Pten deletion causes mTorc1-dependent ectopic neuroblast differentiation without causing uniform migration defects

In rodent brain, thousands of new neuronal precursors are born every day in the subventricular zone (SVZ), the postnatal neural stem cell niche in the lateral walls of the lateral ventricles. Multiple cell types reside in this niche including ‘type B’ Gfap-expressing neural stem cells that generate ‘type C’ Mash1 (Ascl1 – Mouse Genome Informatics)-expressing transit-amplifying cells. Transit-amplifying cells further differentiate into ‘type A’ migrating neuronal precursor cells or neuroblasts that express the microtubule-associated protein doublecortin (Dcx). Neuroblasts migrate from the SVZ through the rostral migratory stream (RMS) to the olfactory bulb (OB) where they differentiate into interneurons and are integrated into local circuits (Alvarez-Buylla and Garcia-Verdugo, 2002).

In contrast to much of embryonic development in which immature neurons migrate along radial glial fibers, in the RMS, neuroblasts migrate along one another in a chain formation, separated from the surrounding brain parenchyma by a glial tube (Ghashghaei et al., 2007). Removal of the OB does not disrupt this rostral migration, indicating that long-distance chemoattractant signals emanating from the final destination are not required (Kirschenbaum et al., 1999). Chemorepulsive cues involving secreted SLIT ligands and their receptors, the ROBO family of proteins, are involved in directing newly generated neuroblasts to enter the RMS from the SVZ, and in mediating interactions in the RMS between migrating neuroblasts and astrocytes that form the glial tube. Extracellular matrix cues, including polysialated neuronal cell adhesion molecule (PSA-NCAM) and integrin signaling, are also important for normal migration through the RMS to the OB, and reelin and tenascin R act as a detachment signal to dictate the switch from tangential to radial migration when neuroblasts reach the OB (Ghashghaei et al., 2007). Additional guidance cues might be provided by the vasculature along the RMS (Snapyan et al., 2009). Thus, diverse extracellular cues dictate the continual process of directional migration of neuroblasts from the SVZ through the RMS to their final destination as integrated interneurons in the OB. However, the intracellular signaling pathways that respond to these extracellular cues, instructing when and where migrating neuroblasts stop, are largely unknown.

The phosphoinositide 3-kinase (PI3K) signaling pathway is an evolutionarily conserved intracellular signaling cascade that transduces extracellular signals to regulate multiple processes, including proliferation, survival, metabolism and cell migration (Engelman et al., 2006). PI3Ks phosphorylate phosphatidylinositol-4,5-bisphosphate (PIP2) to generate phosphatidylinositol-3,4,5-trisphosphate (PIP3), resulting in downstream consequences, including activation of the serine-threonine kinase Akt, suppression of the tuberous sclerosis complex, and activation of mTorc1 (mTOR complex 1). Pten is the major negative regulator of the PI3K pathway, acting as a lipid phosphatase to directly antagonize PI3K by dephosphorylating PIP3 to PIP2 (Chalhoub and Baker, 2009).

Through numerous studies employing different cell types, the PI3K pathway has been shown to influence cell migration at multiple levels, including transducing chemotactic signals, establishing cell polarity and influencing cell adhesion (Cain and Ridley, 2009). In Dictyostelium, PTEN was required for directional migration towards a chemotactic stimulus (Funamoto et al., 2002; Iijima and Devreotes, 2002). However, more recently Dictyostelium simultaneously lacking all Class I PI3K isoforms and PTEN were shown to have normal directional chemotactic migration, but reduced speed. The most substantial effect of PTEN deletion in this study was a decrease in random movement in the absence of a chemotactic signal (Hoeller and Kay, 2007). Thus, even within simple organisms, the precise mechanisms through which PI3K signaling influences migration are not straightforward.

Inherited mutations disrupting PI3K-AKT-mTORC1 regulation are associated with human syndromes that include neurological abnormalities, indicating that the pathway is important in normal neural development and/or function (Inoki et al., 2005; Endersby and Baker, 2008). Conditional knockout mice showed that Pten is important in brain for diverse processes, including proper neuronal positioning during development as well as negative regulation of neural stem cell self-renewal and proliferation, and neuronal morphogenesis and size regulation (Backman et al., 2001; Kwon et al., 2001; Marino et al., 2002; Waite and Eickholt, 2010). However, the underlying mechanism of how Pten is involved in neuronal migration is currently unknown.

