic or hypoplastic fifth finger/toenails (Coffin and Siris 1960;Tsurusaki et al. 2014). Thus, germline and somatic mutations of SMARCB1 can lead to a wide range of pathologies. However, the functional causes are not yet fully understood.
While previously published mouse models have focused on the impact of homozygous (Moreno et al. 2014;Graf et al. 2022) or heterozygous (Filatova et al. 2019) Smarcb1 loss on neuronal development and tumorigenesis, models for non-truncating and/or point mutations of Smarcb1 are lacking. Yet, these are ultimately needed to decipher how the diverse array of mutant SMARCB1 leads to either uninhibited cell division in tumours or aberrant brain development in neurodevelopmental disorders.
In this study, we have established a new mouse model bearing the C-terminal Smarcb1 c.1148del mutation. This mutation results in the synthesis of elongated SMARCB1 proteins, which in turn might alter the composition and function of the BAF complex. We characterized the development of mutant brains by a multi-faceted approach including magnetic resonance imaging (MRI), histology, and single-cell RNA sequencing (scRNA-seq).
Microinjections were performed in B6D2F1 (hybrid between C57BL/6J and DBA strains) fertilized one-cell oocytes. Oocytes were removed from oviducts of superovulated B6D2F1 female mice in M2 medium (Behringer et al. 2014) supplemented with hyaluronidase (400 µg/mL, Sigma-Aldrich, #H3506), washed twice for removal of cumulus cells in M2 medium, transferred to KSOM medium (Behringer et al. 2014), and kept at 5% CO 2 and 37 °C before injection. Cytoplasmic microinjections were performed in M2 medium using the transjector 5246 (Eppendorf), and Narishige NT-88NE micromanipulators attached to a Nikon Diaphot 300 inverted microscope. Oocytes that survived microinjections were transferred to oviducts of pseudopregnant CD1 foster mice and carried to term. Positively targeted F0 and F1 animals were identified by quantitative PCR (qPCR) and sequencing analysis of genomic DNA isolated from ear biopsies. Smarcb1 1148del/+ mice were maintained on the C57BL/6 background and were bred to obtain Smarcb1 1148del/1148del mice. Animals of both sexes were used for the experiments. Whole body weights of at least 3 mutant and 3 wild-type animals were measured daily between days 30 and 40. All mice were monitored regularly. Between 1 and 5 animals were housed per cage. Protocols and animal housing were in accordance with all guidelines provided by the local regulatory authorities (reference number TVA84-02.04.2015. A088 and TVA81-02.04.2018.A214; Government of NRW, Germany).

Genotyping
DNA was extracted from ear biopsies by applying 200 µL lysis buffer (100 mM Tris, 5 mM EDTA, 0.2% (w/v) SDS, 200 mM NaCl), 7µL proteinase K (Roche, #03115828001) and 5µL ribonuclease A (20 mg/mL, Sigma-Aldrich, #R5000) and incubating overnight at 55 °C at 700 rpm. After enzyme inactivation at 85 °C for 5 min, the solution was centrifuged at 20800 g for 10 min at room temperature, DNA was precipitated using 2-propanol and washed with 70% (v/v) ethanol. After centrifugation (20800 g, 5 min) the DNA pellet was left to dry for at least 3 h before resuspension in nuclease-free water.

Magnetic Resonance Imaging
In vivo MRI was conducted on an Achieva 3.0 T Philips MR System operating at 127.74 MHz using a dedicated mouse radio frequency coil (Philips Research Laboratories 3T Solenoid Coil). Mice were narcotized with inhaled isoflurane (initiation 4% and preservation 1.2%, Abbott, #213111). 17 contiguous transversal slices with a slice thickness of 0.5 mm were acquired with a T1IR sequence and reconstructed to 0.15 mm in-plane pixel resolution. The T1-weighed images were analysed with ImageJ (Version 1.53q). Brain and ventricle volumes were determined from the measured brain or ventricle area multiplied by the slice thickness. Data processing and statistical tests were performed with Microsoft Excel 2016 and IBM SPSS Statistics (Version 28.0.1.1(14)).

