Usp9x-deficiency disrupts the morphological development of the postnatal hippocampal dentate gyrus

Within the adult mammalian brain, neurogenesis persists within two main discrete locations, the subventricular zone lining the lateral ventricles, and the hippocampal dentate gyrus. Neurogenesis within the adult dentate gyrus contributes to learning and memory, and deficiencies in neurogenesis have been linked to cognitive decline. Neural stem cells within the adult dentate gyrus reside within the subgranular zone (SGZ), and proteins intrinsic to stem cells, and factors within the niche microenvironment, are critical determinants for development and maintenance of this structure. Our understanding of the repertoire of these factors, however, remains limited. The deubiquitylating enzyme USP9X has recently emerged as a mediator of neural stem cell identity. Furthermore, mice lacking Usp9x exhibit a striking reduction in the overall size of the adult dentate gyrus. Here we reveal that the development of the postnatal SGZ is abnormal in mice lacking Usp9x. Usp9x conditional knockout mice exhibit a smaller hippocampus and shortened dentate gyrus blades from as early as P7. Moreover, the analysis of cellular populations within the dentate gyrus revealed reduced stem cell, neuroblast and neuronal numbers and abnormal neuroblast morphology. Collectively, these findings highlight the critical role played by USP9X in the normal morphological development of the postnatal dentate gyrus.

trafficking and endocytosis 7,8 , cellular polarity 9,10 and cell death 11,12 . USP9X has also been implicated in maintaining embryonic, haematopoetic and neural stem cell identity 13,14 . With respect to the nervous system, USP9X is expressed by neural stem cells in the developing cerebral cortex, as well as within the adult neurogenic niches 15 , and it promotes neural stem cell self-renewal in vitro 9 . In vivo, the conditional ablation of Usp9x from neural stem cells culminates in deficits in cortical architecture and abnormalities within the germinal ventricular zone of the developing cerebral cortex 15 . The most prominent phenotype of Usp9x-deficient mice is, however, the dramatic reduction in the size of the adult hippocampus 15 . The underlying cellular phenotypes contributing to this reduction in hippocampal size remain undefined.
Interestingly, the hippocampal phenotype of mice lacking Usp9x just prior to birth is subtle 9 , suggesting that postnatal hippocampal development is abnormal in these mice. The first three weeks after birth in rodents are critical for hippocampal morphogenesis, as this is the time during which the SGZ of the dentate gyrus resolves and matures 16 . Here we analysed postnatal and adult mice in which Usp9x had been removed specifically from neural progenitor cells within the dorsal telencephalon (via conditional ablation using an Emx1-Cre allele) to study the trajectory of postnatal hippocampal development in mice lacking this DUB. We reveal that hippocampal size, as well as the length of the blades of the dentate gyrus, is significantly reduced in comparison to control littermate mice from as early as postnatal day (P) 7. Furthermore, there were significantly fewer quiescent neural progenitor cells within Usp9x-deficent mice, as well as a dramatic reduction in the number of neuroblasts expressing the microtubule-associated protein doublecortin (DCX). Morphologically, those neuroblasts that were produced in mutant mice exhibited severe deficits in the extension and elaboration of dendritic processes. Collectively, these findings illustrate that abnormal production and/or maintenance of the neural progenitor cell population, coupled with reduced neuroblast numbers, underlie, at least in part, the reduced dentate gyrus size in mice lacking Usp9x, findings that are consistent with this DUB acting to maintain stem cell identity 9,13,14 .

Results
USP9X expression has previously been reported within the dentate gyrus and within the cornu ammonis (CA) regions of the adult mouse hippocampus 9,17 . Postnatally, we revealed a similar pattern of Usp9x expression. At P7 Usp9x mRNA was expressed strongly within the stratum pyramidale of the CA3 subfield, and was also observed within this area of the CA1 subfield, as well as within the hilus and blades of the dentate gyrus ( Fig. 1). A sense riboprobe did not reveal any specific staining (Fig. 1).