In the present study, we demonstrated that Pten inhibits downstream activation of mTorc1 in normal migrating neuroblasts in the RMS. Postnatal deletion of Pten in this population resulted in ectopic positioning of neurons that failed to reach their final destination, reminiscent of defects in radial glia-guided neuronal migration caused by Pten deletion during development. The phenotype was fully rescued by rapamycin, demonstrating that mTorc1 is required for the Pten-mediated regulation of migration in the RMS. We used live imaging of slice cultures to visualize the defect in migrating neurons. Unlike the more uniform phenotypes induced throughout the population in PTEN-deficient Dictyostelium, we found that a subpopulation of Pten-null neuroblasts showed normal directional migration, whereas another subpopulation showed a complete lack of directional migration and showed non-polarized morphology. This altered morphology was associated with the expression of markers of differentiation. This strongly suggests that the neuroblast migration defect associated with Pten loss might be secondary to precocious differentiation rather than a defect in the mechanics of directional migration in the RMS.

Nestin-creER^T2 mice (Cicero et al., 2009) contain a transgene comprising the Nestin promoter and second intron (Zimmerman et al., 1994) driving expression of CreER^T2 (Metzger et al., 1995; Feil et al., 1996; Feil et al., 1997), an internal ribosomal entry site (IRES) and human placental alkaline phosphatase (hPLAP; ALPP – Human Gene Nomenclature Database) specifically in progenitor cells in the central nervous system. To map Cre activity, Nestin-creER^T2 transgenic mice were bred with the R26R-lacZ reporter (R26R) (Soriano, 1999) or R26LSL-EYFP (Srinivas et al., 2001) mice to generate Nestin-creER^T2;R26R-lacZ or Nestin-creER^T2;R26LSL-EYFP mice. To generate Pten^cKO mice (Nestin-creER^T2; Pten^loxP/loxP or R26LSL-EYFP;Pten^loxP/loxP for cre electroporation experiments), Pten^loxP mice (Suzuki et al., 2001) were intercrossed with Nestin-creER^T2 or R26LSL-EYFP mice. For all analyses including Nestin-creER^T2, the transgene was hemizygous to avoid variation from transgene dosage. Controls were littermate Pten^loxP/loxP mice without Nestin-creER^T2. All procedures were reviewed and approved by the Animal Care and Use Committee at St Jude Children’s Research Hospital.

Tamoxifen (TM; Sigma) was dissolved at 20 mg/ml or 5 mg/ml at 37°C in corn oil (Sigma) for Cre induction in adult mice or pups, respectively, filter sterilized and stored at 4°C in the dark for up to 10 days. Cre activity was induced by introperitoneal (i.p.) injection of TM at 3 mg/40 g body weight daily from postnatal day (P) 0 to P1, or from P11 to P12, as indicated, in pups, or 9 mg/40 g body weight daily from P28 to P30 in adults. Daily injections were separated by 24 hours. Sterile corn oil was the vehicle control.

5-Bromo-2′-deoxyuridine (BrdU; Sigma) at 5 mg/ml in sterile PBS was stored at –20°C. BrdU (50 μg/g body weight) was injected i.p. five times at 2-hour intervals for short-term BrdU labeling, or six times at 12-hour intervals from P4 to P6 for BrdU birthdating.

Rapamycin (LC Laboratories) at 20 mg/ml in sterile dimethyl sulfoxide (DMSO) was stored at –20°C and diluted in 5.2% Tween80 immediately before use. Rapamycin (1.5 μg/g body weight) was injected i.p. daily from P8 to P31. Sterile DMSO+Tween80 was the vehicle control. When TM was given on the same day, the two drugs were separated by 6 hours.

For cryosections, mice were anesthetized and perfused transcardially with PBS followed by 2% paraformaldehyde (PFA) in PBS. Following dissection, tissues were post-fixed overnight in 2% PFA in PBS at 4°C, and then equilibrated in 25% sucrose in PBS for an additional 24 hours at 4°C. Tissues were embedded in TBS embedding media (Triangle Biomedical Sciences) on dry ice and cut into 12 μm-thick cryosections. Tissue slides were equilibrated at room temperature for 20 minutes then washed three times in PBS prior to staining. Cre reporter activity from the R26R-lacZ reporter was detected by X-gal staining to detect β-galactosidase activity (Chow et al., 2008). Cre reporter activity from R26LSL-EYFP mice was detected by anti-GFP immunofluorescence (IF) on cryosections.

