WDR81 regulates adult hippocampal neurogenesis through endosomal SARA-TGFβ signaling

Adult hippocampal neurogenesis, a process considered important for hippocampal function, is regulated at multiple molecular levels. Mutations in the gene encoding the WD40 repeat-containing protein WDR81 are associated with neurological disorders, including cerebellar ataxia, mental retardation, quadrupedal locomotion syndrome (CAMRQ2), and microcephaly. In this study, we show that ablation of WDR81 in adult neural progenitor cells (aNPCs) markedly reduced adult hippocampal neurogenesis and impaired hippocampus-dependent learning. WDR81 suppresses endosomal PtdIns3P synthesis, likely by inhibiting the assembly of the PI3K-III complex. In the absence of WDR81, endosomal PtdIns3P levels are greatly elevated, leading to endosomal persistence of the PtdIns3P-binding protein SARA and consequently hyperactivation of SARA-dependent TGFβ signaling. Inhibition of PI3K-III activity or suppression of SARA-dependent TGFβ signaling markedly ameliorated the defective adult neurogenesis in WDR81-deficient mice. Taken together, these findings not only uncover the requirement for the WDR81–SARA–TGFβ axis in adult hippocampal neurogenesis, but also suggest that defective adult hippocampal neurogenesis contributes to the etiology of WDR81-related neurological diseases.


Introduction
Normal brain function requires the precise control of neurogenesis, which is tightly regulated by combinatorial functions of extrinsic signals and intrinsic programs.
Pathological disturbance of this process contributes to the symptomatology of many neurological disorders. In the adult mammalian brain, newborn granular neurons are continuously integrated into hippocampal circuitry throughout life, and the fine-tuning of this process is important for hippocampal function [1,2]. Therefore, understanding the molecular mechanisms that regulate adult neurogenesis will not only shed light on the processes that govern the functional integrity of the adult brain, but will also provide valuable insights into neurological disorders.
The precise regulation of adult neural stem cell proliferation and differentiation requires the coordinated action of a variety of signaling molecules (e.g., receptors and ligands). Through endosomal trafficking, such signaling molecules are distributed in a spatio-temporal manner, thereby determining cell fate, neuronal polarity, migration, and axonal outgrowth and guidance [3,4]. Endosome-associated signaling receptors can either be delivered to lysosomes for degradation and signal termination, or recycled back to the plasma membrane in a tightly regulated way. These events are primarily controlled by endosomal Rab GTPases and phosphoinositides [5][6][7]. The early endosome-specific phosphatidylinositol 3-phosphate (PtdIns3P) plays important roles in the endosome-lysosome pathway [7,8]. PtdIns3P is generated by the These authors contributed equally: Min Wang and Changyong Tang * Chonglin Yang clyang@ynu.edu.cn * Weixiang Guo wxguo@genetics.ac.cn class III PI3K (PI3K-III) complex consisting of VPS34, VPS15,and Beclin1 [5,9]. PtdIns3P can be converted to the late endosome-specific PtdIns(3,5)P 2 by the FYVE finger-containing phosphoinositide kinase PIKfyve, dephosphorylated by myotubularin family phosphatases, or degraded in the endosomal lumen [5][6][7]. Appropriate levels of endosomal PtdIns3P are crucial for endosomal trafficking, disruption of which is implicated in several human neurological disorders [10,11]. Nevertheless, it remains largely unknown how PtdIns3P-dependent endosomal trafficking plays a role in adult neurogenesis.
Recently, we and others revealed that WDR91 and WDR81, two WD40 repeat-containing proteins, act in complex with Beclin1, a subunit of the PI3K-III complex, to inhibit endosomal PtdIns3P synthesis, thereby facilitating early-to-late endosome conversion in the endosome-lysosome pathway [12,13]. We further demonstrated that WDR91 is a Rab7 effector that is recruited to endosomes by GTP-bound Rab7, and its brain-specific depletion leads to defective dendritic arborization and reduced brain size in mice [14]. In addition, we found that brain-specific Wdr81 knockout mice died quickly after birth, with accumulated p62 bodies in the cortical and striatal neurons [15]. Importantly, WDR81 mutations were initially reported in a consanguineous family suffering from cerebellar ataxia, mental retardation, and quadrupedal locomotion syndrome (CAMRQ2) [16], and later in patients with microcephaly [17]. However, it is unclear whether and how WDR81 regulates hippocampal neurogenesis in the adult mammalian brain.
In the present study, we investigated the requirement for WDR81 in adult hippocampal neurogenesis. We found that specific depletion of WDR81 in adult neural progenitor cells (aNPCs) reduced adult hippocampal neurogenesis and impaired hippocampus-dependent learning. Mechanistically, we demonstrated that WDR81 inhibits the activity of PI3K-III, possibly by preventing the assembly of the PI3K-III complex, leading to PtdIns3P-dependent endosomal persistence of SARA, a PtdIns3P-binding protein in the TGFβ signaling pathway. Endosomal persistence of SARA led to hyperactivation of SARA-dependent TGFβ signaling, which in turn suppressed adult hippocampal neurogenesis. These findings unveil a regulatory role for WDR81 and endosomal PtdIns3P in adult hippocampal neurogenesis.

