Knockdown of Son, a mouse homologue of the ZTTK syndrome gene, causes neuronal migration defects and dendritic spine dysgenesis

Zhu-Tokita-Takenouchi-Kim (ZTTK) syndrome, a rare congenital anomaly syndrome characterized by intellectual disability, brain malformation, facial dysmorphism, musculoskeletal abnormalities, and some visceral malformations is caused by de novo heterozygous mutations of the SON gene. The nuclear protein SON is involved in gene transcription and RNA splicing; however, the roles of SON in neural development remain undetermined. We investigated the effects of Son knockdown on neural development in mice and found that Son knockdown in neural progenitors resulted in defective migration during corticogenesis and reduced spine density on mature cortical neurons. The induction of human wild-type SON expression rescued these neural abnormalities, conrming that the abnormalities were caused by SON insuciency. We also applied truncated SON proteins encoded by disease-associated mutant SON genes for rescue experiments and found that a truncated SON protein encoded by the most prevalent SON mutant found in ZTTK syndrome rescued the neural abnormalities while another much shorter mutant SON protein did not. These data indicate that SON insuciency causes neuronal migration defects and dendritic spine dysgenesis, which seem neuropathological bases of the neural symptoms of ZTTK syndrome. In addition, the results strongly suggest that the neural abnormalities in ZTTK syndrome are caused by SON haploinsuciency independent of the types of mutation that results in functional or dysfunctional proteins.

SON is a ubiquitously expressed and evolutionarily conserved gene in vertebrates and is located on the human chromosome region 21q22.11 [4]. It encodes the DNA-and RNA-binding protein SON, which functions in RNA splicing as well as gene repression [7][8][9][10][11][12]. A wide variety of genes are, thus, under the control of SON, and SON has been reported to be involved in cell cycle regulation and stem cell maintenance [7][8][9][10][11]. However, the functional signi cance of SON in neural development is largely unknown, and the pathological consequence of SON haploinsu ciency underlying the neural phenotypes of ZTTK syndrome, such as ID and brain malformation, remains undetermined. In this report, we revealed through knockdown experiments in the developing mouse brain that Son insu ciency caused neuronal migration abnormalities and dendritic spine dysgenesis. Rescue experiments that induced the expression of human wild-type SON protein and truncated SON proteins encoded by disease-associated mutant SON genes provided further information relevant to the pathophysiology of ZTTK syndrome.

Animals
All animals were used in accordance with an animal protocol approved by the Animal Care and Use Committee of the Institute for Developmental Research, Aichi Developmental Disability Center. Timedpregnant ICR mice were purchased from Japan SLC (Hamamatsu, Japan).

Antibodies
We raised an antibody against mouse SON by immunizing rabbits with a keyhole limpet hemocyanin-

Cell Culture
Neuro-2a and HEK293 cells were obtained from ATCC (Manassas, VA) and RIKEN BRC (Tsukuba, Japan), respectively. Each cell line was maintained with DMEM supplemented with 10% FBS, penicillin, and streptomycin under standard conditions. The expression plasmids were transfected with polyethyleneimine (Polysciences, Inc., Warrington, PA) or Lipofectamine 2000 (Invitrogen), according to manufacturer's directions.

In Utero Electroporation (IUE)
Various combinations of plasmids were transfected into neural progenitors on the lateral ventricular surface of E14.5 embryos by IUE as previously described [15]. Electroporation was performed by administering ve consequent electronic pulses at an intensity of 35 V for a duration of 50 ms with 450ms intervals using a NEPA21 SuperElectroporator (NEPA Gene, Chiba, Japan). For neuronal migration analysis, 1 mg of shRNA vector with 1.5 μg of the pCAGGS vectors harboring the various forms of human SON cDNA described above were applied. For dendritic spine formation analysis, a plasmid mixture containing 1 mg of shRNA vector with or without 0.5 μg of the hSON expression vectors was applied.

