The expression and roles of Nde1 and Ndel1 in the adult mammalian central nervous system

Graphical abstract Nde1 (red) is localised to the GFAP-positive (green) stem cell of the subventricular zone of the lateral ventricle, and its overexpression in neural stem cells promotes neuronal differentiation while inhibiting astroglial differentiation.


INTRODUCTION
Nuclear distribution factor E (Nde1, or NudE) and nuclear distribution factor E-like (Ndel1, or Nudel) are ohnologous, evolutionarily conserved proteins with more than 50% amino acid conservation (Sweeney et al., 2001). They are coiled-coil proteins that can form a complex by binding lissencephaly-1 (LIS1) (Derewenda et al., 2007), and can also interact with dynein. They are essential for the function of microtubules which regulate cell mitosis and migration (McKenney et al., 2010). Nde1 and Ndel1 are both critical for cortical development. They regulate developmental neurogenesis and cortical neuronal positioning (Shu et al., 2004;Tsai et al., 2005Tsai et al., , 2007. For example, NDE1 mutations in humans can cause lissencephaly, a cortical developmental disease, and overt microcephaly (Alkuraya et al., 2011;Bakircioglu et al., 2011). Targeted deletion of Nde1 in mice is often embryonically or postnatally lethal and the mice have fewer cortical neurons and a much thinner cortex, reminiscent of lissencephaly. Mutation in Nde1 reduces the neural progenitor pool, dramatically inhibits later stage neuronal migration and blocks the development of the upper cortical layers (Feng et al., 2000;Feng and Walsh, 2004). Ndel1-null mutant mice are also embryonic lethal -Ndel1 À/À embryos die shortly after implantation. Ndel1 null/hypomorph compound heterozygotes (Ndel1 cko/À ) exhibit mild neuronal migration defects (Shu et al., 2004;Sasaki et al., 2005). In addition, Ndel1 can be phosphorylated by Aurora A kinase at position S251 or by the Cdk1/Cdk5 kinase at positions T219/ T245, which are important for the interaction with dynein (Zylkiewicz et al., 2011) and for the regulation of neuronal migration (Mori et al., 2007).
The investigation of copy number variation (CNV) of NDE1 and NDEL1 strongly suggests that they are associated with neuropsychiatric disorders. In genetic association studies, NDE1 has been identified as a strong candidate for epilepsy (Heinzen et al., 2010) and schizophrenia (Ekholm et al., 2003;Hennah et al., 2007;Ingason et al., 2011). The Disrupted in schizophrenia 1 (DISC1) protein is a common binding partner of NDE1 and NDEL1. Both NDEL1 and DISC1 are important genetic risk factors for the development of schizophrenia (Ozeki et al., 2003;Burdick et al., 2008), and they are all involved in regulating the migration of newly generated neurons in the adult hippocampus.  in stem cells in the hippocampal subgranular zone (SGZ) led to the formation of ectopic primary dendrites and mispositioning of the soma during neuronal maturation (Duan et al., 2007). Although Nde1 is involved in epilepsy, and adult hippocampal neurogenesis is up-regulated in rodent epilepsy models (Bengzon et al., 1997;Parent et al., 1997;Jessberger et al., 2007) and after antidepressant treatment (Malberg et al., 2000;Reif et al., 2006), the roles of Nde1 in these processes are less intensively investigated and therefore remains unclear.
Intriguingly, although there are several sporadic reports in the embryonic mouse and human cortex (Bakircioglu et al., 2011;Pawlisz and Feng, 2011), little information for the localization of Nde1 and Ndel1 in the developing and adult brains is available. It is essential to perform more wide and detailed investigations of the expression pattern of Nde1 and Ndel1 in the adult brain. For instance, its expression and dysfunction in astrocytes might play important roles in neurological disorders including epilepsy (Blumcke et al., 2002;de Lanerolle and Lee, 2005;Seifert and Steinhauser, 2013) and schizophrenia (Rothermundt et al., 2004). Although NDE1 and NDEL1 have been shown to be expressed in some human gliomas, their expression in normal astrocytes has not been well described.
In the present study, we have performed a comprehensive regional comparison of the expression profiles of Nde1 and Ndel1 in the adult mouse brain, with emphasis on the subventricular zone (SVZ) in the forebrain and the SGZ in the hippocampus, two regions where adult neural stem cells reside. We also analyzed the distribution and function of Nde1 and Ndel1 in primary neuronal/glia culture and a hippocampal neural stem cell line, HCN-A94. Our further functional studies suggest that overexpression of Nde1 and Ndel1 play important roles in cell fate specification, which potentially links with a range of neurodevelopmental and neuropsychiatric disorders.

