The synaptic scaffolding protein CNKSR2 interacts with CYTH2 to mediate hippocampal granule cell development

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


Introduction
In humans, neurodevelopmental disorders present with a spectrum of diverse phenotypic characteristics such as intellectual disability (ID), cognitive deficits, learning and memory deficits, anxiety, and difficulties with social interaction (1). For ID syndromes such as Fragile X syndrome, Rett syndrome, and Down syndrome, phenotypic traits arise from genetic abnormalities that disrupt the development of multiple regions of the brain, including the hippocampus (2). Comprising the cornu ammonis (hippocampus proper) and the dentate gyrus, separated by the hippocampal sulcus, this brain region is critical for learning and memory. Of note, the development of the hippocampus commences prenatally but concludes postnatally, as evidenced by the finding that up to 85% of dentate granule neurons are born after birth in mice (3).
Presently, the molecular mechanisms of hippocampal neurogenesis and the impact of genetic mutations on these important developmental events in the mammalian brain remain to be better characterized. For example, the X chromosome contains genes associated with cognitive function, including FMR1 (4,5) and NLGN3 (6). Indeed, mutations in either of these genes lead to ID and autism (7)(8)(9)(10), yet numerous other genes remain poorly characterized, despite clinical genetic evidence implicating their essential functions in brain development and disease (11,12). One such gene is CNKSR2 (also known as CNKSR2), associated with ID and epilepsy (13)(14)(15)(16)(17)(18)(19).

J o u r n a l P r e -p r o o f Expression analyses of CNKSR2 in mouse brain
To investigate endogenous CNKSR2 protein, we produced a specific antibody which we confirm by the following analysis. First, we performed western blotting on lysates from cells transduced with a myc-CNKSR2 expression construct together with a control (non-targeting) shRNA vector or with a CNKSR2 targeting shRNA vector. As shown, immunoblotted myc-CNKSR2 signals that accord with the molecular size of approximately 130 kDa (indicated with an asterisk) are visible using our anti-CNKSR2 antibody and anti-myc antibody, but these respective signals are significantly diminished in lysates of cells co-transfected with shCNKSR2 (Fig. S1A). Second, following validation of our antibody, we performed western blotting on whole-brain extracts of mice from postnatal (P) days 0 to 30. Consistent with the molecular weights for signals validated in our specificity testing (Fig.   S1A), CNKSR2 immunoblotted signal of about 110 kDa was faintly detected at P0 and was more intense by P7 to P30, while a 130 kDa signal was weakly detected until P15 through to P30 (Fig. 3A).
We also investigated the expression of CYTH2 and detected two immunoblots of approximately 50 kDa that were prominent from P0 to P30 and a lower molecular weight band that was detected from P0 to P15 (Fig. 3A). These bands may be isoforms or products of post-translational modification such as phosphorylation (26-29). Further analyses should be required to address this issue. In order to substantiate their protein-protein interaction in vivo, we found that endogenous CYTH2 protein, which corresponds to the upper band, could be immunoprecipitated from protein lysates extracted from J o u r n a l P r e -p r o o f mouse cerebral cortices and hippocampal tissues with our anti-CNKSR2 antibody (Fig. 3B). In addition, given the specificity of our anti-CNKSR2 antibody (Fig. S1C), we performed immunohistochemical analyses using sections of a P7 mouse brain to find prominent CNKSR2 signal in the soma of pyramidal cells within the cornu ammonis, as well as in granule cells of the dentate gyrus ( Fig. 3C, a-d). Parallel studies with an anti-CYTH2 antibody enabled us to visualize cytosolic staining in pyramidal and dentate granule cells (Fig. 3C, e-h).