The subtle hippocampal phenotype of embryonic mice lacking Usp9x 15 , coupled with the postnatal expression of this gene (Fig. 1) and the severe reduction in hippocampal size in adult Usp9x −/Y ; Emx1-Cre mice 15 suggests that postnatal hippocampal development is aberrant in mice lacking this DUB. To test this hypothesis, we compared the hippocampal size of Usp9x-deficient and control mice at postnatal ages ranging from P7 to P56. Haematoxylin labeled sections revealed a significant reduction in hippocampal area in Usp9x −/Y ; Emx1-Cre mice from as early as P7 in comparison to control Usp9x loxP/Y mice ( Fig. 2A-L, Supp. Fig. 1). Indeed, the hippocampus of Usp9x −/Y ; Emx1-Cre mice was significantly smaller than controls at each of the ages analysed (Figs 2M and 3, Supp. Figs 1 and 2). Moreover, the size of the mutant hippocampus was at its greatest at P7, unlike controls, which increased in size at each age (Fig. 2M). Analysis of the combined length of the superior and inferior blades of the dentate gyrus revealed a similar trend, with pooled blade length being significantly reduced in comparison to controls (Fig. 2N), and not increasing markedly after P7. Analysis of the individual blades of the dentate gyrus also showed that both the superior and inferior blades were reduced in size at different rostro-caudal levels in the mutant in comparison to the controls postnatally and in the adult brain (Fig. 3, Supp. Figs 1 and 2). Given that the majority of the granule neurons within the dentate gyrus are produced postnatally 18 , these findings are illustrative of deficits in neurogenesis within the dentate gyrus contributing to the reduction in hippocampal size in Usp9x −/Y ; Emx1-Cre mice.
The first 3 postnatal weeks of development are critical for the development of the rodent dentate gyrus, as this is the time in which the neural stem cells coalesce into the SGZ and become quiescent 16 . Entry of stem cells into quiescence is important to ensure that this population of cells does not become prematurely depleted, so ensuring a substrate for neurogenesis throughout life 19 . Given the failure of the dentate gyrus blades to develop postnatally, we investigated neural progenitor cell number, as well as quiescence, using co-immunofluorescence labelling and confocal microscopy (a schematic of basic hippocampal structure, as well hippocampal cell types and specific markers that label these cells can be found in panels C-E respectively of Fig. 1). Progenitor cells were identified as those cells expressing both GFAP and SOX2, and proliferation was assessed using the marker Ki67. Although the total NSC number per unit area was not significantly different (data not shown), in total there were significantly fewer neural progenitor cells (GFAP + /SOX2 + ) within the dentate gyrus of Usp9x −/Y ; Emx1-Cre mice at each of the ages analysed in comparison to controls (Fig. 4). This is in line with the dramatic reduction in dentate gyrus size in mutant mice (Fig. 2). More interestingly, when we partitioned neural progenitor cells into quiescent (GFAP + / SOX2 + /Ki67 − ) and proliferating (GFAP + /SOX2 + /Ki67 + ) we saw a significantly lower proportion of quiescent cells, and a concomitantly higher proportion of proliferating cells within the dentate gyrus of Usp9x −/Y ; Emx1-Cre mice at each of the ages analysed (Fig. 4). From this we infer that there is a premature depletion of neural progenitor cells from the dentate gyrus of postnatal Usp9x −/Y ; Emx1-Cre mice, consistent with previous reports identifying Usp9x as a factor promoting 'stemness' 9,13,14 .
Proliferating neural progenitor cells within the dentate gyrus produce intermediate progenitor cells that are marked by the expression of proteins including TBR2 20,21 . Quantification of the number of intermediate progenitor cells within the dentate gyrus of Usp9x −/Y ; Emx1-Cre and Usp9x loxP/Y mice revealed that there were significantly fewer of these cells in the mutant compared to the control at each of the ages investigated (Fig. 5). Similarly, when we quantified the number of neuroblasts within the dentate gyrus (as evidenced by the expression of DCX) 22 , we observed that there were dramatically fewer neuroblasts within the mutant dentate gyrus at P28, P42 and P56 (Fig. 6A-E). Moreover, those neuroblasts that were produced in the absence of Usp9x exhibited an almost complete lack of dendrites, from which we infer that this DUB is critical for the extension of neuronal processes. We were unable to reliably count neuroblasts in the dentate gyrus of younger Usp9x loxP/Y mice due to the density of these cells and the extent of their dendritic processes. However, qPCR analysis revealed that there was a significant reduction in Dcx mRNA expression in the hippocampus of mutant mice at P14 (Fig. 6F).
