Generation of new transgenic SMARCA4-deficient mouse models results in neuromuscular weakness and paralysis of limbs.

The SWItch/sucrose non-fermentable (SWI/SNF) complex is an ATP-dependent chromatin-remodeling complex, which is crucially involved in the regulation of gene expression. Many physiological processes during the development of the nervous system essentially rely on transcriptional regulation through the SWI/SNF complex, such as the differentiation of Schwann cells, which are required for the myelination of peripheral nerves [1]. Besides its regulatory role in normal cellular development, the SWI/SNF complex is known to be deregulated in various human cancers, either by mutations in genes encoding the subunits of the complex or by overexpression of these genes. In fact, alterations in at least one of the genes that make up the complex are observed in a large number of malignancies [1, 2]. Mutations in SMARCA4, which is one of the two ATPase subunits of the SWI/SNF complex, are not only frequently found in various carcinoma types of adult patients, but have also been described as driver mutations in the development of a subset of pediatric rhabdoid tumors (RT) [3]. RT are rare, highly aggressive embryonal tumors mostly affecting young children, with a poor prognosis, and an overall survival of 17 months. They comprise RT of the central nervous system (CNS), which are known as atypical teratoid/rhabdoid tumors (AT/RT), RT of the kidney (RTK), and malignant RT of the soft tissue (MRT). All RT show a biallelic inactivation of the SWI/SNF subunits SMARCB1 or SMARCA4 [2, 3]. Mutations in SMARCB1 leading to the development of RT have been characterized by various mouse models, whereas, to date, SMARCA4 deficiency has never caused RT development in vivo. So far, there is no cell line for SMARCA4-deficient RT. Therefore, in vitro and in vivo models for SMARCA4-deficient RT are urgently needed to better understand the oncogenesis and to develop potential treatment options. Recently, neural crest cells were suggested as one potential cellular origin of SMARCB1-deficient RT [2]. Such cells express the glycoprotein myelin protein zero (MPZ, P0), which is also known to be the major structural protein in peripheral myelin. In a previous study, P0-Cre mice were used to conditionally delete Smarcb1, which resulted in the development of RT [4]. In addition, genomic and single-cell transcriptomic analyses of human RT-derived organoids revealed a shared cellular lineage of SMARCB1-mutated RT cells and neural crest-derived Schwann cells [2]. To examine, whether P0 also marks cells of origin of SMARCA4-deficient RT, we crossed P0-Cre mice [5] with Smarca4fl/fl mice [6] to generate P0-Cre::Smarca4fl/wt (n = 33) and P0-Cre::Smarca4fl/fl mice (n = 17). As a control, P0-Cre::Smarcb1fl/lf mice (n = 23) were produced by crossing P0-Cre mice [5] with Smarcb1fl/fl mice [7]. All mice were bred on a C57BL/6J background and mice of both sexes were used for the study. No abnormalities were observed in heterozygous mice (P0-Cre::Smarca4fl/wt), but all homozygous mice developed neuromuscular weakness and partial or complete fore- and hind limb paralysis within 10 weeks (Figure 1A). To further analyze and classify this phenotype, a score for the symptoms of P0-Cre::Smarca4fl/fl mice was adapted from Weider et al. [8] (Figure 1B). Symptoms started at 3 weeks of age and increased continuously as visualized in Figure 1C. Body weight control revealed that homozygous mice were significantly lighter compared to heterozygous and control mice after 4 (p = 0.018) and 5 weeks (p = 0.028, Figure 1D). Mice were sacrificed based on the severity of paralysis and weight loss. The median overall survival was 65 days (Figure 1E). However, none of these mice developed tumors, as shown by thorough macroscopic inspection of all mice and histological analyses of the brains and kidneys in a subset of mice. Additionally, control mice with the genotype P0-Cre::Smarcb1fl/fl (n = 23) developed the same neuromuscular phenotype after the age of 10 weeks without any clinical or macroscopic signs of tumor development. Since SMARCA4 loss in other cell types than neural crest cells previously failed to model RT development as well, our goal was to develop a mouse model that mimics the human situation more specifically. Therefore, we next generated Smarca4-p.R1189* knock-in mice carrying a constitutive point mutation. The same mutation had previously been described in the germline of a family with RT predisposition syndrome leading to the development of RT during infancy of two siblings [9]. Knock-in mice were generated by pronuclear injection in 1-cell stage embryos after induction of a c.3565C>T (p.R1189*) point mutation via CRISPR/Cas9. In human patients, this mutation leads to a nonsense-mediated decay of the transcript [9] (Figure 1A). Mice carrying a heterozygous Smarca4-p.R1189* point mutation (n = 17) did not show any abnormalities over the observation period of 