Mice
Mice were housed in the animal facility at the Institute of Genetics and Developmental Biology (IGDB), Chinese Academy of Sciences, on a 12-h reverse light/dark cycle with lights on at 0800. All procedures and husbandry were performed according to protocols approved by the Institutional Animal Care and Use Committee at IGDB. WDR81 f/f mice were bred onto the C57BL/6 background as described previously [15]. The inducible conditional Wdr81 knockout mice were generated by crossing Nestin-CreER T2 [18] driver with Rosa26-CAG-DTR-EGFP (BCG-T0-0001, BIO-CYTOGEN). All mice in the study were backcrossed to the C57BL/6 background for at least six generations. To induce recombination, mice (8-10 weeks old) received Tamoxifen (Sigma-Aldrich, T5648) daily for 5 days (180 mg/kg intraperitoneal injection, 30 mg/ml in 10% EtOH/sunflower oil, Sigma-Aldrich) based on a published procedure [19].

Cell lines and transfection
Wdr81 WT and KO Hela cells were cultured as described previously [12]. For plasmid DNA transfections, 2 μg of plasmid DNA and/or 2 μg of shRNA were used and transfection was performed with Lipofectamine 2000 (Invitrogen, 11668019) according to the manufacturer's instructions.
For deletion of the Wdr81 gene in aNPCs, lentiviruses expressing Cre-GFP were added twice (once per day) under proliferation conditions for 2 days before the initiation of proliferation or differentiation assays. Lentiviruses expressing dCre-GFP were used as control. Cre-induced recombination was confirmed by using both western blotting and immunocytochemistry.
The numbers of marker-positive cells (BrdU + , Tuj1 + , or GFAP + , and cleaved caspase-3 + ), as well as total Cre-GFPinfected cells (GFP + ) were quantified using a Nikon-ECLIPSE 80i microscope with NIS-Elements, BR. 3.00 software. The percentage of differentiated cells was calculated as the number of marker-positive cells divided by the total number of DAPI-or GFP-positive cells.

Recombinant lentivirus production
Lentivirus production was performed as described previously [21]. Briefly, lentiviral transfer vector DNA and packaging plasmid DNA were co-transfected into 293 T cells. The medium was collected and pooled at 40, 64, and 88 h, and then filtered through a 0.2-μm filter. Viruses were concentrated by ultracentrifuge at 19,000 rpm for 2 h at 20°C using a SW27 rotor (Beckman). The virus particles were washed once with phosphate-buffered saline (PBS) and then resuspended in 150 μl of PBS.

Recombinant retrovirus production and in vivo grafting
Retrovirus production was performed as described previously [22]. Briefly, viral transfer vector DNA and packaging plasmid DNA were transfected into cultured 293T cells using calcium phosphate methods. The medium containing lentivirus was collected at 40, 64, and 88 h post transfection, pooled, filtered through a 0.2-µm filter, and concentrated using an ultracentrifuge at 19,000 rpm for 2 h at 20°C using a SW27 rotor (Beckman). The virus particles were washed once and then resuspended in 50 µl of PBS. We routinely obtained 1 × 10 9 infectious viral particles /ml.

3-MA and LY2109761 administration
For cell treatment, 3-MA (5 mmol/ml in PBS) was added to aNPCs, which were then subjected to proliferation and differentiation assay.
Systemic administration of 3-MA (15 mg/kg in PBS) or LY2109761 (20 mg/kg in PBS) was performed by intraperitoneal injection into adult mice every other day for 28 days. For cell proliferation assays, the mice also received one BrdU injection (200 mg/kg), and were killed at 2 h after BrdU injection. For cell differentiation assays, the mice also received BrdU injections (50 mg/kg), twice per day for two days, and were killed at 4 weeks after the final BrdU injection.