For paraffin sections, tissue was processed the same way as above except using 4% PFA for perfusion and 24 hour post-fixation, then embedded in paraffin, and cut into 5 μm sections. Primary antibodies for immunostaining were: anti-GFP (1:1000, Invitrogen A6455; 1:2000, Abcam #13970), anti-Pten [1:100 for immunohistochemistry (IHC), 1:500 for IF with tyramide amplification, Cell Signaling #9559], anti-p-Akt S473 (1:50, Cell Signaling #9271), anti-p-S6 S235/236 (1:500, Cell Signaling #2211), anti-Dcx (1:4000 for IHC, 1:500 for IF, Chemicon AB5910), anti-Mash1 (1:100, BD #556604), anti-NeuN (1:500, Chemicon), anti-Map2 (1:5000, Sternberger SMI52), anti-calretinin (1:2000, Chemicon AB149), anti-tyrosine hydroxylase (1:500, Sigma T1299), anti-Gfap (1:200, Sigma G3893), anti-Ki67 (1:5000, Novocastra NCL-Ki67p), anti-BrdU (1:1000, AbDSerotec OBT0030CX) and anti-active Caspase3 (1:1000, BD #559565). All IHC used microwave antigen retrieval, biotinylated secondary antibodies in conjunction with horseradish peroxidase-conjugated streptavidin (Elite ABC, Vector Laboratories), color development with substrates NovaRed, DAB or VIP (Vector Laboratories) and counterstaining with Hematoxylin or Methyl Green (Vector Laboratories). For IF, AlexaFluor 488-, AlexaFluor 647- (Invitrogen) and Cy3-, Cy5- (Jackson ImmunoResearch) conjugated secondary antibodies were employed along with Vectashield mounting media containing 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories). A Tyramide Signal Amplification Kit (Perkin Elmer) was used for Pten and p-Akt S473 IF. All IF images are shown as z-stack projections from confocal microscopy. Terminal deoxynucleotidyl transferase-mediated nick end labeling (TUNEL) staining was conducted with the ApopTag Peroxidase In Situ Apoptosis Detection Kit (Chemicon, S7100).

Ki67 or BrdU IHC was quantitated from anatomically matched sections. For proliferation, the Bioquant system (R&M Biometrics) was used to count all Ki67+ cells in the SVZ. For BrdU birthdating (Fig. 6), BrdU+ cell number and granule cell density in a 20× objective field in the same areas of OB GCL or total number of BrdU+ cell in the SVZ were counted with ImageJ v1.44 cell counter. Four controls and four Pten^cKO brains were analyzed for each group and positive cells were counted from three sections from each brain for each staining. Two-tailed Student’s t-test was used for statistical significance analysis.

In vivo electroporation was performed similarly to previous reports (Boutin et al., 2008; Chesler et al., 2008). Endotoxin-free pCAG-cre expression plasmid (2 μl of 0.65 μg/μl, in PBS containing 1% Fast Green) was injected into the lateral ventricle of P2 pups with a 30G needle Hamilton syringe using the midpoint between Bregma and Lambda and 1 mm lateral to the midline suture as landmarks. Fast Green dye allowed visual confirmation of correct injection into lateral ventricles. Only successfully injected animals were subjected to five square-wave electrical pulses (50 mseconds separated by intervals of 950 mseconds, 150 V).

Mouse brains were placed in sterile cold (4°C) dissecting Complete Hank’s Balanced Salt Solution (complete HBSS) containing 1× HBSS (GIBCO), 2.5 mM Hepes (pH 7.4, GIBCO), 30 mM d-glucose, 1 mM CaCl₂, 1 mM MgSO₄ and 4 mM NaCO₃ (Polleux and Ghosh, 2002), dissected, embedded in 3% low-melting point agarose (Promega) and cut into 300 μm-thick sagittal sections with a vibratome. Slices containing the RMS were identified by enhanced yellow fluorescent protein (EYFP) expression, equilibrated for 2 hours in artificial cerebrospinal fluid (aCSF) containing 125 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃ and 20 mM d-glucose (285-295 mOsm) with 95% O₂/5% CO₂ at room temperature and then transferred into the submerged recording chamber and superfused (1-2 ml/minute) with warm (36-37°C) aCSF.

Time-lapse live imaging was acquired by two-photon laser-scanning microscopy equipped with an Ultima imaging system (Prairie Technologies), a Ti:sapphire Chameleon Ultra femtosecond-pulsed laser (940 nm) (Coherent) and a 20× 0.95 numerical aperture water-immersion infrared objective (Olympus). z-stacks (512 × 512 pixels, 0.079 μm/pixel, 60-100 sections per stack, 1 μm/section) were collected at the elbow of the RMS every 5 minutes for 3-6 hours.

