Reelin-mediated Signaling Locally Regulates Protein Kinase B/Akt and Glycogen Synthase Kinase 3β*

Reelin is a large secreted protein that controls cortical layering by signaling through the very low density lipoprotein receptor and apolipoprotein E receptor 2, thereby inducing tyrosine phosphorylation of the adaptor protein Disabled-1 (Dab1) and suppressing tau phosphorylation in vivo. Here we show that binding of Reelin to these receptors stimulates phosphatidylinositol 3-kinase, resulting in activation of protein kinase B and inhibition of glycogen synthase kinase 3β. We present genetic evidence that this cascade is dependent on apolipoprotein E receptor 2, very low density lipoprotein receptor, and Dab1. Reelin-signaling components are enriched in axonal growth cones, where tyrosine phosphorylation of Dab1 is increased in response to Reelin. These findings suggest that Reelin-mediated phosphatidylinositol 3-kinase signaling in neuronal growth cones contributes to final neuron positioning in the mammalian brain by local modulation of protein kinase B and glycogen synthase kinase 3β kinase activities.

Reelin is a large secreted protein of ϳ400 kDa that is defective in the ataxic reeler strain (1). In Reelin-deficient mice (2) and humans (3), neurons fail to migrate to their proper positions, resulting in abnormal lamination of the neocortex and the hippocampus. Reelin is also needed for the cortical positioning of Purkinje cells, a requirement for granule cell proliferation and foliation in the cerebellum.
Reelin signaling requires binding to two members of the low density lipoprotein (LDL) 1 receptor gene family, the very low density lipoprotein receptor (VLDLR) and the apolipoprotein E receptor 2 (apoER2), on the surface of the migrating neurons (4,5). The phenotype of knockout mice in which both of these Reelin receptors have been inactivated by gene targeting is indistinguishable from that of reeler mice, suggesting that both receptors are obligate components of the Reelin-signaling pathway (6).
Further transmission of the signal is dependent upon the cytoplasmic adaptor protein Disabled-1 (Dab1). Dab1-deficient mice are indistinguishable from reeler and vldlr/apoer2 mutant mice (7)(8)(9). Dab1 interacts with NPXY motifs in the cytoplasmic domains of several LDL receptor family members (10), including VLDLR and apoER2 (6). Reelin binding to VLDLR and apoER2 induces tyrosine phosphorylation of Dab1 (5,11). Replacement of tyrosine residues in Dab1 that are phosphorylated in response to Reelin by phenylalanines in knockin mice abolished Dab1 function (12). Furthermore, mice that lack both Reelin and Dab1 are no more affected than animals that lack only Reelin or Dab1 (11), suggesting that they are components of the same pathway.
Tyrosine phosphorylation of Dab1 allows it to interact with nonreceptor tyrosine kinases including Abl and Src family members, suggesting that Dab1 itself might function as a regulator of tyrosine kinase signaling in the cell (13). The phosphotyrosine binding (PTB) domain of Dab1, which mediates the interaction of the adaptor protein with the NPXY motifs in the cytoplasmic domains of the Reelin receptors, also has an independent affinity for phosphatidylinositol 4-phosphate-or PI 4,5-bisphosphate-rich microdomains in the plasma membrane (14). Both of these phospholipids are substrates for phosphatidylinositol 3-kinase (PI3K), a lipid kinase that is activated by tyrosine kinases at the cytoplasmic leaflet of the plasma membrane (15)(16)(17).
Here we show that Reelin activates a PI3K-dependent signaling cascade in cultured embryonic neurons, resulting in the activation of protein kinase B (PKB; also called Akt) and the inhibition of glycogen synthase kinase-3␤ (GSK-3␤) by phosphorylation on specific regulatory residues. Genetic evidence indicates that activation of Reelin-induced PI3K signaling is dependent on the presence of VLDLR, apoER2, and Dab1. Reelin signaling results in a reduction of cellular levels of specific phosphorylated forms of the microtubule-associated protein tau. Either inhibition of Reelin binding to its receptors or inhibition of PI3K prevents Reelin-induced activation of this pathway. Furthermore, all components of the Reelin pathway are present or enriched in the axonal growth cones, suggesting that leading processes of migrating neurons are sites where positional cues conferred by Reelin are first received.