RNA isolation and real-time PCR
RNA isolation used Trizol (Invitrogen, 15596018) based on the manufacturer's protocol. The first-strand cDNA was generated by reverse transcription with oligo (dT) primer or random hexamers (Promega, A5001). Standard RT-PCR was performed using GoTaq DNA polymerase (Promega, M3001). To quantify the mRNA levels using real-time PCR, aliquots of first-strand cDNA were amplified with gene-specific primers and SYBR Green PCR Master Mix In vitro PI3K complex activity assay The in vitro PI3K complex activity assay was performed as described previously [12]. aNPCs were harvested and lysed with lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, and protease inhibitor cocktail). Endogenous Vps34 was immunoprecipitated by antibody against Beclin1 (MBL, PD017). Immunoprecipitated beads were extensively washed with lysis buffer and further washed twice with reaction buffer (40 mM Tris-HCl pH 7.5, 20 mM MgCl 2 , and 1 mg/ml BSA). The beads were then incubated with 10 μg of sonicated phosphatidylinositol (Sigma, 79403) and 1 μl of ATP (10 mM) in 30 μl of reaction buffer for 30 min at room temperature. Conversion of ATP to ADP was measured with an ADP-GloTM Kinase Assay Kit by following the instructions provided by the manufacturer (Promega, V6930).

ELISA analysis of PtdIns3P
Cell lysates were collected, and the amount of PtdIns3P was determined using a PtdIns3P ELISA kit (Echelon, K-3300) according to the manufacturer's instructions.

Tissue preparation and immunohistochemistry
Immunohistochemistry was performed as described previously [19,21]. At 1, 7, 14, 28, or 56 days after the last TAM injection, mice were euthanized by intraperitoneal injection of Avertin and then transcardially perfused with saline followed by 4% PFA. Brains were dissected out, post-fixed overnight in 4% PFA, and then equilibrated in 30% sucrose. Forty micrometer brain sections were generated using a sliding microtome and stored in a −20 o C freezer as floating sections in 96-well plates filled with cryoprotectant solution (glycerol, ethylene glycol, and 0.1 M phosphate buffer, pH 7.4, 1:1:2 by volume).
The tissue sections were pre-blocked with TBS++ (TBS containing 3% goat or donkey serum and 0.3% Triton X-100) for 1 h at room temperature, followed by incubation with primary antibodies diluted in TBS++ overnight at 4 o C. After washing three times, sections were incubated with secondary antibodies for 1 h at room temperature. All sections were counterstained with a nuclear counter stain, DAPI.

Quantification and fate mapping of GFP + cells in the DG
For quantification of GFP + cells, 1 in 12 serial sections starting at the beginning of the hippocampus (relative to bregma, −1.5 mm) to the end of the hippocampus (relative to bregma, −3.5 mm) were used [19]. Quantification of GFP + cells and phenotypic quantification of GFP + cells in the granule layer were performed using a Nikon-ECLIPSE 80i microscope with NIS-Elements, BR. 3.00 software.

The fear conditioning test
The fear conditioning test was performed as described with modification [23]. Mice were placed into a shock chamber and allowed to explore for 2 min. Then, a white noise tone (87 dB) sounded for 30 s (conditional stimulus or "CS"). During the last 1.5 s of the tone, mice received a mild footshock (0.5 mA) (unconditioned stimulus or "US"). 2 min later, the same tone-footshock (CS-US) combination was delivered again. This cycle was presented a total of three times with a 60-s interval. The context test was performed 24 h after the training. During the test, mice were placed back into the same training chamber, and monitored by an overhead camera in the chamber for 5 min. No stimuli were applied. Two-hours after the context test, the cue test was performed, in which colored plexiglass inserts were placed into the training chamber to hide the shock grid and to change the "context" of the chamber. Mice were then placed in the chamber and monitored by the overhead camera for 6 min, during which two CS (spaced the same way as in the training session) were given. In the fear conditioning test, all events were programmed and all data were recorded through the Startle and Fear conditioning system (Panlab) and Packwin software (V2.0.05).

Novel object location test
The object recognition test was performed as described with modification [24]. Mice were habituated to an empty white chamber by allowing them to freely explore for 15 min. After 24 h, each mouse was rehabituated to the empty chamber for 1 min and then placed in a holding cage, whereas two identical objects were placed in the corners of the arena 7 cm from the walls. Mice were returned to the chamber for training and allowed to freely explore until they accumulated a total of 30 s exploring the objects (exploration recorded when the front paws or nose contacted the object). Mice were then removed from the chamber, immediately infused, and returned to their home cage. After 24 h, object recognition was tested by moving one of the objects to a novel location counterbalanced across mice. The mice were reintroduced into the chamber and recorded for 10 min via digital video. The time spent exploring each object in the novel location tests was measured by watching the recorded behavior. Scoring was done by using two stopwatches and collecting the exploratory time for each object simultaneously. Novel location preference is expressed as the percentage of time exploring the novel location out of the cumulative time spent exploring both objects.

Open-field assay
Mice were placed in an unfamiliar arena with clear side walls (10 × 10 × 16 inch; RWD life Science) and were allowed to freely explore the arena for 20 min. They were returned to their home cages after the test. Their locomotor activity was tracked by photo beams preinstalled in the arena and then analyzed by Panlab SMART 3.0 Software.