Fig. 1
Enlarged cerebral ventricles in a Smarcb1 c.1148del mutant mouse model. a Scheme depicting the structure of murine Smarcb1 wild-type (wt) mRNA (top left) and the domain structure of both isoforms of the SMARCB1 wt protein (bottom left). Scheme showing the location of the c.1148del mutation in Smarcb1 mutant (mt) mRNA (top right) and the resulting elongated SMARCB1 mt isoforms (bottom right). Striped areas indicate segments at the 3′ end of exon 2 that are present in isoform 1 and missing in isoform 2 due to an alternate in-frame splice junction. b Immunoblot showing different isoforms of wild-type (wt iso 1/2) and mutant (mt iso 1/2) SMARCB1 proteins in whole-cell lysates of brain cells isolated from newborn Smarcb1 +/+ (wt) and Smarcb1 1148del/1148del (mt) mice. Lysates from the mouse embryonic fibroblast cell line NIH3T3 and a SMARCB1-negative peripheral T cell lymphoma cell line (T15) were included as positive and negative controls, respectively, and betaactin antibody was used as a loading control on the same immunoblot. c MRI of wild-type and mutant Smarcb1 1148del/1148del mice mouse brains. The brains of mutant mice show different degrees of enlarged lateral ventricles (asterisks) and smaller aqueducts (arrows). d Graphs showing comparisons of absolute brain volumes and absolute or relative ventricular volumes in age-matched (P35-P40) Smarcb1 +/+ mice (wt) (n = 5: 4 male, 1 female), Smarcb1 1148del/+ (ht) (n = 5: 4 male, 1 female) and Smarcb1 1148del/1148del (mt) (n = 5: 3 male, 2 female) mice. Manual quantifications were performed using ImageJ, and measurements were subjected to a Kruskal-Wallis test. WHD winged helix DNA binding, RPT1 repeat 1, RPT2 repeat 2, CC coiled-coil domain Immunohistochemistry For histological evaluation, mouse brains (week 4-12) or entire heads of embryos/newborn pups (E13.5/P0) were fixed in 4% PFA for at least 24 h. Afterwards, they were dehydrated, embedded in paraffin and 4 μm sections were cut. Hematoxylin and eosin (H&E) stainings were performed according to standard protocols. For immunohistochemical stainings, the ultraView DAB Detection Kit was used on a Ventana Benchmark xt system (both Roche Diagnostics) according to the manufacturers' instructions. The following antibody was used: anti-SMARCB1 (Clone 25/BAF47, BD Biosciences, #612110, 1:50) (Bruder et al. 1999;Carugo et al. 2019). For adolescence and P0, brain sections from at least three mutant animals were analysed. For E13.5, brain sections of two mutant animals were stained.

Single-Cell RNA Sequencing of Mouse Brains
Single-cell suspensions obtained from the supratentorial and infratentorial parts of two Smarcb1 1148del/1148del and two Smarcb1 +/+ P0 mice were stained with 7-Aminoactinomycin D (7-AAD) (eBioscience™, #00-6993-50). Non-viable, 7-AAD-positive cells were removed by fluorescence-activated cell sorting (BD FACS Aria II, BD FACS Diva Software), followed by manual counting with trypan blue staining (Sigma-Aldrich, #T8154). Approximately 25,000 single cells of each sample were processed for scRNA-seq using the Chromium Next GEM Single Cell 3′ GEM, Library & Gel Bead Kit v3.1 (10X Genomics, #1000121), the Chromium Next GEM Chip G (10X Genomics, #1000120) and the Single Index Kit T set A (10X Genomics, #1000213) according to the manufacturer's instructions. In brief, single-cell GEM (Gel Beads-in-Emulsion) were generated by the Chromium Controller, followed by GEM-RT (reverse transcription), Dynabeads cleanup, cDNA amplification, SPRIselect beads cleanup and Library Construction. Quality, purity, size and concentrations of cDNA and libraries were measured by a tapestation 2200 (Agilent Technologies). All libraries were sequenced separately by the NextSeq 500 sequencing platform (v 2.5 chemistry, 75 cycle kit) at the Core Facility Genomics (Medical Faculty of the Westfälische Wilhelms-Universität Münster, Münster).