EXPERIMENTAL PROCEDURES
Materials-PI3K inhibitors wortmannin and LY294002 were purchased from Sigma and Cell Signaling Technology (Beverly, MA), respectively. All other kinase and phosphatase inhibitors were purchased from Calbiochem. All other reagents were from Sigma. Animals deficient for apoER2 (lrp8) and VLDLR (vldlr) were maintained on a mixed C57BL/6J ϫ 129 strain background (6). All wild type, single apoer2 Ϫ/Ϫ and vldlr Ϫ/Ϫ mice, as well as double apoer2 Ϫ/Ϫ ;vldlr Ϫ/Ϫ mice were littermates. Mice homozygous for the reeler spontaneous mutation (reln rl ) were purchased from Jackson Laboratories (Bar Harbor, ME) and were on a B6C3Fe background. Jon Cooper and Brian Howell generously provided Dab1-deficient mice (mixed Balb-c/B6 background) (7). All animals were maintained in accordance with National Institutes of Health and University of Texas Southwestern Animal Resources Center animal care guidelines.
Activity Assays-For PKB and CDK5 activities, 300 g of total protein from cultured neurons were brought to 1 ml in freshly made kinase lysis buffer (0.5% Triton X-100, 0.5% sodium deoxycholate, 50 mM Hepes, pH 7.4, 150 mM NaCl, 5 mM EDTA, 50 mM sodium fluoride, 25 mM sodium pyrophosphate, 80 mM glycerol phosphate, 2 mM sodium vanadate, 0.1 M okadaic acid, and mammalian protease inhibitor mixture (10 l/ml; Sigma). PKB was immunoprecipitated with 2 g of mouse monoclonal anti-PKB antibody (SKB1 clone; Upstate Biotechnology, Inc., Lake Placid, NY) and CDK5 with 2 g of anti-CDK5 antibody (C-8; Santa Cruz Biotechnology). Control immunoprecipitates were done with 2 g of either normal mouse or rabbit IgG. After overnight rotation at 4°C, 20 l of protein G-Sepharose were added, and the tubes were incubated for 2 h. Immunoprecipitates were washed twice with 1 ml of kinase lysis buffer and twice with kinase reaction buffer (KB buffer; 10 mM Hepes, pH 7.4, 10 mM MgCl 2 , 0.1 mM EGTA, 1 mM dithiothreitol).
Immunoprecipitated PKB and CDK5 were assayed in a total of 40 l of KB buffer supplemented with 50 M sodium orthovanadate, 1 M PKI peptide (New England Biolabs, Beverly, MA), 50 nM microcystin, 50 nM okadaic acid, 100 M cold ATP/10 Ci of [␥-32 P]ATP. Tubes containing immunoprecipitated PKB were incubated with 33 M crosstideparamyosin fusion protein as substrate (New England Biolabs) and CDK5 immunoprecipitates with 100 M histone H1 (Calbiochem). After 20 min of incubation at 30°C, reaction mixtures were stopped by adding 20 l of 2ϫ Laemmli buffer. Samples were run on SDS-PAGE. Gels were dried after staining with Coomassie Blue. Kinase activity values were obtained using a Typhoon phosphor imager after overnight exposure. Background kinase activity values from immunoprecipitates with nonimmune control antibodies were subtracted.
Recombinant Reelin-Stably transfected HEK-293 cells expressing recombinant full-length mouse Reelin (18) were kindly provided by Michael Frotscher and Eckardt Förster (University of Freiburg). For a typical batch of Reelin, nontransfected HEK-293 and Reelin-expressing cells were grown on 10 81-cm 2 dishes with ϳ4 ϫ 10 6 cells/dish in 20 ml of Dulbecco's modified Eagle's medium (low glucose with 0.2% bovine serum albumin). Cells were grown for 2 days, and Reelin-or mockconditioned media were collected, centrifuged at 3,000 ϫ g for 15 min, and sterile-filtered. The supernatant was concentrated 20 -40-fold using 100-kDa cut-off centrifugal filters (Millipore Corp., Bedford, MA). To ascertain that activation of the PI3K pathway was dependent on Reelin, concentrated Reelin samples were size-fractionated by gel filtration on Superose 6. Fractions containing full-length Reelin (ϳ50% of total protein in the fractions) were pooled and tested for their ability to stimulate PI3K-dependent phosphorylation of PKB and GSK-3␤. Results from these control experiments were indistinguishable from Reelin-conditioned culture supernatant experiments. Control media from untransfected 293 cells were collected, concentrated, and applied to cultured cells under the exact same conditions and in exactly the same relative concentrations as the Reelin-containing samples.