Dark/light exploration assay
The test was conducted in a two-chamber shuttle box with an opaque divider in the middle (RWD Life Science). Each chamber measured 17 × 17 × 33 cm. An opening measuring 6 × 6 cm was located at the center bottom of the divider. The walls of one chamber were made of black plexiglass (dark chamber) while those of the other chamber were transparent (light chamber). The ceiling of the box was made of a sheet of aluminum with a round hole measuring 1 cm in diameter over the center of the light chamber. Mice were habituated in the room for 1 h before the test with the room lights off. During the test, the room was not illuminated but the light compartment was lit by a 60 W bulb closely placed over the hole in the ceiling. During the test, a mouse was first placed into the dark chamber and allowed to freely travel between the chambers for 5 min. The time spent in the light chamber was scored.

Statistical analysis
All experiments were performed and analyzed by the same experimenter, blind to the animals' genotype or group treatment under assessment. Animals were assigned to groups according to their genotypes, and no randomization was applied. Sample sizes were determined by power analyses based on a previously published study [25]. Variables followed a Gaussian distribution as revealed by the D'Agostino-Pearson normality test. All percentages were arcsine-transformed before statistical analysis. All data are shown as mean with standard error of mean (mean ± SEM). Statistical comparisons between the two groups were made using two-tailed Student's t test. For the in vivo cell fate analysis of WT and cKO mice at different time point after TAM injection, two-way ANOVA were used. For the onesample t test, the treatment group was first normalized by the control group, and then the one-sample t test against a mean of one was used on the normalized values. Probabilities of P < 0.05 were considered as significant. For the in vitro analysis, the number of independent experiments is specified in the legend of each figure.
To further determine the role of WDR81 in neurogenesis in vivo, we generated inducible conditional Wdr81 knockout mice (Wdr81 cKO) by crossing Wdr81 f/f mice with inducible Nestin-CreERT2 transgenic driver mice and Rosa26-stop-DTR-2A-EGFP reporter mice (Fig. S3a). Male littermates without the Wdr81 flox allele were used as wildtype controls (WT). After tamoxifen (TAM) injection, the expression of WDR81 was undetectable in GFP + recombination cells in Wdr81 cKO mice (Fig. S3b). We then analyzed the number of GFP + cells and their phenotypes in Wdr81 cKO and WT mice at 1, 14, 28, and 56 days after TAM injection to cover the critical development stages of adult hippocampal neurogenesis. In general, Wdr81 cKO mice generated significantly fewer GFP + cells in the DG of adult hippocampus compared to WT mice (Fig. 1a, b). Next, we performed fate mapping of GFP + cells using cell lineage markers to determine which cells were affected in Wdr81 cKO mice (Fig. 1c-f). At 1day after TAM injection, no difference was observed between WT and Wdr81 cKO mice in the number of GFP + cells or their lineages (Fig. 1g-l). At all given time-points after TAM treatment, there was a similar number of GFP + GFAP + Sox2 + type 1 NSCs in WT and Wdr81 cKO mice (Fig. 1g). Starting from 7 days after TAM injection, however, Wdr81 cKO mice generated significantly fewer GFP + GFAP − Sox2 + type 2 NSCs than WT controls (Fig. 1h). Using the cell proliferation marker Ki67, we found that Wdr81 cKO mice had fewer proliferating GFP + DCX − Ki67 + progenitor cells than the WT controls from 7 days after TAM treatment (Fig. 1i). Subsequently, Wdr81 cKO mice produced fewer proliferating GFP + DCX + Ki67 + neuroblasts (Fig. 1j), GFP + DCX + Ki67 − postmitotic DCX + immature neurons (Fig. 1k), and GFP + NeuN + mature neurons. (Fig. 1l). In contrast, the numbers of GFP + GFAP + S100β + astrocytes were comparable in WT and Wdr81 cKO mice at all given time-points after TAM injection ( Fig. S3c and d), which suggests that WDR81 is not required for astrocytic cell fate specification of aNPCs. Thus, WDR81 deficiency led to decreased proliferation of neural progenitor cells, and consequently to reduced neuronal differentiation in adult hippocampal neurogenesis. In addition, no elevation in caspase 3 activation was observed in Wdr81 −/− aNPCs, or their differentiated derivatives, or the GFP + cells in Wdr81 cKO mice ( Fig. S3e  and f, Fig. S4). This excludes the involvement of apoptosis in this process.