Preprocessing of Single-Cell RNA Sequencing Data
Eight murine scRNA-seq samples were processed with the 10X Genomics CellRanger v3.0.2 program suite (Zheng et al. 2017). First, the raw 10 × scRNA-seq data was converted to the standard fastq format with CellRanger's mkfastq function. The CellRanger count algorithm was subsequently used to align the sample data against the murine reference transcriptome mm10 v3.0.0 with default values, and the resulting filtered feature matrices were checked for basic quality metrics with the CellRanger web summary reports. Samples with a sequencing saturation of < 35% were re-sequenced.

Analysis of Single-Cell RNA Sequencing Data
Further analyses of the CellRanger data were conducted with R v4.0.5 (RCoreTeam21 2021) and Seurat v4.0.5 (Hao et al. 2021). For all samples, a minimum of 3 cells and a minimum feature number per cell of 200 were required; all cells with more than 25% of mitochondrial genes were removed from the analysis. Furthermore, outlier cells with a high nCount_RNA value were classified as doublets and removed from the analysis; depending on the sample's cell distribution, the threshold for filtering was adapted (30,000 to 50,000, mean = 37,500). The filtered sample data was transformed with Seurat's SCTransform function and default parameters, and a principal component analysis (PCA) was used for dimension reduction. Subsequently, integration anchors were identified based on the normalization method SCT and RPCA as reduction. The dataset was then integrated and clustered with a resolution parameter of 0.5. Uniform manifold approximation and projections (UMAP) and Seurat feature plots were chosen to visualize both the sample distribution and a set of predefined features and gene signatures. Additionally, Seurat's FindMarkers function was used to compute differentially expressed marker genes for each cluster. Based on these calculated gene lists and marker plots, murine cell types were manually assigned to the Seurat clusters. The transcriptomic single-cell atlas of the Developing Mouse Brain (La Manno et al. 2021; www. mouse brain. org/ devel opment) and complementary publications for embryonic and perinatal brain development (Carter et al. 2018;Loo et al. 2019) were used as a reference for cell type annotation. Subclusterings were performed for all clusters with inhomogeneous marker plot results, so that the final clustering consisted of 37 distinct clusters.
As an additional layer of information, eight superclusters were introduced to structure the identified Seurat clusters further, and the resulting distribution of cells was analysed separately for the two murine groups and visualized as pie charts and bar charts with Microsoft Excel 2016. Gene Ontology term analysis was performed using ToppGene Suite (Chen et al. 2009; https:// toppg ene. cchmc. org/). Furthermore, expression values for chosen marker genes were extracted for both mouse groups and visualized with ggplot's boxplot function (Wickham 2016). Heatmaps were used to visualize the expression of chosen marker genes in different cluster sets (https:// cran.r-proje ct. org/ packa ge= pheat map). For the full murine data, a marker list published by Tirosh et al. (2016) was then used as input for Seurat's CellCycleScoring function to identify cell cycle phases in the integrated dataset. Additional data exploration was performed with the R/Bioconductor package iSEE v2.6.0 (Rue-Albrecht et al. 2018).