Neuronal Cultures-Embryonic cortical neurons from mice (E15-E16) or rats (E17-E18) were obtained using standard protocols (19). For wild type mice or rats, embryos were collected at E15. Cortical lobes were isolated, and after removing the meninges, the lobes were chopped into small pieces, pooled, and trypsinized for 15 min at 37°C. Trypsinization was stopped by the addition of fetal calf serum (onetwentieth volume). After centrifugation, cells were washed twice in Hanks' balanced salt solution (Invitrogen). Neurons were dissociated by triturating 40 times in Hanks' balanced salt solution with 12 mM MgCl 2 , 0.025% DNase, 0.4 g/ml trypsin inhibitor, and 2 mg/ml bovine serum albumin with a polished glass Pasteur pipette and transferred to Neurobasal medium containing B27 serum supplement (Invitrogen), penicillin/streptomycin, and 1 mM L-glutamine. Cells were plated at ϳ1000 cells/mm 2 on poly-D-lysine (Sigma)-coated plates. Wild type embryonic neurons were obtained from mouse or from rat embryos and yielded identical results in all experiments. For Dab1 or apoER2/VLDLR-deficient cultures, mouse embryos were treated individually until plating onto cell culture dishes, with a hind portion of the embryo used for genotyping.
Reelin Stimulation-After 3 days in culture, one-half of the neuronal culture medium was exchanged for fresh medium. On day 4 after plating, cells were stimulated either with mock-conditioned medium or with Reelin (see above). The estimated final concentration of Reelin added to the culture medium was 5 nM. Cells were washed once in PBS and then lysed with lysis buffer (PBS with 2 mM EDTA, 1% Triton X-100, 0.25% deoxycholic acid, 0.5% SDS, protease inhibitors (mixture tablets, EDTA-free; Roche Molecular Biochemicals), phosphatase inhibitor mixture 1 and 2 (Sigma)). Lysate was centrifuged for 20 min at 4°C, and supernatants were assayed for protein content.
Specific Inhibitors-The glutathione S-transferase (GST) receptorassociated protein (RAP) fusion protein and GST protein were prepared as described (20). GST-RAP and GST were added at 30 g/ml for 1 h prior to and during Reelin stimulation. PI3K inhibitors LY294002 (catalog no. 9901; Cell Signaling Technology) and Wortmannin (W1628; Sigma) were applied at the indicated concentrations 1 h prior to Reelin stimulation and for duration of the assay.
Isolation of Growth Cone Particles by Subcellular Fractionation-Fetal rat brain (gestational stage E18) was fractionated as described by Pfenninger et al. (21) to obtain growth cone particles. Briefly, the low speed supernatant of fetal brain homogenate was loaded on a discontinuous sucrose gradient in which the 0.75 and 1 M sucrose layers were replaced with a single 0.83 M sucrose step. This facilitated collection of the interface and increased growth cone particle yield without decreasing purity (22). The 0.32 M/0.83 M interface was collected, diluted with 0.32 M sucrose, and centrifuged to yield the GCP fraction, which was resuspended in 0.32 M sucrose for use in experiments.
Immunocytochemistry-Cortical neurons were cultured in Dulbecco's modified Eagle's medium plus B27 supplement (Invitrogen) for 3 days on dishes coated with 0.3 mg/ml poly-D-lysine. After stimulation with mock-conditioned media or with Reelin, cells were processed for immunocytochemistry. Briefly, cells were first washed with warm PHEM buffer (60 mM Pipes, 25 mM Hepes, 10 mM EGTA, 1 mM MgCl 2 , pH 7.4) and fixed in 2% paraformaldehyde, 0.01% glutaraldehyde in PHEM at 37°C for 10 min. After fixation, cells were washed three times with PBS, extracted with 0.2% Triton X-100 for 10 min, and blocked in 10% goat serum, 5% bovine serum albumin for 1 h. Primary antibodies were diluted in block solution and incubated overnight at 4°C. Immunohistochemistry-specific polyclonal rabbit anti-phospho-PKB (Cell Signaling Technology) and 4G10 antibody were used at a 1:100 dilution. Monoclonal anti-tubulin (DM1A clone) was used at a 1:400 dilution.
Secondary antibodies were goat anti-mouse IgG-labeled with OG488 or goat anti-rabbit Texas Red (Molecular Probes, Inc., Eugene, OR) used at a 1:400 dilution in block solution. Images were acquired with an Orca CCD camera (Hamamatsu) controlled by OpenLab software (Improvision) and processed for presentation in Adobe Photoshop.

RESULTS
To identify the biochemical signaling pathways that are activated by Reelin signaling in the embryonic brain, we developed a sensitive in vitro assay in which we could follow the regulation of cellular kinase activities in whole cell lysates in response to Reelin exposure. To achieve this, it was necessary to reduce Reelin-independent cellular signal input that would otherwise differentially affect the various pools of these kinases and thereby obscure any Reelin-specific effect. This was accomplished by culturing primary embryonic neurons for several days in a minimal medium before stimulation with recombinant Reelin. Reelin-independent kinase activation, for instance by cell adhesion molecules such as integrins, was minimized by plating the cells on polylysine instead of other commonly used mixtures of extracellular matrix components.