Maximum projections of 20 consecutive sections in the middle of z-stacks of time-lapse live imaging were analyzed using SlideBook v5.0 (Intelligent Imaging Innovations). Cells that exhibited bipolar morphology and moved at least two cell body lengths were counted as ‘bipolar migrating cells’. All other cells were considered to be ‘stationary or non-polar cells’. For each movie, all the EYFP+ cells were manually tracked but only those that could be observed more than ten consecutive frames (45 minutes) were included in the analysis. The average speeds (total displacement/time) or endpoint speeds (the distance between the first and last time points of the path/time) were binned in 20 μm/hour intervals to obtain a distribution. χ² analysis or Student’s t-test was used to assess the significance of the distributions or means, respectively, of average speeds (or endpoint speeds).

To understand PI3K signaling regulation in the SVZ-RMS-OB (Fig. 1A), we examined expression of several key components of the pathway. Levels of Pten expression were heterogeneous in the SVZ, and all of the cells expressing Pten were Dcx+ neuroblasts (Fig. 1B). In the SVZ and proximal RMS, the majority of cells expressing p-S6, a downstream indicator of PI3K-mTorc1 pathway activity, were Mash1+ transit-amplifying cells, and not Dcx+ migrating neuroblasts (Fig. 1C) (data not shown). By contrast, in the terminal RMS at the OB core, a substantial population of Dcx+ type A cells expressed p-S6 (Fig. 1D). This indicates that the PI3K pathway was not strongly active in migrating neuroblasts until these cells reached their destination in the OB.

To evaluate the function of the PI3K pathway regulator Pten in the postnatal neural stem/progenitors, we generated Nestin-creER^T2 mice with inducible Cre activity in the SVZ (supplementary material Fig. S1) and bred them to Pten^loxP/loxP (Suzuki et al., 2001) to generate inducible Pten conditional knockout mice (hereafter Pten^cKO). Induction of Pten deletion in adult SVZ by tamoxifen (TM) injection in Pten^cKO mice at P30-32 did not cause obvious neurological abnormalities throughout a normal lifespan except for occasional seizures in some aged animals. Analysis of all Pten^cKO brains (n=24) showed dramatic expansion of the SVZ (Fig. 2A) and proximal RMS (supplementary material Fig. S2) that was most pronounced at the dorsal junction of the SVZ abutting the corpus callosum and the ventral tip where it formed a bulb-like structure. The expansion was easily detectable as early as 20 days post induction (DPI) and was similar at 60, 150, 300 and 540 DPI without progressive enlargement at later time points (data not shown). Close examination of the cell morphology in the SVZ revealed that, in contrast to wild-type control brains in which the SVZ was mainly composed of well-organized spindle-like progenitor cells with scant cytoplasm, the expanded SVZ in Pten^cKO mice contained a central region of cells with large round nuclei interspersed with isolated progenitor cells, and with clusters of progenitor cells at the edge (Fig. 2B). IHC analysis showed that the expanded SVZ was negative for Pten staining (Fig. 2C) with marked elevation of p-Akt (supplementary material Fig. S3) indicating overall activation of the PI3K pathway in the entire region. When conditional Pten deletion was induced in newborn Pten^cKO mice by TM injection at P0 and P1, similar dorsal SVZ and RMS expansion were consistently detected by P10-30 in the absence of other obvious abnormalities (data not shown), although the ventral expansion of the expanded SVZ was not detected at this age.