Immunocytochemistry and Immunohistochemistry
Neuro-2a cells were grown on poly-L-lysine coated glass coverslips. The cells were xed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. Immunocytochemical staining was performed with the anti-SON antibody and Alexa-conjugated secondary antibodies (Invitrogen).
Para n-embedded brain tissues were sectioned at a thickness of 4 mm and subjected to immunohistochemical staining and hematoxylin and eosin (HE) staining. Anti-SON immunoreactivity was visualized using EnVision (Dako, Glostrup, Denmark). All images were acquired with BX60 microscope (Olympus, Tokyo, Japan).

Neuronal Migration Analysis and Spine Density Analysis
Mice subjected to IUE were sacri ced at E18.5 or P60, and perfused with 4% paraformaldehyde in PBS.
Then the mice were dissected and obtained brains were post xed in paraformaldehyde solution for 2-24 hours. The brains were embedded in 3% agarose and sectioned at a thickness of 50-100 mm using a VT1200S vibrating microtome (Leica Microsystems Wetzlar, Germany). The coronal sections were stained with an anti-GFP antibody to visualize Son knockdown cells. The distribution of GFP-positive cells at E18.5 was examined to assess neuronal migration. For spine density analysis, dendrites of pyramidal neurons in cortical layer II/III at P60 were examined. The number of spines on each dendrite at between 30 mm and 80 mm from the soma was counted, and the spine density was represented as the number of spines per dendrite length of 10 mm. All images were acquired using a FV1000 (Olympus) or LSM880 (Carl Zeiss, Göttingen, Germany) confocal laser scanning microscope. Image processing was performed with Fiji (http:// ji.sc) and Photoshop (Adobe Systems, San Jose, CA).

Statistical Analyses
Statistical signi cance was determined using one-way ANOVA followed by a Dunnett's post hoc test for multiple comparisons using Prism 8 (GraphPad Software, San Diego, CA).