Reelin was partially purified from the supernatant of stably transfected 293 cells by column chromatography and size exclusion filtration. Conditioned medium from nontransfected 293 cells served as a control in all experiments. Exposure of cultured neurons to Reelin (Fig. 1A, lane 2) greatly increased the levels of phosphorylated Dab1 in immunoprecipitates (upper panel) as well as in whole cell lysates (middle panel) compared with mock medium (lane 1). Since tau is hyperphosphorylated in animals genetically deficient in Reelin signaling (5), we investigated whether Reelin can directly reduce tau phosphorylation in primary neuronal cultures. Immunoblot analysis of extracts with an antibody directed against the tau phospho-Ser 202 /Thr 205 epitope revealed a marked reduction of phospho-tau in response to Reelin. This suggested that one or more kinases that phosphorylate tau at this site are inhibited as a result of Reelin signaling.
Two multifunctional neuronal kinases implicated in tau phosphorylation are CDK5 and GSK-3␤. The activity of the latter is inhibited by phosphorylation of a serine at position 9 by PKB. Enzymatic activity of the total cellular pool of CDK5 was unaffected by Reelin (93.2 Ϯ 4.5% of control). However, the activity of PKB, the primary cellular kinase inhibiting GSK-3␤, was increased 2-fold in the same cells (205.6 Ϯ 5.0% of control). This suggested that PKB and GSK-3␤ might be part of a Reelin signaling cascade. A time course experiment (Fig. 1B) showed that Dab1 phosphorylation reached maximal levels after 20 min of Reelin stimulation (lane 4). Phosphorylation of both GSK-3␤ and PKB showed a similar pattern of activation, with maximal phosphorylation by 20 -30 min and a clear reduction by 60 min. Total levels of GSK-3␤ and PKB remained unchanged under the same conditions. Enzymatic activity of PKB, but not CDK5, was stimulated by Reelin and correlated with phosphorylation of PKB on Ser 473 . All subsequent experiments were thus carried out for 20 min.
Maximal activation of PKB requires phosphorylation at two sites: Ser 473 and Thr 308 . To determine whether Reelin differentially affected the phosphorylation of PKB at these sites, we performed the experiment shown in Fig. 1C. Primary embryonic neurons were unstimulated (lanes 1 and 2), exposed to mock medium (lanes 3 and 4), or exposed to Reelin (lanes 5 and 6). Whereas Thr 308 phosphorylation was unaffected by any of these conditions, Ser 473 phosphorylation was induced only when Reelin was present (lanes 5 and 6) and correlated with Dab1 phosphorylation in both rat and mouse primary cultured neurons (Fig. 1, B and C, respectively). Total levels of Dab1, PKB, and GSK-3␤ remained unchanged with Reelin treatment.
RAP universally blocks the binding of ligands to LDL receptor family members (20,23). RAP also blocks the Reelin-induced phosphorylation of Dab1 in primary neurons by preventing Reelin binding to apoER2 and VLDLR (5). To determine whether the phosphorylation of PKB and GSK-3␤ requires Reelin binding to both receptors and subsequent Dab1 phosphorylation, we incubated primary mouse neurons in the presence (Fig. 2a, lanes 1, 4, and 6) or absence (lanes 2, 3, 5, and 7) of Reelin, GST-RAP fusion protein (lanes 3-5), or the GST control protein (lanes 6 and 7). A similar experiment was performed in primary rat neurons (Fig. 2b). GST-RAP effectively blocked Reelin-induced Dab1 phosphorylation (Fig. 2a, lane 4), whereas GST-RAP alone had no effect (Fig. 2a, lanes 3 and 5). Reelin-stimulated phosphorylation of PKB at Ser 473 was mark-  1, 3, 5, and 7) or with Reelin (lanes 2, 4, 6, and 8) for up to 60 min. Tyrosine phosphorylation of Dab1 is maximal after 20 min of stimulation. Reelin also induced serine phosphorylation of GSK-3␤ at residue 9 and PKB at serine 473. Immunoblots of total GSK-3␤ and PKB showed no change across time or Reelin treatment. Phosphorimaging analysis of 32 P incorporated into crosstide-paramyosin fusion protein by PKB (PKB Activity) was stimulated in a Reelin-dependent manner, whereas incorporation into Histone H1 by CDK5 (CDK5 Activity) appeared unaffected by Reelin treatment. C, treatment of mouse embryonic neurons with mock media (lanes 3 and 4) did not alter the phosphorylation or total level of any of the proteins examined compared with untreated (lanes 1 and 2) control neurons. Stimulation with Reelin (lanes 5 and 6) for 20 min significantly induced Dab1 phosphorylation as determined by immunoblotting with anti-phosphotyrosine antibody 4G10. Reelin greatly induced phosphorylation of PKB at Ser 473 but not at Thr 308 , whereas total levels of PKB remained unaffected. GSK-3␤ phosphorylation at Ser 9 was also increased by Reelin treatment. Total levels of GSK-3␤ remained unchanged under any condition. edly reduced by GST-RAP (Fig. 2a, compare lane 4 with lanes  1 and 6; Fig. 2b, compare lanes 2 and 6 with lane 4). GST control protein did not affect Reelin-induced phosphorylation of Dab1, PKB, and GSK-3␤ (Fig. 2, a and b). Phosphorylation of GSK-3␤ at Ser 9 in response to Reelin was also reduced in the presence of RAP (Fig. 2b, lane 4). Total levels of Dab1, PKB, and GSK-3␤ were comparable between the samples and unaffected by the various conditions (Fig. 2, a and b).