The expanded region of the SVZ and RMS was composed of NeuN+ (Rbfox3 – Mouse Genome Informatics) and Dcx+ cells, which mark differentiated neurons and migrating neuroblasts, respectively (Fig. 2D,E; supplementary material Fig. S2B,C). NeuN was not expressed in any of the wild-type SVZ cells, but was strongly expressed in the cKO SVZ, specifically in the cells with larger, rounder nuclei that were not observed in wild-type SVZ (Fig. 2D, lower panel). The expression of Dcx decreased and NeuN expression increased in the ectopic cells as mice aged from 150 to 300 DPI (data not shown). The Dcx+ cells in the expanded SVZ and RMS showed evidence of neuronal differentiation, including abnormal formation of projections (Fig. 2E; supplementary material Fig. S2C). Whereas wild-type Dcx+ neuroblasts did not express p-S6 until reaching the OB (Fig. 1C), Dcx+ neuroblasts of the Pten^cKO SVZ showed robust p-S6 staining (Fig. 2F), indicating aberrant activation of the PI3K pathway. The ectopic cells also expressed Map2 (Mtap2 – Mouse Genome Informatics) (Fig. 3A,B), a marker of dendrites of differentiated neurons, and calretinin (Calb2 – Mouse Genome Informatics) (Fig. 3C), a marker expressed by mature GABAergic interneurons in the granule cell layer (GCL) and the glomerular cell layer of the OB arising from neuroblasts of the SVZ-RMS (Kohwi et al., 2005; Kohwi et al., 2007; Alvarez-Buylla et al., 2008). This population is the most abundant population of OB neurons produced postnatally. There were also a smaller number of ectopic cells that expressed tyrosine hydroxylase (TH), a marker of dopaminergic interneurons of the glomerular layer, which are less abundant than the calretinin-expressing neurons of the OB (supplementary material Fig. S4) (Batista-Brito et al., 2008). The expression of neuronal markers of differentiated neurons of the OB in Dcx+ cells in the expanded SVZ indicates ectopic differentiation of neuroblasts that normally remain undifferentiated until they reach the OB. Consistent with a failure of migrating neuroblasts to reach the OB, the diameter of the terminal RMS at the core of the OB and the cell density in the GCL were markedly reduced in Pten^cKO mice (Fig. 4, Fig. 5A; supplementary material Fig. S5). The ectopic cells expressing neuronal differentiation markers also showed evidence of PI3K/mTorc1 activation as shown by co-expression of Map2 and p-S6 (supplementary material Fig. S6).

To determine how loss of Pten affected the proliferation in the intact SVZ in vivo, we used both Ki67 IHC and short-term BrdU labeling, which gave consistent results (supplementary material Fig. S7A) (data not shown). The cell numbers in the SVZ of Pten^cKO mice were increased owing to the presence of ectopic post-mitotic differentiated cells. Considering the percentage of proliferating cells would thus underestimate the proliferation capacity of progenitors in the Pten^cKO SVZ. Therefore, we counted the absolute numbers of Ki67+ cells from matched SVZs of both wild-type and Pten^cKO mice. There was no statistically significant difference in proliferation rate in the Pten^cKO SVZ compared with control littermates (supplementary material Fig. S7A,C).

Activated PI3K signaling can also enhance cell survival; therefore, we used active caspase3 and TUNEL staining to determine whether Pten loss altered apoptosis in the SVZ (supplementary material Fig. S8A). The endogenous level of apoptosis was very low, and not significantly different in wild-type and Pten^cKO SVZ. Consistent with the resistance of differentiated neurons to radiation-induced death (Frappart and McKinnon, 2008), the expanded portion of the Pten^cKO SVZ containing differentiated neurons as well as the parenchyma surrounding the SVZ were resistant to irradiation-induced apoptosis (supplementary material Fig. S8B). However, the progenitor cells of both wild type and Pten^cKO underwent apoptosis 5 hours after exposure to 4 Gy of irradiation. Therefore, Pten-deficient neuroblasts did not show enhanced survival.

To determine whether ectopic neurons arose from proliferative cells in the SVZ that underwent premature termination of migration to the OB, we used BrdU birthdating. Pten deletion was induced by TM injection in newborn mice. Four days later, to allow sufficient time for Pten deletion and loss of Pten protein, BrdU was injected to label proliferating cells in the SVZ. As expected, the majority of labeled cells in control mice migrated into the GCL of the OB by 15 days after the BrdU pulse (Petreanu and Alvarez-Buylla, 2002) (Fig. 5A). By contrast, there was significant accumulation of BrdU+ cells in the expanded SVZ and greatly reduced numbers of BrdU+ cells, and reduced granule cell density in the OB GCL in the Pten^cKO mice (Fig. 5A,B; supplementary material Figs S5, S9). A high percentage of BrdU+ cells in the expanded SVZ expressed NeuN, indicating that ectopic differentiated neurons arose from previously proliferating cells in the SVZ (Fig. 5C). In control mice, few BrdU+ cells remained in the SVZ, and none of them expressed NeuN or had the larger rounded morphology seen in ectopic cells in the Pten^cKO SVZ. They might correspond to slow-cycling neural stem cells or local proliferating glia (Fig. 5C). In the control, BrdU and NeuN double-positive cells were only found in the OB and not in the SVZ or RMS (supplementary material Fig. S9), indicating that BrdU-labeled neuroblasts do not differentiate until they reach the OB in the wild-type mice.