Results
The human SON and mouse Son genes encode proteins of 2,426-amino-acid and 2,444-amino-acid proteins, respectively, and share 84.2% homology [12] (Fig. 1a). We raised an antibody against mouse SON and found that it recognized a major band of approximately 260 kDa, which corresponds to the predicted size, and a few lower molecular weight bands upon Western blot analysis of mouse embryonic brain lysates (Fig. 1b). These bands were almost completely absent after the antibody was preabsorbed with the antigen peptide, con rming its speci city (Fig. 1b, right panel). In immunocytochemistry, the antibody recognized nuclear speckles in Neuro-2a cells, and the signals were partially colocalized with SRSF2, a splicing factor that, with SON, forms the core of speckles [16], further con rming the speci city of the antibody (Fig. 1c). The antibody worked well for immunohistochemistry as well, and we found that every neural progenitor in the developing mouse brain at E15.5 abundantly expressed SON and that the expression was maintained in mature neurons at P60 (Fig. 1d).
To reveal the functional signi cance of SON in neural progenitors, we then applied IUE to knockdown Son speci cally in these cells and examined the effects on migration. We generated two shRNA expression constructs targeting independent sites of Son mRNA (Fig. 2a) and con rmed that both shRNAs reduced SON expression levels in Neuro-2a cells to 17.8% (shRNA#1) and 32.6% (shRNA#2) of that in the control (Fig. 2b). IUE was performed at E14.5 to deliver shRNA#1 or shRNA#2 to neural progenitors, and the distribution of SON knockdown cells in the developing cortex at E18.5 was examined. As shown in Fig.  2c, GFP-positive Son knockdown neurons (shRNA#1 and shRNA#2) in the upper cortical plate (UCP) were sparse compared with GFP-positive neurons without Son knockdown (control), while Son knockdown neurons in the lower cortical plate (LCP) and intermediate zone (IZ) appeared less sparse than control neurons. The quanti cation of GFP-positive neurons in each cortical layer revealed that fewer Son knockdown neurons than control neurons were distributed in the UCP (Fig. 2d). Although slightly more Son knockdown neurons than control neurons were distributed in the LCP and IZ, the differences were not statistically signi cant. In addition, we examined SON expression levels in GFP-positive shRNAintroduced cells in electroporated samples, and con rmed that the SON signals in GFP-positive cells was hardly detectable, while that in GFP-negative cells was clearly observed as nuclear speckles (Fig. 2e).
These results indicate that canonical Son expression in neural progenitors is indispensable for normal neuronal migration.
Next, we performed rescue experiments by overexpressing shRNA-resistant human SON (hSONr) in knockdown cells. In addition, we examined constructs expressing two forms of disease-associated mutant SON. hSONm1 is a truncated mutant without most of the known functional domains, while hSONm2 lacks RNA-binding motifs and the C-terminal half of the RS domain (Fig. 3a). The former is derived from a SON mutation reported by Kim et al [1], and the latter is from the most prevalent mutation found in ZTTK syndrome [1,3,4]. The effective production of hSONr, hSONm1, and hSONm2 in HEK293 cells in the presence of shRNA#1 was con rmed (Fig. 3b). Then, the vectors expressing wild-type or mutant SON were introduced along with shRNA vectors into neural progenitors at E14.5, and the distribution of GFP-positive neurons was examined. As shown in Fig. 3c and d, migration defects induced by shRNA#1 were rescued by the overexpression of hSONr (shRNA#1 + hSONr), con rming that the defects were caused by SON insu ciency in neural progenitors. Intriguingly, the overexpression of hSONm2 (shRNA#1 + hSONm2), but not that of hSONm1 (shRNA#1 + hSONm1), rescued the defects as well, indicating that hSONm2, like hSONr, exerts su cient functions for neuronal migration, while hSONm1 does not.
Since ID is not always accompanied by cerebral cortical malformation due to migration abnormalities, we reasoned that SON haploinsu ciency affects other factors essential for intellectual abilities. Therefore, we examined the dendritic spine density on Son knockdown neurons. The density of dendritic spines on Son knockdown cortical neurons at P60 (7.6 ± 0.5 per 10 mm) was decreased by approximately 30% compared to that on control neurons (10.6 ± 0.9 per 10 mm) (Fig. 4a, b). The forced expression of hSONr resulted in a dendritic spine density nearly equal to that on control neurons, con rming that SON insu ciency resulted in decreasing in the numbers of dendritic spines (Fig. 4a, b). These data indicate that SON is necessary for normal spine formation during neural development and that SON haploinsu ciency may cause dendritic spine dysgenesis. Again, the overexpression of hSONm2, but not that of hSONm1, rescued the dysgenesis as well, indicating that hSONm2 retains functions necessary for spine formation, while hSONm1 does not (Fig. 4a, b).