Statistics
A sample size of n ≥ 3 was chosen for all statistical analyses, with the single-cell analyses of two animals (four samples) per genotype being an exception. The allocation of the animals to the statistical groups was conducted according to their genotype. The evaluation was carried out without blinding the experimenter. Statistical tests were performed with IBM SPSS statistics (Version 28.0.1.1(14)) and R v4.0.5 (RCoreTeam21 2021). Data were tested for normal distribution using the Kolmogorov-Smirnov test and the Shapiro-Wilk test or the Anderson-Darling test for single-cell data. For comparisons of two groups, for normally distributed data, a two-tailed unpaired t-test was applied, and for the remaining data, a two-sided unpaired Mann-Whitney-U-Test was performed. When comparing more than two groups that were not normally distributed, the Kruskal-Wallis test was used instead. A subsequent Bonferroni correction was used to adjust for multiple testing. For detailed information on statistical tests, see the statistical report in the supplementary material.
Figures were created using Microsoft Excel 2016 and Adobe Illustrator 2022.
To obtain insight into the functional consequences of a C-terminal, non-truncating mutation, we established a genetically engineered mouse model. These mice carry a cytosine point deletion in exon nine, resulting in a frameshift of 36 amino acids until the following stop codon (Smarcb1 c.1148del/p.P383QfsX36). The resulting protein contains 32 additional amino acids located C-terminal to the coiled-coil region (Online Resource 3a), which, according to computational prediction tools (Gasteiger et al. 2005, https:// web. expasy. org/ compu te_ pi/), increases its molecular weight by 3.4 kDa. Due to alternative splicing within exon two of the Smarcb1 pre-mRNA, the wild-type (wt) SMARCB1 protein is expressed in two isoforms of 385 and 376 amino acids in length (wt iso 1 and wt iso 2, respectively) (Fig. 1a, b) (Bruder et al. 1999). Similarly, the Smarcb1 c.1148del pre-mRNA undergoes alternative splicing and encodes the mutant (mt) SMARCB1 isoforms with a length of 417 and 408 amino acids (mt iso 1 and mt iso 2), respectively (Fig. 1a, b).
In immunoblots, we detected roughly equal expression levels for wt iso 1 and 2 SMARCB1 proteins in NIH3T3 embryonic fibroblasts (Fig. 1b, Online Resource 4a). The same observation was made when using whole-cell lysates from brain cells of newborn wild-type mice, while mt iso 1 and 2 SMARCB1 proteins were exclusively detected in corresponding lysates from Smarcb1 1148del/1148del mice (Fig. 1b, Online Resource 4a). Further immunoblots demonstrated the expression of all isoforms in organs of adult wild-type and Smarcb1 mutant mice (Online Resource 3b, 4b,c,d). Heterozygous mice expressed both wt and mt SMARCB1 isoforms, yet, the Smarcb1 c.1148 mutation does not appear to induce preferential expression of either isoform. One exception can be identified in liver cells, where only isoform 1 was detected in all three genotypes (Online Resource 3b, 4d).
Taken together, the human SMARCB1 c.1148delC mutation has been transferred into a mouse model. It leads to the expression of prolonged SMARCB1 proteins in all organs examined, including the brain.

Smarcb1 1148del/1148del Mutation Impairs Brain Development and Growth
Heterozygous Smarcb1 1148del/+ mice were crossed to obtain homozygous Smarcb1 1148del/1148del mice, which were born at expected Mendelian ratios. Newborn Smarcb1 1148del/1148del mice were viable and did not show any apparent anomalies at the time of birth.
To visualize brain structures and ventricle shape, MRI was performed on five Smarcb1 +/+ mice, five Smarcb-1 1148del/+ mice and five Smarcb1 1148del/1148del mice, each aged between 35 and 40 days. Images showed enlarged lateral ventricles filled with cerebrospinal fluid (CSF) in all Smarcb1 1148del/1148del mice, while their third ventricles appeared barely dilated. In addition, the images of Smarcb1 1148del/1148del brains suggested an aqueduct stenosis (a known cause of hydrocephalus), while other brain structures such as the thalamus and cerebellum appeared largely unaltered (Fig. 1c, Online Resource 5). The lateral ventricles of the heterozygous Smarcb1 1148del/+ mice appear minimally enlarged, but beyond that, the brains show no other abnormalities (Online Resource 5).