Deletion of WDR81 from aNPCs leads to impaired hippocampus-dependent learning
Since hippocampal neurogenesis is involved in hippocampus-dependent learning, we generated another cohort of Wdr81 cKO mice by crossing Wdr81 f/f mice only with inducible nestin-CreERT2 transgenic driver mice (Fig.  S5a) and assessed whether their learning ability is affected (Fig. 2a). Before performing the hippocampus-dependent learning tasks, we treated this cohort of mice with TAM for 5 days, and then labeled dividing cells with BrdU 1 week after TAM treatment to confirm that this cohort of Wdr81 cKO mice had impaired hippocampal neurogenesis (Fig.  S5b). As expected, Wdr81 cKO mice displayed reduced numbers of BrdU + cells ( Fig. S5c and d), as well as BrdU + Sox2 + GFAP − type 2 NSCs (Fig. S5f and h), but not BrdU + Sox2 + GFAP + type 1 NSCs ( Fig. S5e and g) at 2 h after BrdU injection. At 24 h after BrdU labeling, Wdr81 cKO mice displayed a significantly higher cell cycle exit index ( Fig. S5i and j). Through in vitro cell cycle analysis, we found that Wdr81 −/− aNPCs indeed showed an increased number of cells in G1/G0 phase, but a decreased number in G2/M phase compared with Wdr81 +/+ aNPCs (Fig. S6). At 4 weeks after BrdU exposure, Wdr81 cKO mice exhibited a decreased number of BrdU + NeuN + neurons ( Fig. S5k and  l). These data were consistent with the in vivo cell lineage tracing. We therefore subjected this cohort of Wdr81 cKO and WT mice to a fear-conditioning test, in which performance in the context test is dependent on the function of both the hippocampus and the amygdala, while performance in the cue test is only dependent on the function of the amygdala [23]. Wdr81 cKO and WT mice had a similar level of freezing behavior in the cue test (Fig. 2b, d). However, Wdr81 cKO mice exhibited significantly less contextual learning as they showed a reduction of freezing behavior in the contextual test compared with WT mice (Fig. 2b, c). To further confirm that Wdr81 cKO mice have hippocampal learning deficits, we subjected the mice to the novel objective localization (NOL) test, which requires intact hippocampus function [24]. Compared with WT mice, Wdr81 cKO mice displayed no preference for the object in the novel location during NOL assessment (Fig. 2e, f). However, in comparison with control littermates, Wdr81 cKO mice did not change their overall level of ambulation level, quantified as the mean moving speed, moving distance, and moving time on the floor plane (Fig.  S7a-c), nor did they show reduced exploration in the center of the arena, expressed as the number of entries into the center, and the time spent in the center (Fig. S7d-f). In addition, Wdr81 cKO mice did not show a decreased tendency to explore the light/dark compartments: the total number of entries and accumulated time spent exploring the light/dark compartments were not significantly different from their control littermates ( Fig. S7g and h). Taken together, these behavioral assessments suggest that selective deletion of Wdr81 in aNPCs leads to impaired hippocampus-dependent learning, but does not result in abnormal locomotor activity or overt anxiety-like behavior.
PI3K-III is a multimeric complex that consists of the VPS34 kinase, the scaffolding protein VPS15, and several accessory subunits, including Beclin1 [26]. The expression levels of VPS34, VPS15, and Beclin1 were comparable in Wdr81 +/+ and Wdr81 −/− aNPCs ( Fig. S9a and b). Therefore, we investigated whether WDR81 regulates PI3K-III activity by affecting the assembly of the PI3K-III complex. In co-immunoprecipitation (co-IP) assays, WDR81 coprecipitated with VPS15 and Beclin1, but not with VPS34 ( Fig. 3e-g). In WDR81 −/− cells, a significant increase of the co-IP efficacy between VPS34 and VPS15 (Fig. 3h, i) was detected, while the interaction of VPS34 with Beclin1 was not changed (Fig. 3j, k). These results suggested that WDR81 deficiency enhances VPS15-VPS34 interaction, and subsequently increases the activity of VPS34.
To determine if the defective neurogenesis in Wdr81 -/-aNPCs results from the increase in endosomal PtdIns3P levels, we treated in vitro cultured aNPCs with 3methyladenine (3-MA), a widely used PI3K-III inhibitor [27]. Compared with the vehicle (control) treatment, 3-MA caused a significant decrease of the percentage of BrdU + cells and differentiated Tuj1 + cells (Fig. S9c-f). This is consistent with previous findings that PI3K-III activity is required for proliferation and differentiation of aNPCs [28]. Compared with the vehicle-treated Wdr81 -/-aNPCs, however, Wdr81 -/-aNPCs treated with 3-MA exhibited a significant increase in the population of both BrdU + cells and Tuj1 + cells (Fig. S9c-f). In addition, 3-MA treatment rescued the cell cycle progression of WDR81-deficient aNPCs to the control levels (Fig. S10). These results indicated that reducing PI3K-III activity rescued the proliferation and differentiation defects of WDR81 -/-aNPCs in vitro. To prove this further, we assessed the effect of 3-MA treatment on aNPC proliferation in vivo by quantifying BrdU + cells and BrdU + Sox2 + GFAP − type 2 NSCs in the DG of hippocampus 2 h after BrdU injection (Figure 3l). 3-MAtreated WT mice exhibited a significant decrease in the number of BrdU + cells ( Fig. S11a and b) and BrdU + Sox2 + GFAP − type 2 NSCs (Fig. 3m, n) compared with the vehicle-treated control. In contrast, the number of BrdU + cells ( Fig. S11a and b) and BrdU + Sox2 + GFAP − type 2 NSCs (Fig. 3m, n) was significantly increased in 3-MAtreated Wdr81 cKO mice. Similarly, 3-MA treatment caused a decrease in the number of BrdU + cells (Fig. 3l and Fig.  S11c and d) and BrdU + NeuN + neurons (Fig. 3o, p) in WT mice, but significantly increased the number of BrdU + cells ( Fig. S11c and d) and BrdU + NeuN + cells in Wd81 cKO mice (Fig. 3o, p) at 4 weeks after BrdU injection. Furthermore, 3-MA treatment significantly improved the hippocampus-dependent learning abilities of Wdr81 cKO mice in both the contextual fear conditioning test and the novel objective location test (Fig. S12). Altogether, these in vitro and in vivo data suggested that reducing PI3K-III activity has a beneficial effect on the abnormal hippocampal neurogenesis in Wdr81 cKO mice.