Since RAP will inhibit ligand binding to all LDL receptor family members, we designed a genetic approach to determine whether apoER2 and VLDLR specifically were responsible for the Reelin-mediated phosphorylation of PKB and GSK-3␤. To achieve this, we isolated neurons from mice lacking various numbers of alleles for both receptors. Since apoER2/VLDLR double deficient mice do not survive to breeding age (6), we used crosses between mice that lacked three of the four receptor alleles. This generated pregnancies in which 25% of the embryos were deficient for both receptors, with remaining embryos containing one or two wild type alleles to serve as controls. Fig. 3 shows that Reelin is capable of stimulating phosphorylation of Dab1 or PKB, as long as the embryos from which the neurons had been isolated retained at least one functional apoER2 or VLDLR allele (Fig. 3a). In mice that were completely deficient in apoER2 and VLDLR, no appreciable Reelinstimulated phosphorylation of either Dab1 or PKB was detected (Fig. 3a). Total levels of PKB or Dab1 (not shown) were not changed by treatment with mock or Reelin.
Dab1 is one of several cytoplasmic adaptor proteins that bind to the intracellular domains of apoER2 and VLDLR (24,25). To determine whether Dab1 or some other adaptor or scaffolding protein were required for Reelin-mediated activation of the PKB pathway, we prepared neurons from embryos that were wild type, heterozygous, or homozygous for the dab1 knockout allele. As expected, no phosphorylated or total Dab1 protein was detected in dab1 Ϫ/Ϫ neurons (Fig. 3b, lanes 1 and 2), and Reelin-induced phosphorylation of PKB on Ser 473 or of GSK-3␤ on Ser 9 occurred only in cells that contained at least one functional dab1 allele (lanes 3-6). Total levels of PKB and GSK-3␤ were independent of dab1 genotype or Reelin treatment.
To determine whether Reelin stimulation of PKB and inhibition of GSK-3␤ required activation of PI3K, we incubated neurons with Reelin in the presence or absence of PI3K inhibitors: LY294002 (Fig. 4A) or wortmannin (Fig. 4B). Dab1 phosphorylation by Reelin was unaltered whether or not LY294002 was present (Fig. 4A). However, Reelin-induced phosphorylation of both PKB and GSK-3␤ were completely blocked by LY294002 (lanes 5 and 6), although total levels of both proteins remained unchanged. Wortmannin also markedly inhibited phosphorylation of PKB at Ser 473 (Fig. 4B). Activation of PI3K and subsequent phosphorylation of PKB can also be achieved by inactivation of a negative regulator of PI3K, the lipid phosphatase PTEN. However, no changes in the phosphorylation state of PTEN at its regulatory Ser 380 residue were observed under any conditions (data not shown), suggesting that PTEN protein was not involved in Reelin signaling.