EXPERIMENTAL PROCEDURES Animals and section preparation
All the animal work was approved by the Local Ethics Review Board of the University of Aberdeen. All animal procedures were performed with UK Home Office licence under the Animals (Scientific Procedures) Act 1986, with guidance of the Medical Research Facility, Aberdeen University, UK. Particular care was taken to minimize the number of animals to achieve statistical significance. Human brain tissue work in the present study was approved by the Ethics Committee of Universidade Metropolitana de Santos, SP, Brazil and by the Brazilian Health Research Committee on 4th of April 2011, under the number CONEP 16168, documents registered under the number 25000.169694/ 2010-18. Human brain tissue samples were collected from prefrontal, frontal and temporal cortices of adult male individuals aged 55 years or less at the Guilherme Alvaro Hospital, Santos, SP, Brazil, with written consent from the relatives of the deceased. None of the subjects presented a major neurological or psychiatric condition during life and the cause of death excluded head trauma, brain infection and stroke.
For immunohistochemistry, the brains of adult C57BL/ 6J mice were dissected out, embedded in Optimal Cutting Temperature compound (OCT) and snap-frozen on dry ice (n = 3). Serial coronal sections with thickness of 12 lm were yielded in a cryostat (Leica, CM1850, Germany) and mounted on the poly-lysine-coated slices (VWR, UK). These sections were stored in À80°C just before use.
To visualize the immunoreactivity, some sections were incubated with biotin-conjugated secondary antibodies (Sigma, UK) followed by Vectastain Elite avidin/biotin complex reagent (Vector Laboratories, UK) according to the manufacturer's instructions. After being washed with PBS, immunoreaction was visualized with 3,3 0 -diaminobenzidine (Sigma, UK). One of each series of the immunostained sections was counterstained with cresyl violet for the study of cytoarchitecture. Internal standards were employed to evaluate the intensity of immunostaining semi-quantitatively by two independent researchers without the knowledge of the experiment. The intensity of immunoreactivity was categorized as intense (+++), moderate (++), weak (+) and negative (À), which represented the relative expression levels among different regions and had no indication of the actual amount of the protein. The negative staining was performed with the same protocol except that the primary antibodies were replaced with normal IgG from the same animal species. The anatomical nomenclature of Paxinos and Watson (1986) was adopted to describe the regions of the CNS.
Immunofluorescence staining was also performed on 7-lm thick human prefrontal, frontal and temporal sections prepared from formalin-fixed and paraffinembedded blocks. Sections were de-waxed in histoclear (National Diagnostics, US) and rehydrated through decreasing ethanol concentrations (100%, 95%, 80% and 70%). Antigen retrieval was carried out by incubating sections in sodium citrate buffer (pH 6.0) and microwaving (800 W, 6 min) followed by incubation in icecold NaHBO 4 (0.5%, in 0.1 M pH 7.4 phosphate buffer, 10 min). Blocking and subsequent steps followed the same protocol as for mouse tissue fluorescence staining. Lipofuscin autofluorescence was quenched by applying Sudan Black-B (1% in 70% ethanol). The samples were then mounted in Aquamount (DAKO, Denmark).
The HCN-A94 cell line was a gift from Prof. Fred H. Gage (Salk Institute, US) and was maintained as described (Gage et al., 1995;Shen et al., 2011). DMEM medium with 0.5% fetal bovine serum (FBS), 1 lM transretinoic acid and N-2 (Invitrogen, US) were used for differentiation. To achieve in vitro overexpression, undifferentiated HCN-A94 cells were transfected with the pCMV-Nde1-EGFP, pCMV-Ndel1-EGFP or empty pCMV-EGFP vectors by using the rat neural stem cell Nucleofector kit (Lonza, UK). From the second day after transfection, the expression of constructs was maintained by applying selective antibiotic G418 (Sigma, 400 lg/ml) into the culture. The same dose of G418 also was used during the process of differentiation.