1 year. However, none of the 75 mice that were born after further crossing of heterozygous mice and none of the embryos genotyped at embryonal day 10.5 (E10.5) and E16.5 had a homozygous genotype, suggesting an early embryonal death of mice homozygous for Smarca4-p.R1189* (Figure 1F). Finally, P0-Cre::Smarca4fl/wt mice were crossed with Smarca4-p.R1189*+/− mice for the generation of P0-Cre::Smarca4fl/p.R1189* mice. In this way, a constitutive mutation–similar to a human germline mutation - and a conditional mutation in neural crest cells were combined to recapitulate the potential situation in pediatric patients with RT predisposition syndrome and to potentially avoid embryonic death of the mice. All P0-Cre::Smarca4fl/p.R1189* mice (n = 6) showed a uniform phenotype similar to P0-Cre::Smarca4fl/fl mice, with neuromuscular weakness and paralysis of the fore- and hind limbs (Figure 1A) and a median overall survival of 41 days (Figure 1E). These results indicate that a homozygous Smarca4-p.R1189* mutation is impairing SMARCA4 function in mice, but they also indicate that a recombination of Smarca4 through the above described P0-Cre strain is still insufficient for tumor development, even if the other allele carries a constitutive loss-of-function mutation. In conclusion, homozygous Smarca4 mutations introduced in the P0-Cre mouse line used in this study cause neuromuscular weakness and paralysis of the fore- and hind limbs. A similar phenotype was previously observed in Schwann cell-specific transgenic DHH-Cre::Smarca4fl/fl [8] and DHH-Cre::Smarcb1fl/fl mice [4]. In both DHH-Cre mouse lines, the observed phenotype could be attributed to an impaired myelination of peripheral nerves. Further analyses of our P0-Cre::Smarca4fl/fl mouse model confirmed genomic recombination of Smarca4fl/fl alleles (Figure S1A) and loss of SMARCA4 (Figure S1D,E) together with Cre-expression (Figure S1A) in the peripheral nerves of P0-Cre::Smarca4fl/fl mice. Here, we could prove that myelination of peripheral nerves is also impaired in our P0-Cre::Smarca4fl/fl mouse model. Histopathological assessment using toluidine blue staining (TBO) revealed thinner and impaired myelin sheaths surrounding axons of sciatic nerves of P0-Cre::Smarca4fl/fl mice (Figure S1H,I) compared to Smarca4fl/fl control mice (Figure S1F,G). Histological assessment of thigh muscles of P0-Cre::Smarca4fl/fl mice (Figure S1L,M) also revealed abnormalities in contrast to Smarca4fl/fl control mice (Figure 1J,K) showing a neurogenic image of muscle tissue with angular atrophic myocytes (Supplementary Figure 1M) in muscles that were innervated by impaired sciatic nerves. No histological abnormalities were detected in the spinal cord of P0-Cre::Smarca4fl/fl mice (Figure 1P,Q) compared to Smarca4fl/fl control mice (Figure S1N,O). The similarity of phenotypes caused by a Smarca4-knockout in DHH-Cre, which starts to be expressed at E12.5 [4], and our P0-Cre, which is presumably expressed from E13.5 onwards [10] indicates that both Cre-lines target a cellular compartment, which requires a functional SWI/SNF complex for proper myelination of peripheral nerves. However, these results highlight that the choice of the Cre-driver is critical for targeting the exact cellular population of origin for a tumor type of interest. For example, previous tumor generation in a P0-Cre::Smarcb1fl/fl mouse strain, in which P0-Cre has been shown to be expressed at E9.5 [4], could not be reproduced in another P0-Cre mouse strain, for which the expression first starts at E13.5 [10]. Overall, our results indicate that the initiation of RT from neural crest-derived cells might require an inactivation of Smarca4 earlier than E13.5 or that RT originate from a different cell type. While such early time points have previously been identified for the development of Smarcb1-mutated RT [4], the exact temporal window for the initiation of SMARCA4-deficient RT and the cells of origin remain to be determined. In fact, the developmental time point of origin may even be earlier than for SMARCB1-mutated RT, since the median age of diagnosis is significantly lower in patients with SMARCA4-deficient AT/RT [3]. We thank Vanessa Thaden and Jacqueline Tischendorf for excellent technical support. We thank Michael Reuter and Helen Morrison for the generous gift of P0-Cre mice. We acknowledge the great support of the Small Animal Core facility at the Leibniz-Institute of Virology. We also want to thank Irm Hermanns-Borgmeyer and Uwe Borgmeyer from the Transgenic Animal Unit at the Center for Molecular Neurobiology (ZMNH) for assistance with generating Smarca4-p.R1189* mice. This work was supported by the Wilhelm-Sander Stiftung, the Fördergemeinschaft Kinderkrebszentrum Hamburg and the University Cancer Center Hamburg. The authors declare no conflicts of interest. Data sharing is not applicable to this article as no new data were created or analyzed in this study. Figure S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.