WDR81 regulates endosomal SARA-TGFβ signaling
Given that WDR81 deficiency led to a significant increase in endosomal Ptdlns3P and consequently caused defective neurogenesis, we sought to determine the signaling pathway that is misregulated by the elevated PtdIns3P in Wdr81 -/-aNPCs. We firstly examined the gene expression levels of several FYVE-domain-containing proteins that are able to bind PtdIns3P and are implicated in signal transduction [29], including SARA (SMAD anchor for receptor activation), FGD1 (FYVE, RhoGEF, and PH domain containing Fig. 2 Wdr81 deletion in Nestinexpressing cells impairs hippocampus-dependent learning. a Experimental scheme for hippocampus-dependent learning after TAM injection. b-d Analysis of the percentage of freezing behavior in WT and Wdr81 cKO mice during fear conditioning tests (c, the contextual test, WT vs. cKO, t test, p = 0.0037; d, the cue test, WT vs. cKO, p = 0.778. WT, n = 7 mice, cKO, n = 7 mice). e and f Analysis of the percentage of time spent exploring the newly-located object in WT and Wdr81 cKO mice (WT vs. cKO, t test, p < 0.001. WT, n = 10 mice, cKO, n = 13 mice). Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001 1) and Mtmr3 (myotubularin-related protein 3). Among them, SARA has the highest expression levels in aNPCs (Fig. S13a). In addition, SARA is located on endosomes (which are named SARA endosomes) and regulates cell division of NPCs in zebrafish spinal cord [30]. In Wdr81 -/-aNPCs, FLAG-tagged SARA was enriched on the abnormally enlarged endosomes (Fig. 4a, b), though the total protein levels of endogenous SARA were not changed ( Fig. S13b and c). However, SARA did not co-precipitate with WDR81 (Fig. S13d). Importantly, 3-MA treatment fully rescued the enlargement of FLAG-SARA-positive endosomes in Wdr81 -/-aNPCs to the WT level (Fig. 4c, Fig.  S14). This suggested that SARA probably persisted on the enlarged endosomes in Wdr81 -/-aNPCs. The percentage of EEA1 + endosomes that were also positive for SARA was indeed significantly higher in Wdr81 -/-aNPCs than in Wdr81 +/+ aNPCs (Fig. 4d, e). This increase was rescued by 3-MA treatment (Fig. 4f).
SARA is believed to be required for the recruitment of Smad2 to the ligand-bound transforming growth factor-β receptor (TGFβR), an event that mediates downstream signal transduction [31,32]. We therefore investigated whether TGFβ signaling is changed in Wdr81 -/-aNPCs. WDR81 deficiency led to attenuated degradation of TGFβR induced by TGFβ1 treatment in aNPCs ( Fig. S13e and f). Consistent with this, the expression level of p-Smad2 (an indicator for activation of TGFβ signaling) was increased in Wdr81 -/-aNPCs, while the levels of p-Smad1/5/9, which indicate the activation of BMP signaling, remained similar to those in Wdr81 +/+ aNPCs (Fig. 4g-i). In addition, immunofluorescence analysis of Wdr81 cKO mice also revealed an increase in p-Smad2 compared with WT mice (Fig. 4j and k). Accordingly, Wdr81 -/-aNPCs exhibited increased mRNA expression of Smad2 downstream target genes that trigger cell cycle exit of aNPCs [33,34], such as p15, p57, and p21; the protein levels were also elevated ( Fig. 4l-n). In contrast, the expression of non-Smad2 downstream target genes, such as p27, CDK4, and cyclin D1, was comparable in Wdr81 +/+ and Wdr81 -/-aNPCs ( Fig. 4l-n). Taken together, these data suggested that WDR81 deficiency impaired SARA-endosome trafficking, leading to prolonged SARA-mediated TGFβ signaling in aNPCs.