Primary culture and immunocytochemistry
Mouse (C57BL/6) cortices and hippocampi were dissected at the embryonic stage E18.5. Neuronal and glial primary culture was established as described previously (Viesselmann et al., 2011). Primary neuronal cells were maintained in Neurobasal-A medium (Invitrogen) supplemented with GlutaMax (Invitrogen, US) and B27 (Invitrogen, US) up to the time points of analysis (24 h, day 2, day 4, day 7). Primary astrocytes were plated on poly-L-lysine-coated glass coverslips (13 mm, VWR, UK) at a density of 3000 cells per ml and maintained in DMEM-F12 medium with 10% FBS. They were cultured for two weeks before analysis.
Both the HCN-A94 cell lines and the primary cultured neurons/astrocytes were fixed in paraformaldehyde (4%) for 5 min, permeabilized in À20°C methanol for 2-3 min and treated with blocking buffer (5% donkey serum, 0.3% Triton X-100 in PBS) at room temperature for 1 h. Immunofluorescence and negative staining were performed using the same protocol as mouse tissue staining.

Image analysis
To quantify neural and astrocytic differentiation ratio in over-expression experiments, fifteen microscopic fields (10Â objective) were picked randomly in each sample and the number of GFP single-labeled cells, GFAP/GFP double-labeled cells or Tuj1/GFP double-labeled cells were counted in each field. The proportion of differentiated astrocytes with GFP was calculated by the number of GFAP/GFP double-labeled cells against the total number of GFP single-labeled cells. The proportion of neuronal differentiation was presented by the ratio of the number of Tuj1/GFP double-labeled cells against the total number of GFP single-labeled cells. The number of total GFP-positive cells is larger than 10,000 in each group. The experiment was repeated three times. Results of Nde1-overexpressing group or Ndel1overexpressing group were compared independently with the pCMV-EGFP vector-transfected group (negative control).
To quantify the NDE1 expression ratio in astrocytes of white matter or gray matter in the human cortex, two different cortical sections were taken from each donor. Ten to fifteen microscopy fields (10Â objective) were taken randomly in gray matter or white matter and the number of NDE/GFAP double-labeled and GFAP singlelabeled cells was counted. For each section, the number of NDE/GFAP double-labeled cells was divided by the number of GFAP single-labeled cells. The results were then analyzed between gray matter and white matter by the Mann-Whitney U test. p < 0.05 was considered as significantly different. Strongest staining was also found in projecting nerves and the major white matter tracts. Negative controls showed only faint background staining after overdevelopment. I-VI, sublayers of cortex; cc in B3, corpus callosum; cc in A6 and B6, central canal; Cereb, cerebellum; CPu, caudate putamen; Ctx, cortex; cu, cuneate fasciculus; DG, dentate gyrus; MI, mitral cell layer; EPI, external plexiform layer; gr, gracile fasciculus; Hp, hippocampus; IPI, internal plexiform layer; LV, lateral Ventricle; OB, olfactory bulb; Puj, purkinje cell layer; py, pyramidal tract; SP, spinal cord. Scale bar = 100 lm.