Next, we investigated the impact of the Smarcb1 c.1148del mutation on a histological level. We performed H&E and anti-SMARCB1 immunohistochemical (IHC) stainings on brain sections from adolescent (week 4-8), newborn (P0) and embryonal (E13.5) Smarcb1 +/+ and mut ant Smarcb1 1148del/1148del mice (Fig. 2a-h). This analysis revealed that mutant adolescent mice have enlarged lateral ventricles and a thinned cortex (Fig. 2a-d). Whether the latter is the cause or consequence of the hydrocephalus remains unclear. Other brain regions like the choroid plexus, the ependyma incl. the subventricular zone, and the cerebellum show no abnormalities . Hyperproliferation of the choroid plexus as a cause of hydrocephalus can thus be largely excluded. No differences to wild-type animals were observed at the earlier time points P0 and E13.5 (Fig. 2e-h). In general, all cells throughout the brain are SMARCB1-positive ( Fig. 2b-II Areas of tumour formation or cellular hyperproliferation, as found in Smarcb1-negative ATRT models, are not detectable (Graf et al. 2022) (Fig. 2b, d, f, h). In addition, the brains of heterozygous Smarcb1 1148del/+ mice were examined at different time points, but showed no abnormalities compared to wild-type mice (Online Resource 6).
In conclusion, Smarcb1 1148del/1148del mice are 10-30% lighter compared to Smarcb1 +/+ mice and develop enlarged ventricles detectable in MRI with a penetrance of 100%. Among them, 60% of mutant animals develop a severe phenotype with visible hydrocephalus within 50 days after birth. Enlarged ventricles and thinner cerebral cortices can only be observed in adolescent Smarcb1 1148del/1148del mice. The excess of CSF can neither be explained by hyperproliferation of the choroid plexus nor by tumour growth. Thus, we concluded that the development of enlarged lateral ventricles takes place within the first 4 weeks of life. Therefore, we decided to further investigate newborn mice for indications of anomalies in brain development at the transcriptomic level.

Single-Cell Transcriptomics Shows the Cellular Heterogeneity of Smarcb1 +/+ and Smarcb1 1148del/1148del Mouse Brains
We performed scRNA-seq of cells isolated from the brains of two wild-type and two Smarcb1 1148del/1148del animals at their day of birth (P0) using the 10X Genomics platform.
Brains were cut between the mid-and hindbrain into supraand infratentorial parts, which were further processed and sequenced separately (Online Resources 7a, 8). After preprocessing and quality control, the eight samples with a total of 71,985 cells were integrated and projected onto a twodimensional map, on which wt and mt cells were evenly distributed (Fig. 3a, Online Resource 7b). We identified 37 clusters representing distinct cell types (Fig. 3a, Online Resource 9). Differentially expressed gene analyses and a cell cycle state plot enabled us to annotate each cluster by the dominant cell type (Fig. 3a, Online Resources 7c, 9, 10, 11). As a reference, we used a comprehensive transcriptomic single-cell atlas of the Developing Mouse Brain (La Manno et al. 2021; www. mouse brain. org/ devel opment) and complementary publications for embryonic and perinatal brain development (Carter et al. 2018;Loo et al. 2019). Each cluster was assigned to one of seven superordinate cell classes (Fig. 3b), except for cluster 9, where a low number of unique molecular identifiers and a low gene count prevented this. Cells of all clusters express Smarcb1 (Online Resource 7d).
Ciliated cells (Cfap126)  A comparison of the proportion of wild-type and mutant cells in the different cell classes revealed a slight shift from neurons to neuroprogenitors in the mutant brain (Fig. 3c). However, no significant difference between wild-type and mutant samples was found (Online Resource 7e).
These analyses demonstrate that at the time of birth the brains of Smarcb1 1148del/1148del mice still reflect the entirety of all cell types in a physiologic mouse brain. From this, we deduce that despite the SMARCB1 mutation, a complete brain including all cell types can be formed.