The final step in neurogenesis within the postnatal and adult dentate gyrus is the migration of immature neurons into the overlying granule cell layer, and their maturation, a process which entails elaboration of dendritic processes and axonal pathfinding, as well as the expression of mature neuronal markers such as NeuN 20 . Given the reduction in the length of the blades of the dentate gyrus, we posited that there would be fewer NeuN-positive neurons within the granule cell layer of Usp9x −/Y ; Emx1-Cre mice in comparison to controls. At P7 we were unable to count the number of NeuN-expressing cells within the dentate gyrus of Usp9x loxP/Y mice, as the density of these cells was too high to preform reliable counts. However, at P14 and beyond we saw that there was a significant reduction in the number of mature neurons within the knockout dentate gyrus compared to the control ( Fig. 7A-E). Furthermore, the width of the NeuN-positive granule cell layer was also significantly reduced ( Fig. 7F), although we did not observe any significant differences in the density of NeuN-positive nuclei between mutant and control samples at any of the ages analysed (data not shown). Collectively, these findings, when considered in light of the diminished numbers of neural stem cells and intermediate progenitors, are consistent with deficits in neurogenesis within the postnatal dentate gyrus being responsible, at least in part, for the reduction in the size of this critical brain region in mice lacking Usp9x.

Discussion
The hippocampus is one of the few regions of the adult mammalian cerebral cortex in which neurogenesis persists throughout life 19 . This phenomenon is not confined to rodents, but has also been revealed, through the post mortem analysis of C 14 ratios, to occur in humans 23 . Neurogenesis is central to ongoing learning and memory, and abnormalities in this process have been linked to epilepsy and cognitive decline 24,25 . Recent work has begun to elucidate the mechanisms regulating the development and the maintenance of neural stem cells within the  dentate gyrus, and has revealed that factors intrinsic to neural stem cells are pivotal for their perdurance, as is the local niche microenvironment in which they reside 26 . However, far less is known of how the post-translational modification of proteins may influence adult neurogenesis. Here, we reveal that the DUB USP9X regulates multiple aspects of neurogenesis within the postnatal dentate gyrus, including regulating neural stem cell quiescence, as well as the production and morphological development of neuroblasts.
USP9X is expressed within the developing cerebral cortex 9 and within the adult dentate gyrus 17 . Whereas the loss of this gene results in subtle changes to neocortical size, it culminates in a markedly smaller hippocampus within adult mice 15 and from as early at P7 (this study). Indeed, the fact that deficits in dentate gyrus morphology are evident from as early as P7 indicates that a more rigorous analysis of the late embryonic phenotype of these mice is warranted. For instance, given that the infrapyramidal blade of the dentate gyrus develops after the suprapyramidal blade, a comprehensive analysis of blade development in younger mice may provide insights into the developmental timeline for the effects of Usp9x-deficiency. Conversely, over-expression of USP9X within cultured embryonic stem cell-derived neural progenitors promotes increased self-renewal capacity 9 . Collectively, these findings suggest that USP9X may regulate neural stem cell biology within the postnatal hippocampus. One way in which this could be mediated is via the regulation of cellular polarity. Hippocampal neural stem cells are highly polarised, with their cell bodies within the SGZ, and their processes extending into the granule cell layer 26 . Ablation of genes modulating polarity, such as Apc, results in premature exhaustion of the neurogenic niche within the dentate gyrus 27 , highlighting the importance of stem cell polarity for stem cell maintenance. Interestingly, USP9X promotes the polarisation of neural progenitor cells in vitro 9 , and co-localises with highly polarised proteins including β -catenin, p120 catenin, ZO-1 and E-cadherin in cultured intestinal cells [28][29][30] . The Par family of proteins is also central to apico-basal cellular polarity 31 , and the Par-1 homolog MARK4 is a