Expression of Nde1 and Ndel1 in the adult mouse CNS
Although Nde1 and Ndel1 play important roles in neurodevelopment and neuropsychiatry, little information is available about their expression and distribution in the adult brain. We therefore performed immunohistochemical studies and investigated the detailed localization of Nde1 and Ndel1 in adult mouse brains. Nde1 and Ndel1 were found widely distributed in the CNS of adult mice. Their distribution patterns were similar but not identical (Fig. 1). Immunodouble-labeling showed that both proteins were largely co-expressed with the neuronal marker NeuN in CNS (Fig. 4C, D, I).
In the olfactory bulb, Nde1 and Ndel1 were strongly expressed within the internal/external plexiform layers, mitral cell layer and granular cell layer (Fig. 1A1, B1). They were also detected in most cortical neuronal cells (Fig. 1A2, B2), the lateral ventricle regions (especially for Nde1, Fig. 1A3) and in the hippocampus (Fig. 1A4,  B4). Nde1 and Ndel1 proteins were not detected to the same intensity in all brain regions. For example, in the hippocampus, Nde1 was present at high levels in CA1-3, the dentate gyrus granular cell layer (arrowed) and the polymorph layer. In contrast, though Ndel1 staining revealed a slightly higher expression in the polymorph layer, Ndel1 is only moderately expressed in CA1-3 and dentate gyrus (Fig. 1A4, B4, arrowed). In the cerebellum, somatic expression was most abundant in Purkinje neurons (Fig. 1A5, B5, arrows). Both proteins were broadly and strongly localized in white matter tracts, and projecting nerves, including in the spinal cord ( Fig. 1A6, B6). We further performed western blotting assays and also detected robust expression of Nde1 and Ndel1 in different brain sub-regions (Fig. 2). Two antibodies against different peptide sequences of Nde1 were used and the results were consistent. A detailed summary of the distribution of Nde1 and Ndel1 proteins in the mouse CNS is given in Table 1.

Distribution of Nde1 and Ndel1 in the adult mouse SVZ
Because Nde1 functions in stem cell mitosis (Feng and Walsh, 2004), we hypothesized that Nde1 may be expressed in the adult stem cells of SVZ in vivo. Two subtypes of SVZ neural progenitors have been identified so far: (1) infrequently or slowly dividing type B cells (putative stem cells) which are GFAP-positive and share some characteristics with radial glia; (2) type C active mitotic cells (Suh et al., 2009) which express Ki67 instead of GFAP. Our results showed that Nde1 (not Ndel1) was extensively expressed in the adult mouse SVZ (Fig. 3), and weakly co-localized with Ki67, a proliferation marker for actively dividing cells in the SVZ (Fig. 3). There was no co-localization of Nde1 and NeuN, markers for mature neurons, in the SVZ (Fig. 3C, D). In addition, some of the Nde1-stained cells had a unique radial process (Fig. 3D, white arrows) which may indicate astrocytic or stem cell morphology. We speculated that the Nde1-expressing cells may include type B stem cells. GFAP staining was carried out and the co-expression of GFAP/Nde1 was confirmed ( Fig. 3E-G, white arrows). These results suggest that Nde1 may be localized to SVZ-type B stem cells. In contrast, Ndel1 was barely detectable in the SVZ of the lateral ventricle (Fig. 3H, I), which suggested that Ndel1 is probably not expressed in SVZ adult stem cells at a significant level.

High levels of Nde1 in adult SGZ
We then examined the expression pattern of Nde1 and Ndel1 in the hippocampal SGZ, another niche where adult stem cells reside (Christian et al., 2010), and found an overlapping, but still distinct pattern (Fig. 4A, B, D-F). Nde1 was expressed in Ki67-positive cells in the SGZ (Fig. 4D). In contrast, there was no or only very weak co-localization of Ndel1 and Ki67 in the same brain region (Fig. 4E). Sox2, a stem cell marker, also was coexpressed in a subset of Nde1-positive cells in the SGZ ( Fig. 4F). Data suggest that Nde1 but not Ndel1 is expressed in actively dividing adult stem cells in the SGZ region.
To confirm the potential expression of Nde1 in SGZ adult stem cells, we performed immunocytochemistry and RT-PCR assays in HCN-A94 cells, an adult rat hippocampal neural stem cell line (Gage et al., 1995;Shen et al., 2011). High levels of Nde1 were found in undifferentiated HCN-A94 cells and intriguingly, these Nde1-expressing cells consistently showed co-immunostaining of Nestin, a specific marker for neural stem cells (Fig. 4G-I). In contrast, weak expression of Ndel1 was detected in these cells (Fig. 4J-L). Similar mRNA expression profile was detected by RT-PCR (Fig. 4C). We therefore concluded that Nde1 but not Ndel1 is abundantly expressed in adult hippocampal progenitor (stem) cells.