Single-Cell transcriptomes of Smarcb1 1148del/1148del neurons revealed altered signalling pathways
Next, we focussed on the expression pattern of neuronal clusters forming the largest cell class at time point P0 (Fig. 4a). We compared neurons of Smarcb1 wt and mt brains by performing gene ontology analysis (Chen et al. Fig. 4 Comparison of the transcriptome of Smarcb1 +/+ and Smarcb1 1148del/1148del brain cells. a Merged UMAP of brain cells. All neurons are indicated in blue. b Bar chart showing the enrichment of biological processes related to neurogenesis and neuron development in Smarcb1-mutant neurons compared to neurons in wild-type animals, as revealed by gene ontology analysis using ToppGene Suite (Chen et al. 2009; https:// toppg ene. cchmc. org/) c expression of AP-1 transcription factor family members, and d of neurite outgrowthassociated genes, in Smarcb1-mutant and wild-type neurons. *** Indicates p ≤ 0.001 (two-sided unpaired t-test, Bonferroni correction applied) 2009). In Smarcb1 mutant neurons, we detected an enrichment of transcription factors (e.g. Neurod2,Lhx2,Sox11) and biological processes connected to neuronal development (Fig. 4b). Moreover, several genes encoding for activator protein 1 (AP-1) transcription factor family members, such as Fos, Fosb, Jun, and Jund, showed significantly reduced expression levels in mutant neurons (Fig. 4c).
Furthermore, Smarcb1 1148del/1148del neurons showed a significantly lower expression of several important factors involved in neurite outgrowth, such as Stathmins (Stmn2 and Stmn3), Tubulins (Tubb2a and Tubb3), or Gap43 (Fig. 4d). While stathmins and tubulins are needed for microtubule organization in axons, dendrites and synapses, GAP43 regulates actin dynamics in neuronal growth cones (Hartl and Schneider 2019). Interestingly, Basp1, which can partly replace the function of Gap43 in neurite outgrowth (Korshunova et al. 2008), is upregulated in Smarcb1 1148del/1148del neurons (Fig. 4d). This points to an altered formation of neurites and axons in mutant neurons, which has already been described for other C-terminal mutations of SMARCB1 (Valencia et al. 2019) as well as for mutations of other BAF subunits, i.e. for Arid1b and Actl6b Ka et al. 2016;Jung et al. 2017).

Discussion
From the broad spectrum of SMARCB1 mutations, we investigated the C-terminal Smarcb1 c.1148del mutation in a newly established mouse model. Homozygous mutant mice showed delayed weight gain and enlarged lateral ventricles in MRI and histology. This phenotype was not found in heterozygous mice. ScRNA-seq revealed the differentiation of progenitor cells into all expected cell types but altered AP-1 and neurite outgrowth signalling pathways in Smarcb1 1148del/1148del neurons.
The human SMARCB1 c.1148 cytosine point deletion (SMARCB1 c.1148del/p.P383RfsX100 (COSM1057)) is closely associated with the formation of ATRT of the central nervous system and was also reported in tumours of the soft tissue and the thyroid gland (Tate et al. 2019). ATRT with this mutation are SMARCB1-negative on IHC staining, indicating that the frameshift of 100 amino acids in the mutated human protein causes unstable or misfolded proteins (Tsai et al. 2012). In contrast, no tumour formation was detected in Smarcb1 1148del/1148del mice and in IHC stainings, all cells were SMARCB1-positive. The cause for this deviant phenotype might be that the same mutation at DNA level leads to a frameshift of only 36 amino acids in mutated murine proteins, resulting in an elongated SMARCB1 protein that might still be partially functional.