To determine whether inhibition of a downstream effector in the aberrantly activated PI3K pathway could prevent the early termination of tangential migration and ectopic differentiation of Pten^cKO neuroblasts in the SVZ, we treated the mice with rapamycin, an inhibitor of mTorc1 activity. We previously showed that rapamycin required several days of administration to maximally block mTorc1 signaling in brain (Kwon et al., 2003). Therefore, we pre-treated the mice daily with rapamycin for three days, then induced the deletion of Pten with TM at P11-12 and continued rapamycin treatment until P31. Rapamycin treatment completely prevented SVZ expansion in Pten^cKO mice (Fig. 6A), although the SVZ remained Pten-null in rapamycin-treated mice (Fig. 6B). Rapamycin selectively blocked mTorc1 activity, as shown by lack of p-S6, a downstream indicator of mTorc1 activity (Fig. 6C) without altering levels of p-Akt in the Pten-deficient SVZ (Fig. 6D). Neither apoptosis nor proliferation was affected by rapamycin treatment (supplementary material Fig. S7).

To understand how Pten loss affected the movement of tangentially migrating neuroblasts, we used time-lapse live imaging on acute brain slices. We generated R26LSL-EYFP;Nestin-creER^T2;Pten^wt/wt (wild type) or R26LSL-EYFP; Nestin-creER^T2;Pten^loxP/loxP (Pten^cKO) mice so that the EYFP cre reporter expression could be used to visualize cells in which cre-mediated recombination occurred. We induced Nestin-creER^T2 activity at P0 and imaged the RMS in acute brain slices at P16 with two-photon microscopy at the ‘elbow’ of the RMS, where the RMS curves from ventral towards rostral migration halfway between the SVZ and the OB (see supplementary material Fig. S10A for the location). The majority of visible neuroblasts in the Pten^cKO brain slices (supplementary material Movie 2) showed directional migration towards the olfactory bulb similar to that observed in wild type (supplementary material Movie 1). However, a small subpopulation of larger cells with rounded morphology that maintained their position were visible in the Pten^cKO, but not in wild type. Although these movies showed that the directional movement of the majority of Pten^cKO neuroblasts did not seem to be perturbed, the density of EYFP+ cells was too great to quantify the movement of individual cells. Therefore, we used in vivo electroporation to introduce cre recombinase, allowing visualization of a smaller number of isolated cells that could be reliably tracked over time. We transfected the expression construct pCAG-Cre into the SVZ by in vivo electroporation at P2 and visualized transfected cells by expression of the EYFP Cre reporter in Rosa26-loxP-STOP-loxP-EYFP (R26LSL-EYFP);Pten^wt/wt (wild type) or R26LSL-EYFP;Pten^loxP/loxP (Pten^cKO) mice. Neuroblast migration was imaged at P19-20. The majority of EYFP+ cells in the wild-type RMS showed a bipolar morphology, and migrated effectively towards the OB (Fig. 7A,C; supplementary material Movie 3). By contrast, there were two distinct populations of EYFP+ cells in the Pten^cKO RMS (Fig. 7B; supplementary material Movie 4). One population maintained bipolar morphology and was not significantly different in size or directional migration compared with wild-type control (Fig. 7C-E). Another population lost polarity, showed increased soma size and showed only local non-directional movement. Interestingly, the cells that had stopped directional migration showed active spontaneous membrane ruffling, which distinguished them from completely static cells seen at very low frequency in both the wild-type and Pten^cKO RMS. The average speed of all EYFP+ cells in the Pten^cKO RMS (mean: 30.6 μm/hour; n=368), including both cells with normal bipolar morphology and static cells with rounded morphology, was significantly slower than that of the wild type (mean: 47.1 μm/hour; n=194) (P<10^–13). However, the average speed of bipolar migrating neuroblasts in the Pten^cKO RMS (mean: 59.2 μm/hour; n=110) was not slower. Indeed, Pten^cKO neuroblasts that were actively migrating showed a slight increase in speed compared with wild type (mean: 53.8 μm/hour; n=160) (P=0.046) (Fig. 7C). We also measured endpoint speed, a representation of directional movement calculated as the distance between the starting and final position over time. The Pten^cKO bipolar migrating neuroblasts moved effectively in the appropriate caudal-to-rostral direction, as their endpoint speeds were not significantly different from the wild type (mean 47.6 μm/hour and 44.4 μm/hour, respectively) (P=0.15). Accordingly, a substantial number of Pten-deficient neuroblasts were able to migrate to the OB (supplementary material Fig. S10B,C,E). Both the EYFP+ bipolar and round non-polar cells in the RMS expressed Dcx, indicating that they were migrating neuroblasts or neuroblasts that had stopped migration recently (supplementary material Fig. S10D). These data indicate that Pten deletion does not cause a uniform intrinsic defect that compromises neuroblast directional migration, consistent with premature differentiation resulting in ectopic positioning.