Nde1 overexpression decreases astrocytic differentiation of HCN-A94 cells
The different expression patterns of Nde1 and Ndel1 in the adult mouse hippocampus suggest that Nde1 and Ndel1 may have different functions during adult neurogenesis. In order to examine the precise functions of Nde1 and Ndel1 in adult hippocampal stem cells, and to determine how CNV of Nde1 and Ndel1 in the human population may underlie abnormal brain development, we have generated specific plasmids to over-express rat full length Nde1 or Ndel1 and performed the transfection experiments. Using the pCMV-EGFP, pCMV-Nde1-EGFP or pCMV-Ndel1-EGFP expression constructs described earlier, each gene was transiently overexpressed independently in undifferentiated HCN-A94 cells and protein up-regulation was detected by Immunostaining of the adult mouse hippocampus. Nde1 is highly expressed in both the polymorph layer and granular layer of the dentate gyrus (A); Ndel1 is more highly localized to the somata of some polymorph layer cells, but is also expressed in the granular layer of the dentate gyrus (B). (C) RT-PCR reveals more expression of Nde1 mRNA than Ndel1 mRNA in undifferentiated HCN-A94 cells. (D, E) In the SGZ, Ki67-labeled putative stem cells also express Nde1 (D) but not Ndel1 (E). (F) Double-labeling of Nde1 and Sox2 shows that Sox2-labeled adult stem cells also express Nde1. (G-L) Immunocytochemistry of undifferentiated HCN-A94 cells revealed high expression of Nde1 (G-I) and low expression of Ndel1 (J-L). PoDG, polymorph layer, dentate gyrus; GrDG, granular layer of the dentate gyrus; CtlÀ, negative control; Scale bar = 20 lm. ++ +++ S1 (somatosensory cortex) ++ ++ S1BF (somatosensory1, barrel field) ++ ++ S1DZ (primary somatosensory cortex, dysgranular zone) ++ ++ S1DZO(primary somatosensory cortex, oral dysgranular zone) ++ ++ S1HL (primary somatosensory cortex, oral dysgranular zone) ++ ++ S1J (somatosensory1, jaw reg) ++ ++ S1Tr (somatosensory1, trunk reg) ++ ++  (Fig. 5). Nde1 overexpression increased Nde1 protein levels as expected (Fig. 5A-C) but did not affect the expression of Ndel1 (Fig. 5D-F).
The same pattern was also demonstrated in Ndel1overexpressing cells (Fig. 5G-L). None of the empty EGFP-vector transfected cells showed a significant increase in Nde1 or Ndel1 proteins (Fig. 6). Further confocal analyses of differentiated HCN-A94 cells have also demonstrated the specificity of the antibodies we   have used (Fig. 7). To determine the roles of Nde1 and Ndel1 during adult stem cell differentiation, the transfected HCN cells were differentiated for 6 days in serum-containing medium as described before, and then immunostained with anti b-tubulin III (Tuj1) (Fig. 8A, C, E) and anti-GFAP (Fig. 8B, D, F). The percentage of Tuj1/GFP-positive and GFAP/GFP-positive was measured in transfected cells. Compared to EGFP vector-only transfected cells, Nde1-overexpression significantly decreased the number of differentiated astrocytes (P < 0.001), whereas the number of differentiated neurons was increased slightly (Fig. 8G,  H). A smaller, but still significant, decrease in the number of astrocyte was observed also in Ndel1overexpressing cells (P < 0.05). These results strongly suggest that the Nde1 and Ndel1 protein levels modulate the neuron/astrocyte differentiation decision in differentiating HCN-A94 cells and Nde1, which shows high levels of expression in hippocampal stem cells, has a more predominant effect than Ndel1. It should be noted that our immunohistochemistry indicates that the level of fusion protein produced from the plasmids is far greater than that normally found in un-transfected HCN-A94 cells, possibly by an order of magnitude. It is prudent therefore to be cautious about extrapolating these supranormal dosages to the mechanism of human neurological disease.