Inhibition of SARA-mediated TGFβ signaling rescues the adult hippocampal neurogenesis defects in Wdr81 cKO mice
Since excessive activation of PI3K-III in Wdr81 -/-aNPCs led to persistent recruitment of SARA onto endosomes, and subsequent hyper-activation of SARA-mediated TGFβ signaling, we hypothesized that inhibition of SARA-mediated TGFβ signaling may rescue the adult neurogenesis deficits in Wdr81 cKO mice. Compared with the control shRNA treatment, shRNA knockdown of SARA in Wdr81 -/-aNPCs in vitro led to a significant increase in the percentage of BrdU + cells and differentiated Tuj1 + cells (Fig. S15a-c). To confirm the specificity of shRNA manipulations, we generated lentiviruses expressing a fusion protein of FLAG and a shRNA-resistant form of SARA (SARA R ) that harbors two silent mutations within the sequence targeted by shSARA (Fig. S15a). Western blot analysis confirmed the resistance of SARA R to shSARA in vitro (Fig. S15a). In addition, re-expression of SARA R eliminated the rescue effect of shSARA on aNPC proliferation and neuronal differentiation caused by WDR81 deficiency (Fig. S15b and  c). SARA shRNA also downregulated the expression level of p-Smad2 in Wdr81 -/-aNPCs ( Fig. S15d and e). Knockdown of SARA also rescued the cell cycle progression of WDR81-deficient aNPCs to the control levels (Fig. S16). We then stereotaxically injected retroviruses co-expressing Cre-2A-GFP and SARA shRNA (Fig. 5a) into the DG of adult hippocampus to inactivate WDR81 and SARA expression in aNPCs. Compared with the control shRNA injection, retrovirus-mediated SARA shRNA led to a significant increase in GFP + BrdU + progenitor cells and GFP + DCX + differentiated neurons in Wdr81 cKO mice ( Fig. 5b-e). Thus, downregulation of SARA rescued the proliferation and differentiation defects of aNPCs caused by loss of WDR81. Finally, we evaluated the rescuing effect of LY2109761 (LY), a small molecule inhibitor of the TGF-β receptor type 1/type II kinases, on hippocampal neurogenesis in Wdr81 cKO mice (Fig. 5f). LY-treated WT mice exhibited an increased number of BrdU + cells, BrdU + Sox2 + GFAP − type 2 NSCs, and BrdU + NeurN + neurons compared with the vehicle treatment ( Fig. 5g-j). This is consistent with the previous study showing that inhibition of TGFβ signaling promoted proliferation and neuronal differentiation of aNPCs [35]. Interestingly, LY treatment induced a significant increase in BrdU + cells, BrdU + Sox2 + GFAP − type 2 NSCs and BrdU + NeurN + neurons in Wdr81 cKO mice (Fig. 5g-j). Taken together, these data suggested that inhibition of SARA-mediated TGFβ signaling rescued the adult hippocampal neurogenesis defects caused by WDR81 deficiency.