Mutations of the Smarcb1 C-terminal region can predispose to the neurodevelopmental disorder Coffin-Siris syndrome (Holsten et al. 2018;). Valencia et al. (2019) hypothesize that this is due to disruption of the binding between the coiled-coil C-terminal domain and the nucleosome acidic patch. Most Coffin-Siris syndrome characteristics such as microcephaly and abnormal corpus callosum did not apply to Smarcb1 1148del/1148del mice. Shared abnormalities were growth delay, which has been observed especially in SMARCB1 pathogenic variants (Santen et al. 2013), and hydrocephalus, which has only recently been described as a common prenatal characteristic in ARID1A pathogenic variants (but not in SMARCB1 variants) (van der Sluijs et al. 2022).
The most striking characteristic of Smarcb1 1148del/1148del mice is hydrocephalus development. In general, hydrocephalus can be classified into acquired hydrocephalus due to haemorrhage, infection or neoplasm, and congenital/ developmental hydrocephalus (CH) without any extrinsic cause (Tully and Dobyns 2014). Studies of CH have led to the identification of several pathogenic human genes (Tully and Dobyns 2014;McKnight et al. 2021), including genes of the BAF complex: de novo and transmitted SMARCC1 mutations were identified in CH patients with stenosis of the aqueduct (Furey et al. 2018;Jin et al. 2020;). Diets et al. (2019) described a SMARCB1 mutation in exon 2 (c.110G > A; p.Arg37His) that causes intellectual disability and hydrocephalus due to choroid plexus hyperplasia. In addition, BAF mouse models showing hydrocephalus were established for Smarca4/Brg1 (Cao and Wu 2015;Holdhof et al. 2020) and Arid1b (Celen et al. 2017;Shibutani et al. 2017).
In our mouse model, Smarcb1 1148del/1148del mice developed enlarged ventricles visible in MRI in all cases and externally visible hydrocephalus with a penetrance of 60%. Histology revealed an unremarkable choroid plexus and no evidence of tumour formation, inflammation, or haemorrhage. Since P0 brains show no reduced neuron count in neither histology nor scRNA-seq, we conclude that hydrocephalus e vacuo is unlikely. MRI indicates that obstruction of the aqueduct might be causing the hydrocephalus. We interpret this as an indication that Smarcb1 1148del/1148del mice might develop congenital non-communicating hydrocephalus within their first 50 days of life, whereby the incomplete penetrance might be a consequence of varying degrees of restricted CSF circulation. Further investigation is needed to understand the exact aetiology.
Apart from hydrocephalus and thinned cortices in adolescent mice, we could not observe any further structural brain abnormalities in Smarcb1 1148del/1148del mice. Therefore, the Smarcb1 c.1148del mutation seems to have a rather small influence on murine brain development compared to other Smarcb1 mutations, which have been described to cause a broad variety of brain abnormalities. For example, the biallelic loss of Smarcb1 in defined cell populations was described to cause embryonic lethality, rhabdoid tumours or cerebellar hypoplasia (Moreno et al. 2014;Graf et al. 2022). Heterozygous Smarcb1 disruption in neural stem/progenitor cells led to microcephaly, brain midline abnormalities and choroid plexus hyperplasia (Filatova et al. 2019). None of these severe neurodevelopmental malformations were present in Smarcb1 1148del/1148del mice, suggesting that the function of SMARCB1 within the BAF complex is largely preserved.
Although the Smarcb1 c.1148del mutation still allows the formation of a complete brain, we hypothesized that it might alter protein interactions and thereby affect cell differentiation and/or signalling pathways in developing neural cells. To investigate these differences triggered by the mutation on a transcriptomic level, we performed scRNA-seq on two wild-type and two mutant P0 mouse brains.
The comparison of wild-type and mutant brains revealed a similar distribution and similar numbers for all cell classes. Only a slight non-significant shift from neurons to neuronal progenitor cells could be observed. Hence, a similar block of neuronal differentiation as observed in cerebral organoids upon knockdown of SMARCB1 during early neural differentiation (Parisian et al. 2020) could not be detected. However, neuron differentiation in mutant brains might be somewhat delayed, since we observed an enrichment of biological processes and transcription factors connected to neuronal development in the mutant post-mitotic neurons.