Subcellular distribution of Nde1 and Ndel1 in cultured cells
We next examined the detailed subcellular localization of Nde1 and Ndel1 in cultured cells. In primary cultured neurons, we observed similar subcellular localization of Nde1 (Figs. 9A,B and 10E,F) (Bradshaw et al., 2009(Bradshaw et al., , 2011. While Nde1 was extensively located in the cytoplasm and nuclear or perinuclear areas, Ndel1 was present significantly in axonal and dendritic extensions as well as growth cones (Fig. 9C, D). Similar distribution patterns were also found in neuronaldifferentiated HCN-A94 cells (Fig. 10A-F).

as in previous reports
Astrocytes have critical roles in brain function and tissue homeostasis and abnormalities in their function are linked to disorders including schizophrenia and epilepsy. Interestingly, subcellular localization of Nde1 and Ndel1 in astrocytes has not been reported before. In light of the evidence presented above that Nde1 dosage may affect astrocytic differentiation, we performed this examination by immunocytochemistry and detected the expression of Nde1 and Ndel1 in primary cultured astrocytes generated from the cortex of E18.5 mice (Fig. 11A-D). Nde1 and Ndel1 were found to have different subcellular localizations in astrocytes, similar to those observed in primary cultured neurons. While Nde1 was highly expressed in nuclei and also the perinuclear region (Fig. 11A, B, white arrows), Ndel1 was mostly distributed in the cytoplasm, especially in filopodial processes of astrocytes (Fig. 11C, D, white arrows).
These localization patterns broadly recapitulated their subcellular localization in neurons. Meanwhile, Nde1 and Ndel1 were also found in astrocytic HCN-A94 cells after differentiation . It was demonstrated that Nde1 showed cytoplasmic localization primarily in vimentin-positive cells (Fig. 11I, J), a cell marker for immature astrocytes. Nde1 protein level and localization therefore may change during glial differentiation. Our results suggest that both Nde1 and Ndel1 exist in multiple astrocytic morphological subtypes, but their subcellular distributions do not overlap extensively.

Neuronal and astrocytic distributions of NDE1 in the adult human cortex
Surprisingly, there is almost no histological description of the localization of NDE1 or NDEL1 in the adult human brain. To address this question, immunohistochemical staining was performed on human cortical sections. Although the Ndel1 antibody did not produce any specific staining in these sections, we successfully detected NDE1 in the prefrontal, frontal and temporal lobes of the adult human cortex. NDE1 protein was distributed in both cortical neurons (Fig. 12A-C) and astrocytes ( Fig. 12D-G). In all cortical layers, NDE1 was distributed broadly in the majority of NeuN-labeled mature neurons, while some GFAP-expressing astrocytes expressed NDE1 (Fig. 12E, G, white arrows). Interestingly, we found that gray matter of human cortex contained more NDE1-positive astrocytes than white matter (Fig. 12H). Because most gray matter astrocytes are protoplasmic while most of the white matter astrocytes are of the fibrous type, our results suggest that NDE1 is heterogeneously expressed in different astrocyte subtypes. This heterogeneous NDE1 localization was consistently found in the cortical regions of all four donors. qPCR experiments were then carried out to compare the mRNA expression abundance of NDE1 and NDEL1 in different human cortical regions. The assay included samples from the prefrontal, frontal and temporal cortices. After normalization against controls, no significant variation was demonstrated among these regions for either NDE1 or NDEL1 (Fig. 11I), suggesting that both NDE1 and NDEL1 are widely and homogenously expressed in the human cortex. Our results convincingly show that NDE1 protein is distributed widely in the human cortex, as expected, but is localized heterogeneously in astrocytes in gray and white matters.