CNKSR2 and CYTH2 control the neonatal development of dentate granule cells
Given the expression of CNKSR2 and CYTH2 in the developing P7 hippocampus (Fig. 3C), we investigated the roles of these two factors in the development of dentate granule cells in neonatal mice.
To achieve this, we used an electroporation-based in vivo gene transfer method, which we previously established (25). This method is highly effective for the analysis of hippocampal cell differentiation and maturation. We previously took this approach to clarify the functions of the small GTPases, Rac1, Rac3, and Cdc42, and their roles in the localization and differentiation of dentate granule cells (30).
Our approach enables us to study neonatally-born dentate neurons and their migration and terminal differentiation ( Fig. 4A and see Methods for further details) (25). Specifically, we investigated the consequences of suppression of CNKSR2 using two shRNA targeting vectors (shCNKSR2#1 and shCNKSR2#2) on the development of P0 dentate granule cells. We began by validating their capacity J o u r n a l P r e -p r o o f for knockdown of exogenously-derived myc-CNKSR2 in transiently transfected COS7 cells. As shown, both shCNKSR2#1 and shCNKSR2#2, but not a non-targeting control shCont vector, efficiently suppressed immunodetectable myc-CNKSR2 signal (Fig. 4B, panel a). We also confirmed that these targeting shRNA vectors could suppress endogenous CNKSR2 in mouse neuroblastoma cell line Neuro2a cells (Fig. S1B). Next, we co-electroporated a GFP-expression vector (pCAG-GFP) with either shCNKSR2#1, shCNKSR2#2, or shCont into neonatal dentate granule precursors by in vivo electroporation in P0 brains, after which mice were allowed to recover before processing for analysis at P21 (25). As shown, the control treatment, in which cells were transduced with GFP and non-targeting (shCont) shRNA vector, GFP-positive cells were predominantly located within the granule cell layer (GCL), as deduced by quantifying labeled cells within the GCL, GCL/hilus junction and the hilus (Fig. 4B, panels b and e). In contrast, cells transfected with shCNKSR2#1 or #2 were frequently mislocalized at the GCL/hilus region or hilus (Fig. 4B, panels c-e). Furthermore, the mislocalization of dentate neurons following shCNKSR2 knockdown could be ameliorated by co-transfection with an expression construct for CNKSR2 that is resistant to shRNA-mediate knockdown (Fig. S2). Therefore, this result indicates that CNKSR2 is necessary and sufficient to influence the positioning of P0-born granule cells within the postnatal P21 hippocampus. To account for the possibility that knockdown affects the viability of electroporated cells, we quantified GFP-labeled cells in the dentate gyrus to find that the absolute numbers of labeled cells were not significantly different between the treatment of targeting shRNAs or non-targeting control vector ( Given that CYTH2 interacts with CNKSR2 in granule cells in vivo to form a stable complex, we investigated the impact of CYTH2 shRNA-mediated knockdown using a similar approach in P0-born granule cells. We designed two shRNA vectors targeting unique sequences, namely shCYTH2#1 and #2, that could efficiently knockdown steady-state levels of CYTH2 protein in transiently transfected cells (Fig. 4C, panel a). Following in vivo electroporation of neonatal dentate granule precursors in the P0 hippocampus with shCYTH2#1 or #2, in conjunction with the GFP-expression construct (pCAG-GFP) to label cells, we observed that shCYTH2-treated cells were abnormally localized to the GCL/hilus border or hilus in a distribution that was significantly different to control-treated (GFP and shCont) cells (Fig. 4C, panels b-e). This effect of shCYTH2 knockdown in granule cell positioning within the dentate gyrus is reminiscent of the effect of shCNKSR2 knockdown. We quantified GFP-labeled cells between treatments to find that the absolute number of labeled cells was not significantly different across shRNA treatments, indicating that knockdown with CYTH2-targeting shRNAs did not influence cell viability (Fig. 4C, panel f).
While the positioning of P0-born granule cells within the hippocampus at P21 was significantly influenced by knockdown with shCNKSR2 and shCYTH2 targeting vectors compared with control treatment, their mispositioning was unclear represented a delay or a defect in migration. To test this, we performed electroporation and then analyzed coronal sections of treated brains at P4, analyzing the J o u r n a l P r e -p r o o f distribution of transfected cells within the stream of migration, as described. As shown in Figures 5A and B, we observed that the distribution of GFP-labeled cells was not significantly different across shRNA treatments. Therefore, this result suggests that the mislocalization of P0-born granule cells within hippocampus at P21 may be explained by the fine-tuning of granule cell positioning within the GCL after P4 in the mouse.
Given that the terminal differentiation of GCL cells is critical to their cellular functions, we next analyzed the effects of CNKSR2 and CYTH2 knockdown on treated cells that co-express GFP. In the control (shCont) treatment, labeled cells exhibited a polarized shape with dendrites projecting to the molecular layer (Fig. 6A), a finding consistent with previous reports (25, 30). In contrast, however, cells transduced with shCNKSR2 and shCYTH2 vectors showed features consistent with a loss of neuronal polarity and abnormal neurite morphology ( Fig. 6A-C). Moreover, this abnormal morphology appeared to be more severe with shCNKSR2 treated cells ( Fig. 6B and C).
Next, we examined the effects of the knockdown of ARHGAP39 on the localization of P0-born granule cells. We did this because ARHGAP39 is a binding partner for CNKSR2, and the presence of ARHGAP39 could enhance CNKSR2 levels, based on our biochemical studies (see Fig. 1B). As shown in Figure S3, treatment with two unique shARHGAP39 vectors did not significantly change the placement of labeled cells in the GCL/hilus or hilus. However, treatment with shARHGAP39#1 significantly increased the proportion of labeled cells within the GCL.