Migrating neurons in the mammalian brain probably receive extracellular signals through their leading processes or growth cones. We found that the Reelin receptor apoER2 and the adaptor protein Dab1 as well as PI3K were relatively enriched in purified axonal growth cones from fetal rat brains (Fig. 5A). PKB was also present in growth cones but appeared similarly distributed between growth cones and total cell lysate. The enrichment of nonphosphorylated tau and GAP-43 and the relative depletion of the dendritic marker MAP2 confirm the high degree of enrichment of axonal growth cones that was achieved in this preparation (22). To determine whether Reelin treatment preferentially affected signaling components in growth cones, we used immunocytochemistry to detect activated components of the Reelin signaling pathway in growth cones of cortical neurons after 3 days in culture. Compared with mock-treated neurons, anti-phosphotyrosine immunoreactivity was specifically increased in growth cones and in the axons of Reelin-treated neurons, presumably because of retrograde transport of tyrosine-phosphorylated Dab1 (Fig. 5, B-D, mock; E-G, Reelin-treated). Dab1 is concentrated in growth cones and was the only protein in total neuronal lysates whose phosphorylation on tyrosine residues dramatically increased with Reelin (Fig. 1A). Furthermore, the levels of all other tyrosine-phosphorylated proteins were unchanged in total extracts from primary embryonic neurons (Fig. 1A), making it likely that the increase in anti-phosphotyrosine immunoreactivity in the growth cones is primarily due to Dab1 phosphorylation. Similarly, phospho-PKB immunoreactivity at Ser 473 was preferentially increased in growth cones with Reelin treatment (Fig. 5, H and J) and enriched in filopodia as compared with microtubules (Fig. 5, I and J). These data indicate that Reelin can specifically activate the PI3K/PKB pathway in growth cones of embryonic cortical neurons. DISCUSSION Genetic and biochemical evidence places the signaling protein Reelin, the neuronal cell surface receptors VLDLR and apoER2, and the cytoplasmic adaptor protein Dab1 in a linear  1, 4, and 6) for 20 min in the presence (lanes 3-5) or absence (lanes 1, 2, 6, and 7) of GST-RAP. RAP markedly reduced Reelin-dependent phosphorylation of both PKB at Ser 473 and Dab1 (lane 4) compared with controls (lanes 1 and 6). GST alone did not affect Reelin-induced phosphorylation of PKB or Dab1 (lane 7). Total levels of PKB and Dab1 protein were unaffected by Reelin or RAP treatment. b, rat embryonic neurons were stimulated with Reelin for 20 min (lanes 2, 4, and 6), treated with mock-conditioned medium (lanes 1, 3, and 5) or not treated (lane 7). GST-RAP decreased Reelin-mediated phosphorylation of PKB and GSK-3␤ (lane 4) to levels comparable with mocktreated samples that were also exposed to GST-RAP (lane 3). The addition of GST did not affect Reelin-induced protein phosphorylation (lanes 5 and 6).
pathway that is critical for neuronal migration and proper organization of the mammalian brain (2, 4 -9). Reelin-induced Dab1 phosphorylation in primary cortical neurons (11) can be inhibited by blocking Reelin binding to the lipoprotein recep-tors apoER2 and VLDLR (5). Here we provide genetic and biochemical evidence that the signaling cascade induced by Reelin downstream of Dab1 inhibits GSK-3␤ and suppresses phosphorylation of the microtubule-associated protein tau. This cascade involves PI3K and PKB kinase activities acting downstream from and dependent on the presence of apoER2, VLDLR, and Dab1.
Our results suggest a model, shown in Fig. 6, whereby Reelin would initially bind to a protein complex on the surface of migrating neurons that contains as obligate components one or both of the lipoprotein receptors apoER2 and VLDLR. A likely place where this might occur is the axonal growth cone, the leading edge of the migrating neuron. Growth cones are the first to make contact with the Reelin-expressing Cajal-Retzius neurons at the border between the cortical plate and the marginal zone (26), and several components of the Reelin signaling complex are enriched in growth cones (Fig. 5). An as yet unidentified tyrosine kinase would then be recruited to the receptor-signaling complex, leading to phosphorylation of Dab1 bound to the cytoplasmic tails of VLDLR and apoER2. Tyrosine phosphorylation of Dab1 is required to activate PI3K, which in turn leads to stimulation of PKB by phosphorylation on Ser 473 and inactivation of GSK-3␤ by phosphorylation on Ser 9 . Activation of PI3K and PKB-mediated signaling pathways can impact on a broad range of cellular functions (15)(16)(17), including gene transcription, protein translation, metabolism, axonal transport, cell survival, growth, and migration. Specificity is probably maintained by temporal as well as spatial restriction of the signal (i.e. primarily to the growth cone). In our model, Reelin-induced suppression of GSK-3␤ activity may block phosphorylation of GSK-3␤ substrates (e.g. the microtubule-associated protein tau or the microtubule motor kinesin) (27). Conversely, failure to suppress GSK-3␤ activity of a specific GSK-3␤ pool (for instance by defects in Reelin, VLDLR, apoER2, or Dab1) could explain the observed increases in tau phosphorylation in these different mutant mouse strains (5) as well as changes in other GSK-3␤-sensitive processes.