substrate for USP9X activity 32 . Together, these findings support the notion that USP9X may play a role in the induction and/or maintenance of stem cell polarity with the nascent SGZ. Neural stem cell quiescence is also important to ensure that the stem cell niche does not become prematurely depleted within the adult brain 33 . Recent research has shown that signals emanating from the vasculature 34 as well as from the extracellular matrix 35 modulate cellular quiescence in adult neural stem cell niches, as does neuronal activity 36,37 . At a molecular level, numerous factors have also been shown to mediate neural stem cell quiescence, including members of the Notch signalling pathway 38,39 , the cyclin-dependent kinase inhibitor p27 kip1 40-42 , the chromatin remodelling factor CHD7 43 and the transcription factor NFIX 44 . Could USP9X also contribute to cellular quiescence in vivo? Evidence suggests that this is possible, as this DUB can potentially modulate Notch pathway activity through its interactions with elements of this pathway, including Epsin 45 , Itch 7 and mind bomb 1 46 . Moreover, UPS9X has been shown to interact with SOX2, a transcription factor localised to quiescent adult neural stem cells 20,47 . Future work aimed at deciphering how USP9X may mediate both polarity and quiescence in neural stem cells within the adult dentate gyrus will enable greater mechanistic insights into the in vivo role of this DUB. This work will be assisted through the use of Cre-drivers that will enable the ablation of Usp9x specifically   from adult neural progenitor cells, such as a nestin-Cre ERT2 allele 48 , coupled with proteomic approaches aimed at determining USP9X substrates and interactors within stem cell populations. The use of different Cre-drivers, such as CaMKIIα-Cre, which drives recombination within neurons of the forebrain, including the CA regions of the hippocampus 49 , will further elucidate the role of USP9X within the broader hippocampal formation. Indeed, Usp9x mRNA is very highly expressed within the stratum pyramidale of the postnatal hippocampus (Fig. 1), and the loss of Usp9x culminates in the entire hippocampus being reduced in size (Fig. 2). These data are suggestive of a role for USP9X in modulating the activity of post-mitotic neurons within the CA regions of the hippocampus, but whether this pertains to the development, maturation and/or survival of these neurons remains unresolved.
Our findings also illustrate that the morphology of neuroblasts within the postnatal hippocampus is severely abnormal in the absence of Usp9x. Why this occurs is unclear, but previous research has provided potential insights 6 . For instance, the elaboration of neuronal processes from cultured hippocampal neurons lacking Usp9x in vitro is reduced, and neocortical neuronal projections are abnormal in Usp9x −/Y ; Emx1-Cre mice. This is suggestive of a role for USP9X in mediating cytoskeletal dynamics and process outgrowth, a concept supported by findings revealing that USP9X interacts with DCX within the brain. The molecular logic underpinning this interaction, and how loss of USP9X disrupts cytoskeletal dynamics are at this point unknown, but work aimed at detailing the significance of the USP9X/DCX interaction will be critical to clarify this facet of USP9X function.
The implications of these data may also extend to humans, where mutations in USP9X are associated with neurodevelopmental disorders. Loss of USP9X function has been implicated in intellectual disability, autism 50 , epilepsy 51 and lissencephaly 52 . USP9X mutations associated with intellectual disability were unable to restore axonal growth and neuroblast migration when expressed in Usp9x-null hippocampal neurons derived from knock-out mice, whereas wild-type USP9X could 50 . Therefore, interrogating the molecular basis whereby USP9X regulates postnatal hippocampal development may provide insight into the normal development of both mouse and human brain.

Ethics statement. All mouse breeding was performed under the ethical clearance approved by Griffith
University Animal Ethics Committee (ESK 06 14 and BPS 02/11). The work performed in this study also conformed to The University of Queensland's Animal Welfare Unit guidelines for animal use in research (AEC approval number: QBI/353/13/NHMRC). The study followed the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes.