Multiple conditional knockout models have shown that Pten deletion during development results in incomplete neuronal migration, disrupting the laminar structure in cerebral cortex and cerebellum. The accumulation of ectopic differentiated Pten^cKO neurons in the SVZ and RMS observed in the present study is consistent with defects in tangential migration, similar to the failure of radial glia-guided migration observed in previous models (Backman et al., 2001; Groszer et al., 2001; Kwon et al., 2001; Marino et al., 2002; Yue et al., 2005).

Pten and the PI3K-Akt-mTorc1 pathway have been implicated in cell migration regulation in multiple contexts. Mechanistically, this was mostly clearly demonstrated in Dictyostelium where subcellular localization of PTEN and PI3K to the trailing and leading edge of the cell, respectively, established a PIP3 gradient that was required for directional migration and chemotaxis (Funamoto et al., 2002; Iijima and Devreotes, 2002). Deletion of the Akt homolog PkbA, rescued the cytokinesis and chemotaxis defects in PTEN-deficient Dictyostelium (Tang et al., 2011). In mammalian cells, the PI3K pathway can regulate cell motility and migration through Akt activation as well as through Akt-independent mechanisms such as Rac1/Cdc42 signaling (Kölsch et al., 2008). In addition, overexpression of wild-type PTEN inhibited migration of human glioma cell lines in an in vitro wound healing assay, and this effect was dependent on the protein phosphatase activity of PTEN independently of PI3K signaling (Raftopoulou et al., 2004).

Unexpectedly, we found that postnatal deletion of Pten divided the RMS neuroblasts into two distinct populations: one subset maintained their bipolar morphology and had normal directional migration, and the other population lost polarity and stopped migration. This non-polar morphology, in association with increased frequency of spontaneous protrusions of the plasma membrane, was reminiscent of PTEN-null Dictyostelium. However, the population of Pten^cKO neuroblasts that maintained bipolar morphology continued moving normally, demonstrating that Pten was not required for directional migration. This is consistent with the observation that some Pten-null cells reach the OB in Pten^cKO mice. A previous study in which Pten was deleted in Gfap-expressing neural stem cells also showed that Pten-deficient neurons were capable of reaching the olfactory bulb. This study also noted an expansion of Dcx+ cells in the SVZ and RMS, although they did not analyze expression of mature neuronal markers in these cells (Gregorian et al., 2009). Postnatal deletion of Pten in cerebellar granule neurons also showed that a subset of Pten-null granule cells migrated to the proper position in the internal granule layer, whereas many were ectopically located in the molecular layer and at the pial surface (Backman et al., 2001; Kwon et al., 2001). Failed migration of Pten-deficient cerebellar neurons has been suggested to be secondary to defects in the glia guiding their migration (Yue et al., 2005). However, the in vivo electroporation approach used in the present study targeted small numbers of cells and identified stationary Dcx+ neuroblasts without Cre-mediated deletion in surrounding other cell types. Therefore, our results in the RMS are consistent with cell autonomous defects leading to ectopic positioning.

Pten-deficient neuroblasts demonstrated an ‘all or none’ phenotype of normal directional migration or abnormal morphology associated with arrested migration. The absolute proportion of stationary cells will vary over time as the differentiated cells accumulate, and also die. This dichotomy of phenotypes is distinct from a number of other migration defects that caused an expansion of the SVZ and proximal RMS due to ectopic differentiated neurons. For example, deletion of Slit1 showed an expansion of the SVZ due to decreased neuroblast migration speed (Kaneko et al., 2010). Live imaging of migrating neuroblasts showed that Dcx deletion caused an increase in the number of secondary branches in the leading processes, causing a defect in neuroblast speed, but not in direction of migration (Koizumi et al., 2006). The Pten-deficient phenotype also differs from migration defects caused by mutations in other components of the PI3K pathway. Deletion of Tsc1 in the neonatal SVZ, which also activates mTorc1, but induces different feedback signaling compared with Pten loss, caused a decrease in neuroblast migration speed (Huang and Manning, 2009; Feliciano and Bordey, 2012). Deletion of Erbb4, a receptor tyrosine kinase that has been shown to signal through PI3K as well as other downstream pathways, disrupted proper directional orientation, with a significant decrease in the proportion of neuroblasts migrating towards the OB (Anton et al., 2004).