Moreover, our dataset revealed two important signalling pathways affected by the mutation. First, we showed a lower expression of AP-1 transcription factors in mutant neurons. These factors include various protein dimers made up of the members of the Jun, Fos, Maf and ATF families (Bejjani et al. 2019). FOS/JUN dimers were shown to interact with the SMARCD1 subunit of the BAF complex (Ito et al. 2001), and probably recruit the BAF complex in order to mediate chromatin accessibility (Vierbuchen et al. 2017). This close relationship seems to be influenced by the Smarcb1 c.1148del mutation, resulting in lower levels of AP-1 transcription factors, which in turn might alter related neuronal functions, including neuronal differentiation (Pagin et al. 2021), neurite outgrowth (Dragunow et al. 2000;Jessen et al. 2001;Seijffers et al. 2006), and response to neuronal-activity (Su et al. 2017).
Second, we detected a reduced expression of factors that are important for neurite outgrowth via growth cones. These include Stathmin and Tubulin, which participate in microtubule organization, and growth-associated protein-43 (Gap43) and brain acid-soluble protein 1 (Basp1), which are involved in neurodevelopment, synaptic function and nerve regeneration (Chung et al. 2020). In Smarcb1 mutant neurons, decreased levels of Stmn3, Stmn2, Tubb2a and Tubb3 suggest a reduced promotion of neurite outgrowth. Basp1 levels, on the other hand, were increased, which suggests an alternative activation route to compensate for low Gap43 levels. It is known that BASP1 can substitute GAP43 in terms of the induction of NCAM-independent neurite outgrowth, but not NCAM-mediated neurite outgrowth (Korshunova et al. 2008). The altered expression levels could possibly be a consequence of reduced AP-1 signalling, as ATF-3 (Seijffers et al. 2006), c-Jun (Dragunow et al. 2000) and c-Fos (Jessen et al. 2001) have been shown to promote neurite outgrowth.
Our findings are in line with previous studies showing neurite outgrowth deficits in neurons with BAF mutations. Differentiated neurons derived from induced pluripotent stem cells harbouring a heterozygous SMARB1 p.K364del mutation showed less neurite outgrowth than wild-type controls (Valencia et al. 2019). Actl6b −/− hippocampal neurons displayed shorter and less complex dendritic growth . Arid1b haploinsufficiency in cortical and hippocampal pyramidal neurons caused fewer and shorter dendritic spines. Additionally, decreased levels of neurite-associated transcripts (Stmn2, Gap43), Arc and the AP-1 transcription factor c-Fos were observed, just as in our mouse model. Ka et al. (2016), therefore, hypothesize that ARID1B participates in neurotrophinmediated c-Fos and/or Arc expression and thereby regulates dendritic development. These findings suggest that a well-functioning BAF complex including a functional DNA-binding SMARCB1 protein is crucial for neurite outgrowth and dendritic development and allows for an appropriate reaction to external stimuli. Impaired SMARCB1 function could disturb neurite outgrowth and synapse formation and may lead to intellectual disability.
Overall, in this study, we have presented the establishment of a new mouse model with a Smarcb1 c.1148del mutation. We have shown that this mutation leads to hydrocephalus and weight loss in homozygous mice. Using scRNA-seq, we have illustrated that although the mutation still allows the formation of differentiated cell types, it interferes with important signalling pathways such as AP-1 transcription factors and neurite outgrowth. This study thus confirms the crucial role of SMARCB1 in neuronal development and in the pathogenesis of neurodevelopmental disorders.

Data Availability
The single-cell RNA sequencing data generated in this study have been deposited in NCBI's Gene Expression Omnibus (Edgar et al. 2002) and are accessible through GEO Series accession number GSE212672, https:// www. ncbi. nlm. nih. gov/ geo/ query/ acc. cgi? acc= GSE21 2672.

Declarations
Competing interests The authors have no relevant financial or nonfinancial interests to disclose.

Consent for Publication Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International 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/.