The differentiation of neonatally-born dentate granule cells is influenced by knockdown of CNKSR2 and CYTH2 expression in vivo
During their development, dentate granule cells express stage-specific differentiation markers, including Prox1, which defines postmitotic dentate granule cells, and NeuN in immature and mature neurons and calbindin, detected in mature neurons dentate granule cells (31). We performed immunostaining for these markers on sections of brains transduced with non-targeting (shCont), shCNKSR2, or shCYTH2 vectors to examine potential effects of knockdown on these cellular features.
We found that for cells located within the GCL, the immunostaining signals for Prox1, NeuN, and calbindin were not significantly different in intensity, irrespective of treatment with control (shCont), shCNKSR2, or shCYTH2 vectors (data not shown). In contrast, however, when we examined immunostaining of these markers for cells localized to the GCL/hilus boundary or within the hilus, we observed a significant effect with CNKSR2 and CYTH2 knockdown, as follows: figure 7A-B shows that compared with control (shCont) treatment, treatment with shCNKSR2 and shCYTH2 knockdown vectors led to a significant decrease in the intensity of Prox1 immunostaining signal within cells, with a more severe effect observed for shCYTH2-treated cells. These results suggest that the knockdown of either gene has affected the maturation of these granule cells. We also quantified the signal for NeuN to observe a significant decrease in cells transduced with shCNKSR2 and shCYTH2 knockdown J o u r n a l P r e -p r o o f 13 vectors ( Fig. 7C-D). In the case of calbindin, a marker of mature granule neurons, we found that knockdown with shCNKSR2 or shCYTH2 shRNAs led to a decrease in signal shCNKSR2 treatment leading to a more severe effect. Taken together, the expression of CNKSR2 and CYTH2 is essential to the development of granule cells of the mouse hippocampus.

Discussion
In this study, we have investigated the biological significance of CNKSR2 and its binding partners within cells and their putative roles in developing neonatally-born dentate granule cells in mice. We found that its binding partners CYTH2 and ARHGAP39 can influence steady-state levels of

Preparation of extracts from mouse brain for western blotting
Brains were dissected from mice at various stages and prepared extracts as previously described (50).
Briefly, tissues were homogenized with the lysis buffer, sonicated at 0°C for 1 min, and centrifuged at 125,000 x g for 20 min at 4°C. The supernatants were used as whole tissue extracts. Protein concentration determination and western blotting were carried out as described above.
Fixed brains were embedded in paraffin and cut into sub serial coronal sections (6-µm thickness) at the level of the dorsal hippocampus. After deparaffinization, sections were treated with citric acid and then processed for immunohistochemistry as reported previously (48). Images were obtained using a

In vivo electroporation into the hippocampus of neonatal mice and immunostaining
In vivo electroporation into neonatal mice was performed with the previously published method (25).
In brief, 1 µl of DNA solution was injected into the lateral ventricle of postnatal day 0 (P0) mice.
Successfully injected pups were immediately electroporated with a tweezers-type electrode (CUY650P5, Nepa Gene, Chiba, Japan) using NEPA21 or CUY21 device (Nepa Gene). Five pulses of 100V were given of 50 msec duration with a 950 msec interval, and electroporated animals were returned to their dam. After indicated days, animals were transcardially perfused with 4% paraformaldehyde in PBS. Brains were then dissected and placed at 4°C for overnight in the same solution, washed with PBS, and mounted in 3% agarose in PBS. Sections (70 µm-or 100 µm-thickness) were cut using the HM 650V vibrating-blade microtome (Thermo Scientific).
Immunostaining was performed with free-floating sections (30). To quantify GFP-labeled cells in the dentate gyrus, we usually prepared serial brain sections containing dentate gyrus (100 µm-thickness), picked one slice from every three slices (six slices in total), stained for GFP, and quantified. For the quantitation of the localization of cells, we used z-stacked images captured by a fluorescent microscope (BZ-9000; Keyence). For the quantitation of the expression of differentiation markers, we used images captured by a confocal laser microscope (FV1000, Olympus, Tokyo, Japan). For the J o u r n a l P r e -p r o o f morphological analyses, we presented z-stacked images captured by FV1000. Image analyses were performed with ImageJ (US National Institutes of Health, Bethesda, MD, USA; https:// imagej.nih.gov/). As for the quantitation of the distribution of GFP-positive cells in brain slices at P4, GFP-signal intensity in distinct regions (bin 1-4) was measured by ImageJ software. Relative fluorescence intensities in each bin to total fluorescence intensities were calculated.

Statistical analysis
Statistical analyses were GraphPad PRISM (GraphPad Software, San Diego, CA). Comparisons between the two groups were performed by t-test (Fig. 1C). For other experiments, we analyzed data by one-way ANOVA with a post-hoc Dunnett's test (Fig. 1E, Fig. 4B-f and C-f and Fig. 7), or two-way ANOVA with a post-hoc Dunnett's test (Fig. 4B-e and C-e, Fig. 5B and Fig. S3) or Tukey test (Fig.   S2).
Data availability: All data are contained within the manuscript.