Discussion
In this report, we clari ed that the canonical expression of Son is necessary for normal neuronal migration and dendritic spine formation in the developing mouse cerebral cortex. In addition, a truncated form of SON encoded by the most prevalent mutant SON identi ed in ZTTK syndrome patients can ameliorate the neural abnormalities induced by Son knockdown. These ndings shed light on the pathophysiological mechanisms underlying the neural symptoms of ZTTK syndrome.
Among the 31 de novo mutations in the SON gene reported to be associated with ZTTK syndrome so far, twenty-eight encode truncated SON proteins due to either frameshift or nonsense substitutions that generate premature termination codons (Table 1). These truncated SON proteins, if produced, vary in length, with their C-termini distributed widely over the normal full-length SON protein and function differently from one another. However, mRNAs bearing a premature termination codon are often targeted by NMD, and are degraded [17]. Kim et al revealed that some mutant SON mRNAs and proteins are indeed highly downregulated in the peripheral blood mononuclear cells of patients [1], suggesting that ZTTK syndrome is caused by SON haploinsu ciency; truncated proteins encoded by mutant SON are not involved in pathogenesis of ZTTK syndrome. The pathophysiological consequence of SON insu ciency in brain development is therefore of great interest for understanding the neural symptoms of ZTTK syndrome. In this respect, our nding that the number of neurons that migrated into the UCP decreased by approximately 20% due to Son knockdown provides a concrete evidence that SON insu ciency in neural progenitors results in migration defects, which seems to be the pathological basis of brain malformation in ZTTK syndrome. More importantly, we found reduced spine density on Son knockdown neurons. In humans, dendritic spine dysgenesis, such as a reduction in the number of spines and morphological abnormalities of spines, in cortical neurons is regarded to be a common pathological feature found in ID patients [18]. To our knowledge, postmortem reports of ZTTK syndrome patients are unavailable. Therefore, reduced spine density on Son knockdown neurons is an important nding and suggests that spine dysgenesis is the pathological basis of ID in ZTTK syndrome. # These two types of mutation were examined in this study. ## An in-frame deletion and a frameshift insertion were identi ed in one allele; the latter was regarded as pathogenic. ### Two substitutions were identi ed in one allele.
Rescue experiments that involved the introduction of two forms of disease-associated mutant SON proteins con rmed that the truncated SON proteins encoded by mutant SON genes differ in their residual functions, even though both mutations cause ZTTK syndrome. The overexpression of a mutant SON protein lacking an RNA-binding motif and part of the RS domain (hSONm2), like wild-type SON, successfully rescued the neuronal abnormalities, while the overexpression of a mutant SON protein lacking additional regions, including repetitive amino acids and a DNA-binding region (hSONm1), failed to do so. These data indicate that the hSONm1-coding mutation is a loss-of-function mutation, but that the hSONm2-coding mutation maintains function comparable, at least in the neural development, to that of wild-type SON. Therefore, it is strongly suggested that hSONm2, which encoded by the most prevalent SON mutation found in ZTTK syndrome, is not expressed in the brains of patients, possibly due to NMD. Otherwise, the mutation would not result in the disease because of the residual functions of the mutant SON gene. Such unfavorable degradation of mutant gene mRNAs bearing a premature termination codon that encodes a functional protein occurs in many other genetic diseases, and NMD inhibitors have been tested as potential therapeutic agents [17,19]. Based on our ndings, NMD inhibition therapy may be effective in a majority of ZTTK syndrome patients.
The molecular mechanisms underlying the neural abnormalities caused by SON mutations remain unclear because the roles of the multifunctional nuclear protein SON are diverse and not fully understood. The nding that hSONm2, which lacks an RNA-binding motif, behaved like wild-type SON in the rescue experiments may provide hints for these mechanisms. A simple explanation for this result is that the loss of the RNA splicing function of SON does not play a signi cant role in the observed neural abnormalities. Instead, other functions, such as transcriptional regulation, are more relevant to neural pathology since SON interacts with more than a thousand of genes via its DNA-binding region and is involved in the transcriptional repression of many target genes [9]. This is supported by the fact that rare nontruncating mutations, i.e., missense mutations [4] and an in-frame deletion [1], identi ed in ZTTK syndrome patients, are located exclusively in and around the genomic region that encodes the DNA-binding region (Table 1). However, there is a possibility that hSONm2 in uences SON-mediated RNA splicing because of its structural similarity to SON E, a physiological isoform of SON. This truncated isoform, which lacks an RNA-binding motif, has been reported to enhance full-length SON-mediated RNA splicing [9]. Therefore, it is possible that hSONm2 together with residual mouse full-length SON rescues RNA splicing de cits caused by SON insu ciency. Many more investigations are necessary to understand the detailed molecular mechanisms underlying the neural pathology of ZTTK syndrome, and accumulation of clinical and genetic information of ZTTK syndrome is also important.
In conclusion, this study revealed clearly that Son insu ciency results in neuronal migration defects and dendritic spine dysgenesis in the mouse brain. Since information about the neuropathology of ZTTK syndrome is extremely limited, these ndings provide important neuropathological basis possibly responsible for the neural symptoms, i.e., brain malformation and ID, of ZTTK syndrome. In addition, rescue experiments provided further evidence suggesting that putative neural abnormalities in ZTTK syndrome are caused by SON haploinsu ciency regardless of the residual functions of mutant SON genes.  One-way ANOVA followed by Dunnett's test was used for each statistical analysis. *p<0.05, n≥17 neurons