As Usp9x is located on the X chromosome, the male offspring from this cross that inherited the Emx1-Cre allele lacked Usp9x (hereafter referred to as Usp9x −/Y ; Emx1-Cre). Cre-negative males were used as controls (hereafter referred to as Usp9x loxP/Y ). Female mice were not used in this study. Genotyping was performed via PCR 15 . The day of birth was designated as postnatal (P) day 0. Mice were collected at P7, P14, P28, P42 and P56 (n ≥ 3 mice per genotype per age). Interestingly, adult Usp9x −/Y ; Emx1-Cre mice were slightly, but significantly, lighter than their control counterparts (Usp9x loxP/Y − 25.67 ± 0.58 g; Usp9x −/Y ; Emx1-Cre − 23.43 ± 0.54 g; p < 0.05, t-test), though the reason underlying this is unclear. All experiments were performed in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, and were carried out in accordance with The University of Queensland Institutional Biosafety committee.
Immunofluorescence labelling. Mice were perfused transcardially with phosphate-buffered saline (PBS), followed by 4% paraformaldehyde (PFA), then post-fixed for 48-72 hours before long term storage in PBS at 4 °C as described previously 53 . Brains were embedded in noble agar and sectioned in a coronal plane at 50 μ m using a vibratome. Sections were placed sequentially across the wells of a 6-well plate to ensure appropriate sampling from different rostro-caudal regions of the hippocampus. For the smallest brains (P7), there were 5 sections containing the hippocampus in each well. For the larger brains there were between 6-10 hippocampal sections per well. Thus, for all analyses we had at least 5 sections per animal to image and count, with each section also containing the left and right hippocampi. Sections were mounted on slides before heat-mediated antigen retrieval was performed in a 10 mM sodium-citrate solution (pH 6.0) at 95 °C for 15 minutes as described 54 . Sections were incubated for 2 hours in a blocking solution [2% vol/vol normal goat or donkey serum (Vector Laboratories) with 0.2% vol/vol Triton-X 100 in PBS] containing serum that matched the species in which the secondary antibody was raised. Primary antibodies diluted in blocking solution were applied to sections and incubated overnight at 4 °C. Following overnight incubation, sections were washed in PBS before the corresponding secondary antibodies were diluted in blocking solution and applied to sections for 2 hours. Sections were counter-stained with 4′ ,6-diamidino-2-phenylindole (DAPI) for 10 minutes and coverslipped using fluorescent mounting medium (DAKO).
Antibodies. The following primary antibodies, and the dilutions at which they were used in this study were: CAGAGAAGGT3′ . The fragment was cloned into the vector pGEM-T Easy (Promega), and anti-sense and sense riboprobes were generated using a DIG RNA labeling kit (Roche). In situ hybridisation was conducted as described previously 56 with minor modifications. Sections were subjected to heat-mediated antigen retrieval, and hybridisation was conducted at 70 °C. The colour reaction was performed with BM Purple (Roche).
Image acquisition and analysis. All brightfield images were captured using Axio Imager Z2 upright microscopes (Zeiss) attached to a digital camera (Zeiss AxioCam HRc). All fluorescent images were captured on an inverted spinning-disk confocal system using a 20 X objective (Axio Observer Z1 Carl Zeiss; CSU-W1 Yokogawa Corporation of America). We took 10 consecutive 1 μ m-thick optical sections to generate a 10 μ m-thick z-stack. In all cases the 10 μ m z-stack was taken from the middle of the section to minimize potential artefacts arising from the sectioning process such as damage to the tissue. Image acquisition was performed using SlideBook 6.0 (3I, Inc). Hippocampal area measurements and cell quantification were completed using Image J. To measure hippocampal area and the lengths of the dentate gyrus blades, images were taken of haematoxylin-stained sections, and quantification of area and length were performed in ImageJ. Measurements were performed on the left and right dentate gyri from both rostral, middle and caudal regions of the hippocampus. Hippocampal area, as well as the length of the dentate gyrus blades, was reduced within rostral, middle and caudal hippocampal sections of mutant mice in comparison to controls (Fig. 3, Supp. Figs 1 and 2). As such, data from different rostro-caudal sections of Usp9x loxP/Y and Usp9x −/Y ; Emx1-Cre mice were pooled to provide the mean values presented in Fig. 2.