This model is based on several lines of evidence. First, Reelin induces phosphorylation of Dab1, PKB, and GSK-3␤, an event that requires the presence of VLDLR and apoER2 and binding of Reelin to these receptors. Second, cells deficient in Dab1 are unable to induce phosphorylation of PKB or GSK-3␤ in response to Reelin. Third, inhibitors of PI3K such as wortmannin and LY294002 inhibit Reelin-induced phosphorylation of PKB and GSK-3␤ but not of Dab1. Finally, several mediators of FIG. 3. Genetic deficiency of apoER2, VLDLR, and Dab1 prevents Reelin-induced PKB phosphorylation. a, mouse embryonic neurons were prepared from E16 embryos derived from two separate matings of apoer2 Ϫ/Ϫ ;vldlr ϩ/Ϫ and apoer2 ϩ/Ϫ ;vldlr Ϫ/Ϫ mice (lanes 1-6 and 7-14, respectively), generating the genotypes shown at the bottom of the panel. Stimulation with Reelin for 20 min increased phosphorylation of Dab1 and PKB at Ser 473 in neurons that retained at least one wild type allele (lanes 1-4 and 7-12). No stimulation of Dab1 or PKB phosphorylation was observed in neurons derived from double receptor knockout cells (lanes 5, 6, 13, and 14). Each pair of samples represents neurons from a single embryo. b, mouse embryonic neurons were prepared from E16 embryos from a dab1 ϩ/Ϫ ϫ dab1 ϩ/Ϫ mating, generating the three genotypes indicated at the bottom. In wild type (lanes 5 and 6) or heterozygous (lanes 3 and 4) neurons, Reelin induced phosphorylation of Dab1, PKB, and GSK-3␤. No induction of PKB or GSK-3␤ phosphorylation was observed in cells lacking Dab1 protein (lanes 1 and 2).  1 and 4), exposed to mock medium (lanes 2 and 5), or induced with Reelin (lanes 3 and 6) for 20 min. LY294002 markedly reduced Reelin-induced phosphorylation of PKB and GSK-3␤, whereas Dab1 phosphorylation remained unaffected. Total protein levels for Dab1, PKB, and GSK-3␤ remained unchanged by LY294002 treatment, although background phosphorylation levels of PKB and GSK-3␤ were reduced. B, mouse embryonic neurons (E15) were either preincubated (lanes 3 and 4) with the PI3K inhibitor wortmannin (100 nM) or not (lanes 1 and 2) and exposed to Reelin (lanes 2 and 4) or mock-conditioned medium (lanes 1 and 3). Reelin-induced PKB phosphorylation was markedly reduced by Wortmannin treatment, but phosphorylation of Dab1 was unaffected.
Reelin signaling, including apoER2 and Dab1, are highly enriched in axonal growth cones.
Reelin-induced Phosphorylation of PKB and GSK-3␤ Requires ApoER2, VLDLR, and Dab1-The present results provide genetic (Fig. 3) and biochemical (Fig. 2) evidence that apoER2 and VLDLR are required for Reelin-induced phosphorylation of Dab1, PKB, and GSK-3␤. They do not exclude a role for co-receptors in the process. Reelin-induced phosphorylation of Dab1, PKB, and GSK-3␤ was greatly reduced by RAP, an inhibitor of Reelin binding to both receptors (Fig. 2). Furthermore, PKB phosphorylation was abolished in neurons from mice in which both apoER2 and VLDLR had been genetically disrupted (Fig. 3a). In contrast, cells that contained at least one remaining receptor allele (either VLDLR or apoER2) were still able to activate PKB to some extent. This result correlates with our finding that only those mice that lack all four receptor alleles show the same severe neuronal migration defects as the reeler mutants (6,28), whereas those missing only three alleles develop milder phenotypes similar to the respective single receptor knockouts.
Reelin-mediated activation of PKB and phosphorylation of GSK-3␤ was also dependent on Dab1 (Fig. 3b). Cells that were heterozygous for Dab1 retained the ability to phosphorylate PKB and GSK-3␤ in response to Reelin. Taken together, these results show that apoER2, VLDLR, and Dab1 are required for Reelin-induced regulation of PKB-and GSK-3␤-mediated signaling and that activation of this pathway appears to be necessary for proper neuronal positioning in the developing mammalian brain.