Discussion
Adult neurogenesis recapitulates the entire process of embryonic neural development and has been implicated in learning/memory and affective behaviors [1,2,36]. Using adult mouse hippocampal neurogenesis as a model, we showed that selective deletion of Wdr81 in aNPCs markedly reduced hippocampal neurogenesis and impaired hippocampus-dependent learning. The interaction of WDR81 with VPS15 and Beclin1 likely inhibits the assembly of PI3K-III complex, thereby preventing endosomal PtdIns3P synthesis. As a result, WDR81 depletion in aNPCs led to elevated levels of endosomal PtdIns3P, which in turn led to abnormal neurogenesis in adult hippocampus. Supporting this conclusion, chemical inhibition, or shRNA inactivation of PI3K-III markedly ameliorated the defective neurogenesis in Wdr81 cKO mice. Together these results suggest that WDR81 promotes adult neurogenesis by controlling endosomal PtdIns3P levels in aNPCs (Fig. S17).
Although exhaustive efforts have been devoted to understanding the underlying mechanisms that regulate neurogenesis, significant questions still remain to be addressed regarding the molecular mechanisms regulating different aspects of neurogenesis under physiological and pathological conditions. PI3K signaling is a vital pathway controlling cell survival, proliferation, and apoptosis [37], and is involved in a number of neurological disorders, such as autism and schizophrenia [38]. Based on primary structure, regulation, and lipid substrate specification, the PI3K family is divided into three different classes (Class I, Class II, and Class III), which are involved in various cellular functions [26,39]. Class I-PI3K is responsible for the production of phosphatidylinositol-3,4,5-trisphosphate (PI (3,4,5)P3), a second messenger for the translocation and activation of Akt. The PI3K/Akt pathway is known to play important roles in adult neurogenesis. Class III-PI3K (PI3K-III) is responsible for Ptdlns3P production, which recruits downstream effector proteins containing Ptdlns3P-binding domains and is involved in autophagy and endocytic trafficking. Inhibition of class-III PI3K activity by pharmacological treatment or by genetic mutation of PI3K-III complex components has been shown to decrease both the proliferation and differentiation of aNPCs [28,40], which suggests that the activation of PI3K-III is essential for neurogenesis. In this study, we found that ablation of WDR81 inhibited the proliferation and differentiation of aNPCs by increasing the activity of PI3K-III. Inhibition of PI3K-III activity rescued the proliferation and differentiation deficits of aNPCs in vitro and in vivo caused by loss of WDR81. Thus, the above data suggest that precise regulation of PI3K-III activity and PtdIns3P levels are important for proper neurogenesis in the brain.
Mutations of WDR81 are associated with neurological disorders including CAMRQ2 and microcephaly [16,17]. Our previous studies revealed that loss of WDR81 and its interaction partner WDR91 disrupted PtdIns3P-dependent endosome conversion in the endosome-lysosome pathway [12,14]. Here, we further established that the WDR81-SARA-TGFβ axis regulates appropriate adult neurogenesis. Together these findings provide mechanistic explanations for WDR81-associated neurological disorders. In addition, WDR81 can also function independently of PtdIns3P by coordinating with the p62 autophagic receptor to remove ubiquitinated protein aggregates through autophagy [15]. The multifunctional properties of WDR81 suggest that the WDR81 mutation-associated human diseases could result from defects in diverse cellular processes. Particularly, dysregulation of neurogenesis and neural development are likely responsible for the pathology of WDR81 mutationassociated CAMRQ2 and microcephaly. Of note, our findings that pharmacological inhibition of PI3K-III and SARAdependent TGFβ signaling markedly ameliorated the defective neurogenesis caused by Wdr81 deficiency offer potential therapeutic strategies for treating the mental illness seen in WDR81-associated neurological disorders. Fig. 5 Inhibition of endosomal SARA-TGFβ signaling rescues the adult hippocampal neurogenesis defects in Wdr81 cKO mice. a Schematic illustration of the retroviral vectors and the experimental timeline of in vivo retrovirus labeling for analyzing aNPC proliferation and differentiation in the adult DG. b Representative images of retrovirus-labeled GFP + cells co-stained with BrdU in the adult DG. Scale bar = 50 μm. c Analysis of the number of BrdU + GFP + cells in the DG of Wdr81 cKO mice treated with retrovirus-mediated SARA shRNA (cKO + shNC vs. cKO + shSARA, t test, p = 0.043; shNC, n = 3 mice, shSARA, n = 3 mice). d Representative images of retrovirus-labeled GFP + cells co-stained with DCX. Scale bar = 50 μm. e Quantification of DCX + GFP + cells in the DG of Wdr81 cKO mice treated with retrovirus-mediated SARA shRNA (cKO + shNC vs. cKO + shSARA, t test, p = 0.042; shNC, n = 3 mice, shSARA, n = 3 mice). f Experimental scheme for assessing the proliferation and differentiation of aNSCs in the DG of adult hippocampus after treatment with the TGFβ inhibitor LY2109761 (LY). g, h Quantification of BrdU + cells (g, cKO vs. cKO + LY, t test, p = 0.008) and BrdU + GFAP -Sox2 + cells (h, cKO vs. cKO + LY, t test, p = 0.016) in the DG of Wdr81 cKO mice and WT mice treated with LY at 2 h after BrdU injection (WT, n = 5 mice, WT + LY, n = 6 mice, cKO, n = 6 mice, cKO + LY, n = 5 mice). I, j Quantification of BrdU + cells (i, cKO vs. cKO + LY, t test, p = 0.033) and BrdU + NeuN + cells (j, cKO vs. cKO + LY, t test, p = 0.014) in the DG of Wdr81 cKO mice and WT mice treated with LY at 4 weeks after BrdU injection (WT, n = 7 mice, WT + LY, n = 6 mice, cKO, n = 5 mice, cKO + LY, n = 6 mice). Data are presented as mean ± SEM; *p < 0.05; **p < 0.01