For cell counts, the following cellular populations were analysed: quiescent NSCs (GFAP + /SOX2 + /Ki67 − ), proliferating NSCs (GFAP + /SOX2 + /Ki67 + ), IPCs (TBR2 + ) and neuroblasts (DCX + ). Cell counts were performed in both the left and right dentate gyrus for each section. To perform the cell counts, images were viewed in Image J. Immuno-positive cells in the first optical section were counted, and were marked using the counting tool in this program. The next optical section was then overlaid over the first, such that cells that had already been counted were still visible and marked with the counting tool. Counts were then performed on the second optical section. We then proceeded to do this with the remaining sections within the z-stack. In this way we were able to avoid counting cells twice. Cell counts per unit area of the dentate gyrus were also calculated (data not shown), but given the markedly reduced size of the mutant dentate gyrus, we chose to present the data as cell numbers within the z-stack per dentate gyrus as this was more reflective of the phenotype observed. We did not see any significant changes in the cell counts performed at different rostro-caudal levels in either control or knockout sections (data not shown). Mature neurons (NeuN + ) were quantified in a similar fashion, with the exception that the entire dentate gyrus was not analysed, due to the large number of immuno-positive cells. Instead, four sampling areas, each of which was 50 μ m wide running parallel to the SGZ, were used (two in the upper blade, and two in the lower blade, of the dentate gyrus). Supplementary Fig. 3 provides a schematic depicting the sampling areas selected. Counts were performed on dentate gyri from rostral, middle and caudal regions of the hippocampus, and were pooled to provide the mean values presented. The density of NeuN-positive nuclei was not significantly different between mutant and control cohorts at the ages analysed (data not shown). Samples were stained and imaged simultaneously, and cell counts were performed blind to the genotype of the sample. Quantitative real-time PCR. Quantitative real-time PCR (qPCR) was performed using standard protocols as described previously 57 . Briefly, the hippocampal formation was microdissected from the hippocampus of P14 Usp9x −/Y ; Emx1-Cre and Usp9x loxP/Y mice 58 and snap frozen. An RNeasy Micro Kit (Qiagen) was used to isolate total RNA. Superscript III (Invitrogen) was used to perform reverse transcription on total RNA (0.5 μ g) using random hexamers. qPCR reactions were carried out in a Rotor-Gene 3000 (Corbett Life Science) using the SYBR Green PCR Master Mix (Invitrogen). All of the samples were diluted to 1:100 with RNase/DNase-free water and 5 μ L of these dilutions were used for each SYBR Green qPCR reaction that also contained 10 μ L SYBR Green PCR Master Mix, 10 μ M of each primer and deionised water. The reactions were incubated for 10 minutes at 95 °C followed by 40 cycles with 15 seconds denaturation at 95 °C, 20 seconds annealing at 60 °C and 30 seconds extension at 72 °C. For this analysis, we isolated the hippocampi from 4 Usp9x loxP/Y mice, and 6 Usp9x −/Y ; Emx1-Cre mice. All the samples were tested in triplicate, and each experiment was repeated a minimum of three times. The house keeping gene glyceraldehyde 3-phosphate dehydrogenase was used to calculate relative changes in gene expression level, which is presented as − Δ Δ Ct. The primer sequences used were: Dcx For: 5′ TGGAAGCATGGATGAACTGG3′; Dcx Rev: 5′ CATGTTGGCAGATGTCTTTACG3′; Gapdh For: 5′ GCACAGTCAAGGCCGAGAAT3′; Gapdh Rev: 5′ GCCTTCTCCATGGTGGTGAA3′ .
Statistical analysis. Student's t-tests were used to compare the datasets described here. For all haematoxylin and immunocytochemical analyses, we used n = 3 animals per age, per genotype. For the qPCR analysis, we interrogated mRNA from 4 biological replicates for the Usp9x loxP/Y group, and 6 biological replicates for the Usp9x −/Y ; Emx1-Cre group. Statistical significance was established at a p-value of <0.05. Error bars represent standard error of the mean (SEM). Data analysis was performed blind to the genotype of the sample. The mean and SEM of all the data points reported in this study are provided in Supplementary Table 1.