Normal directional migration of Pten-null neuroblasts is consistent with the finding that directional chemotaxis can still occur in Dictyostelium in the absence of a PIP3 gradient (Hoeller and Kay, 2007). Additionally, several studies showed that Pten was not essential in directional migration and chemotaxis of mammalian leukocytes (Lacalle et al., 2004; Ferguson et al., 2007; Nishio et al., 2007; Subramanian et al., 2007; Heit et al., 2008). Pten-deficient neutrophils showed an increased migration speed (Subramanian et al., 2007). Thus, context-dependent signals drive the effect of Pten and PI3K signaling on migration.

Taken together, these data suggest that the ectopic localization of neurons is more likely to be a result of premature differentiation and secondary loss of migration ability rather than intrinsic defects in migration. Indeed, the PI3K-Akt-mTorc1 pathway plays an important role in several crucial neuronal differentiation processes, including neuronal polarity, axon guidance, dendrite arborization and spine morphogenesis (Campbell and Holt, 2001; Jaworski et al., 2005; Kumar et al., 2005; Tavazoie et al., 2005; Chadborn et al., 2006; Kwon et al., 2006; Wildonger et al., 2008; Chow et al., 2009). Direct connections between PI3K signaling and neuronal differentiation were shown in Drosophila retina, in which increased insulin receptor/Tor signaling caused precocious differentiation of photoreceptors, and decreased signaling caused delays in differentiation (Bateman and McNeill, 2004). TOR signaling was also required to initiate neuronal differentiation in the chick neural tube (Fishwick et al., 2010).

The PI3K-Akt-mTorc1 pathway is dynamically regulated through the different developmental stages of neuronal lineage along the SVZ-RMS-OB. From birth in the SVZ to differentiation and integration into the local circuit of the OB, neural stem cells progress in phases to transit-amplifying progenitors, migrating neuroblasts and ultimately terminally differentiated mature neurons. Accordingly, cell signaling pathways must be modulated to meet the varying requirements of these differentiation states or to induce transition between consecutive phases. In this study, we found that the PI3K-Akt-mTorc1 pathway is active in the transit-amplifying cells and inactive in the migrating neuroblasts until it is activated again in the OB. Consistent with these findings, previous studies also found high expression of Egfr, a potent PI3K-Akt-mTorc1 pathway activating receptor, in the transit-amplifying progenitors but not in the migrating neuroblasts in the SVZ (Doetsch et al., 2002; Kim et al., 2009; Kokovay et al., 2010). Upregulation of the PI3K-Akt-mTorc1 pathway might act to drive proliferation of transit-amplifying progenitors, whereas inactivation of PI3K-Akt-mTorc1 signaling through Pten expression might play a role in the transition to migrating neuroblasts that exit the cell cycle. The proliferation consequences downstream of Pten loss are likely to be influenced by the extent of upstream activation of PI3K driven by the growth factor environment at different developmental stages.

In wild-type RMS, the PI3K-Akt-mTorc1 pathway is not activated in Dcx+ cells until they reach the OB, correlating well with the time at which they begin further differentiation. This suggests that the PI3K-Akt-mTorc1 pathway might be involved in the transition from a migratory neuroblast to a differentiating neuron in response to local environmental cues. In the absence of Pten, premature activation of PI3K signaling in neuroblasts in the RMS probably triggers precocious differentiation resulting in ectopic positioning. Inhibition of mTorc1 activity effectively rescued ectopic differentiation of neuroblasts and prevented the expansion of the SVZ and RMS. mTorc1 modulates protein synthesis and cell growth and plays an important role in neuronal differentiation processes (Campbell and Holt, 2001; Jaworski et al., 2005; Kumar et al., 2005; Tavazoie et al., 2005; Kwon et al., 2006; Wildonger et al., 2008; Chow et al., 2009). Taken together, our results, along with previous studies, suggest that proper temporal and spatial regulation of the PI3K-Akt-mTorc1 pathway plays an important role in regulating the appropriate timing and positioning for neuronal differentiation, but is not required for directional tangential neuronal migration.

We thank J. Mitchell and K. Cox for genotyping the St Jude Transgenic Core Facility for expertise in generating Nestin-creER^T2 mice; and the St Jude Cell and Tissue Imaging Shared Resource for confocal microscopy. We thank Tak Mak (University of Toronto) for providing Pten^loxP mice and Pierre Chambon (IGBMC) for Nestin-creER^T2 cDNA.