Reelin Signaling Requires PI3K Activity-PKB activity can be regulated by at least two distinct mechanisms: directly through PI3K activation or by inactivation of PTEN, a phosphatase that hydrolyzes the PI3K product phosphatidylinositol 3,4,5-trisphosphate (29). No changes were observed at the Ser 380 phosphorylation site in PTEN, which regulates stability of this enzyme (data not shown). Our results using the PI3K inhibitors wortmannin and LY294002, however, demonstrate that PI3K activity is required for phosphorylation of PKB at FIG. 5. Components of the Reelin signaling pathway are present in growth cones. A, total cell lysate and axonal growth cones were prepared from E18 rat brain as described above. 10 g of protein were loaded in each lane and immunoblotted for the indicated proteins. ApoER2, DAB1, and PI3K are all enriched in growth cone particles relative to total lysate, and PKB is readily detectable, indicating that components of the Reelin pathway are present in growth cones. GAP43 and tau are also enriched in growth cones to different extents, whereas actin is present but not enriched in growth cones relative to lysate. Growth cone particles are primarily derived from growing axons, since MAP2, a dendritic marker, is depleted in this fraction relative to total lysate. B-G, Primary rat cortical neurons were cultured and immunostained for phosphotyrosine after exposure to mock medium (B-D) or to Reelin  Reelin binds directly to apoER2 and VLDLR on the neuronal surface, probably in axonal growth cones. RAP can block signaling by inhibiting Reelin binding to the receptors. Dab1 is recruited to the cell surface by binding to phospholipids as well as to the cytoplasmic NPXY motif in both receptors. Dab1 tyrosine phosphorylation is thought to involve a Reelin co-receptor with intrinsic or associated kinase (K) activity. Phosphorylated Dab1 recruits and activates nonreceptor tyrosine kinases such as Src, Fyn, and Abl that act on downstream components of the signaling pathway. Dab1 phosphorylation results in the activation of PI3K, which can be inhibited by wortmannin and LY294002. This leads to activation of PKB and inactivation of GSK-3␤, which can act on various downstream targets, including the microtubule-associated protein tau and the microtubule motor kinesin. Another tau kinase, CDK5, and its activators, p35 and p39, are also involved in the regulation of neuronal migration, although currently available data suggest that CDK5 activity is not directly regulated by Reelin. Ser 473 , since both kinase inhibitors effectively eliminated the Reelin-induced phosphorylation of PKB (Fig. 6). Consistent with this finding is a recent report that shows that PI3K activity is required for migration of cortical interneurons (30).
Neither PI3K inhibitor affected the ability of Reelin to induce Dab1 phosphorylation, indicating that Dab1 phosphorylation precedes activation of PI3K in this sequence of events. Phosphorylation of PKB at Thr 308 , a second site that is important for its activation, was not affected by Reelin. These findings identify Ser 473 phosphorylation of PKB as a distinct branch of a PI3K-dependent signaling pathway that is controlled by Reelin in embryonic neurons.
The sequence of signaling events initiated by Reelin as proposed here is naturally linear. However, most signaling pathways are multifaceted, branching and converging at various points. For example, CDK5, another kinase that can phosphorylate tau, is also implicated in neuronal migration (31,32). Mice deficient in CDK5 or its activators p35 and p39 display severe neuronal migration defects similar to but distinct from those seen in reeler mice (32,33). We observed no changes in CDK5 activity in any of the mutant mouse strains or in response to Reelin stimulation of primary neuronal cultures. These findings suggest that CDK5 is not a direct downstream target of Reelin and its receptors, although it may act on a common target.
The identity of such a target remains to be determined. Tau hyperphosphorylation could be a causal event for neuronal migration defects or a secondary consequence. GSK-3␤ and CDK5 are major kinases responsible for normal and pathological phosphorylation of tau protein in vivo, and Reelin appears to affect tau phosphorylation through GSK-3␤. However, tau knockout animals display normal brain organization with no indication of neuronal migration defects (34). This suggests that changes in tau phosphorylation might occur parallel to alteration of neuronal migration patterns.
Interestingly, both CDK5 and GSK-3␤ have been implicated in regulation of kinesin-driven motility in neurons. GSK-3␤ specifically inhibited anterograde, but not retrograde, fast axonal transport in vivo (27). GSK-3␤ was enriched in neuronal growth cones, where delivery of membrane proteins and Reelin signaling occurs. Similarly, inhibition of CDK5 specifically inhibits anterograde fast axonal transport (35). The essential role of fast axonal transport in neuronal function and the finding that both GSK-3␤ and CDK5 play a role in regulating fast axonal transport suggest that regulation of these pathways by extracellular cues such as Reelin may be a key step in neuronal migration. Consistent with this model, GSK-3␤ in growth cones has been implicated in signaling by the axonal growth inhibitor protein semaphorin 3A (36). Regulation of GSK-3␤ through extracellular cues such as Reelin and semaphorins could modulate delivery of newly synthesized material to a specific subcellular compartment.
In conclusion, our results provide unequivocal genetic evidence that Reelin activates a PI3K-dependent signaling cascade through the lipoprotein receptors apoER2 and VLDLR and the adaptor protein Dab1. This activation occurs in axonal growth cones, suggesting that integration of the Reelin signal at the leading processes of migrating neurons may be crucial for activating the molecular transport machinery that moves the cell body to its final position, thereby determining the lamination of the mammalian cortex.