DISCUSSION
Neurological diseases such as schizophrenia and epilepsy are highly complex disorders. Their etiology and its relationship with therapeutics used for treatment are still poorly understood. NDE1 in particular is genetically associated with both diseases. In this study, Nde1 and Ndel1 protein distribution and function have been compared in both neurons and astrocytes, to address some outstanding questions about their cellular and subcellular localization in the mouse and human CNS, and to investigate possible roles in modulating the differentiation of neurons and glia from neural stem cells.
Nde1 and Ndel1 are essential for neuronal development and are important cell cycle regulators (Mori et al., 2007;Kim et al., 2011). Most studies have is also distributed in cortical astrocytes. In gray matter, subsets of GFAP-expressing astrocytes are labeled with NDE1 (E, white arrows). (F, G) GFAP and NDE1 double-labeled cells also were detectable in cortical white matter. (H) The proportion of NDE1-expressing astrocytes was significantly greater in gray than in white matter of the human cortex (10 samples from five donors; Mann-Whitney U test (P < 0.02)). (I) qPCR experiments also revealed the mRNA expression of NDE1 and NDEL1 in the three different cortical regions (frontal, prefrontal and temporal lobe) with no significant variations (n = 3). Scale bar = 20 lm.
focused on their functions during brain development, including the proliferation of neural precursor cells and neuronal migration (Bengzon et al., 1997;Sasaki et al., 2005;Taya et al., 2007;Kim et al., 2011). Nde1 deletion leads to cortical developmental defects (Feng and Walsh, 2004;Pawlisz et al., 2008), and these abnormalities are mostly correlated with defective embryonic neurogenesis. Here, we have shown that both Nde1 and Ndel1 are expressed broadly in adult mouse and human brains, including the majority of cortical neurons. This expression pattern of both proteins suggests that they have important functions in mature neurons. Additionally, although Nde1 regulates radial glial functions during development (Bengzon et al., 1997) and its function is pivotal for human glioma formation (Suzuki et al., 2007), the astrocytic distribution of Nde1 and Ndel1 has not been investigated extensively. We have demonstrated that Nde1 and Ndel1 are expressed in multiple subtypes of adult astrocytes with different morphological characteristics both in mice and humans. Our in vitro studies further showed that Nde1 and Ndel1 have similar subcellular localization in astrocytes which is comparable to their distribution pattern in neurons.
Based on the multi-compartmental existence of Nde1 and Ndel1 in astrocytes, together with their important roles during cell division and spindle orientation (Feng and Walsh, 2004), we speculate that they may be critical for brain homeostasis and therefore contribute to a range of disease states. Interestingly, NDE1 and NDEL1 have been found in gliomas and their expression is highly associated with the activity of glioma cell migration and proliferation (Suzuki et al., 2007). We also found a heterogeneous expression of NDE1 in the gray matter and white matter of the human brain, and also in different sub-layers of the human cortex. These results suggest that NDE1 is expressed selectively in certain populations of adult cortical astrocytes, although the properties of these astrocytic subtypes require further study.
Intriguingly, NDE1 CNV has been classified as a potential risk factor for a series of neuronal disorders including schizophrenia and epilepsy. Ndel1 regulates aspects of embryonic neuronal migration and neurogenesis and cooperates with its binding partner DISC1 in regulating adult neurogenesis. Knockdown of Ndel1 in the SGZ leads to abnormal soma size and migration defects (Parent et al., 1997). In contrast, the function of Nde1 in adult neurogenesis had not been studied. We found that Nde1 rather than Ndel1 was highly expressed in GFAP-positive SVZ cells with radial-like processes. This suggests Nde1 is present in SVZ-type B stem cells. In the SGZ, Nde1, but not Ndel1, was co-localized with Ki67 in actively dividing neural stem cells. These data suggest that Nde1 is more likely than Ndel1 to be involved in adult neural stem cell function. HCN-A94 cells can be differentiated into a mixed cell population containing neurons and astrocytes. In this study, Nde1 overexpression resulted in the decreased ability of HCN-A94 cells to differentiate into astrocytes compared to the control. This suggests that Nde1 and Ndel1 may participate in cell fate determination during adult hippocampal stem cell differentiation and potentially links with observations of CNV of the genes underlying human disease. Our findings are supported by previous experimental data suggesting that Nde1 has an important regulatory function in the dividing cells. This is further evidenced by the small cerebral cortex phenotype observed in Nde1 À/À mice, which was apparently caused by mitotic spindle defects (Feng and Walsh, 2004). Our results suggest that it might be important in future to investigate whether Nde1-overexpressing neural stem cells have mitotic spindle defects and how this alteration modulates their mitotic delay or cell fate specification.

CONCLUSIONS
We have provided the most comprehensive documentation of the expression patterns of Nde1 and Ndel1 in cultured cells as well as in mouse and human brains, and also highlight that dosage effects of these two proteins might contribute to some cases of mental disorder.