Persistence of the cell-cycle checkpoint kinase Wee1 in SadA- and SadB-deficient neurons disrupts neuronal polarity

The establishment of neuronal polarity results in the specification of a single neurite as the axon while the remaining neurites that also have the potential to become an axon are prevented from acquiring axonal properties (Bradke and Dotti, 2000; Da Silva and Dotti, 2002). Axon formation depends on a pathway that includes R-Ras, phosphoinositide 3-kinase, mTOR, Rap1B and several additional factors that initiate the rapid growth of a single neurite (Arimura and Kaibuchi, 2007; Choi et al., 2008; Li et al., 2008). In addition, the serine/threonine kinases LKB1, SadA and SadB are required for the formation of axons (Barnes et al., 2007; Kishi et al., 2005; Shelly et al., 2007). LKB1 phosphorylates and activates the Sad kinases (Barnes et al., 2007; Lizcano et al., 2004; Shelly et al., 2007). SadA phosphorylates the microtubule-associated protein (MAP) tau at S262 and neurons lacking Sad kinases show a decrease in tau phosphorylation (Barnes et al., 2007; Kishi et al., 2005). Although the loss of either SadA or SadB alone does not affect neuronal differentiation, Sada^−/−;Sadb^−/− double mutant mice show defects in axon formation. Cultured hippocampal neurons from Sada^−/−;Sadb^−/− double knockout mice develop multiple long neurites that are positive for both axonal (Tau-1 staining) and dendritic (MAP2) markers instead of a single axon (Kishi et al., 2005). The cortex of Sada^−/−;Sadb^−/− mice shows increased Tau-1 immunoreactivity, indicating defects in neurite differentiation similar to those seen in cultured neurons (Kishi et al., 2005).

In addition to its function in neurons, human SADA mediates G2-M arrest after UV-induced DNA damage in HeLa cells by phosphorylating WEE1 at S642 (Lu et al., 2004). Wee1 is a highly conserved kinase that regulates the entry into mitosis by inhibiting Cdc2 activity through the phosphorylation of cyclin-dependent kinase 1 (Cdk1) at Y15 (Harvey et al., 2005; Kellogg, 2003; McGowan and Russell, 1993; McGowan and Russell, 1995; Watanabe et al., 2005; Watanabe et al., 2004; Watanabe et al., 1995). A null mutation in Wee1 results in cell-cycle defects and cell death before E3.5 in mice (Tominaga et al., 2006). Wee1 activity is regulated by inhibitory and stimulatory phosphorylation at multiple sites (Harvey et al., 2005; Kellogg, 2003; Kim and Ferrell, 2007).

Here we show that the Sad kinases are required to initiate the downregulation of Wee1 in polarized neurons. Wee1 is expressed in postmitotic neurons during early stages of differentiation but is no longer detectable after neuronal polarity is established. Although Wee1 is not required for the formation of axons, it impairs neuronal polarity in hippocampal neurons when its expression is maintained. Overexpression of Wee1 or the loss of SadA and -B result in the formation of multiple neurites positive for both axonal and dendritic markers. Suppression of Wee1 rescues the phenotype of neurons deficient in SadA and SadB (SadA/B) and restores normal polarity. Our results identify a new function of Sad kinases: limiting the activity of Wee1 after neuronal polarity is established.

To investigate if Wee1 is a target of Sad kinases not only in dividing cells (Lu et al., 2004) but also in post-mitotic neurons, we analyzed the expression of Wee1 and Sad kinases in rat hippocampal neurons. Hippocampal neurons first develop several short, undifferentiated neurites (stages 1-2) (Bradke and Dotti, 2000; Da Silva and Dotti, 2002). At the transition from stage 2 to stage 3, one of these neurites is selected to become an axon and extends rapidly. At stage 3, neurons have developed a single Tau-1-positive axon and several MAP2-positive minor neurites that eventually differentiate into dendrites (stage 4). Both Wee1 and SadA could be detected in hippocampal neurons by immunofluorescence and western blotting (Fig. 1, supplementary material Fig. S1A). Wee1 was detectable between 0.5 (stage 1) and 3 (stage 3) days in vitro (d.i.v.) but was no longer present from 4 d.i.v. on (stage 4; Fig. 1). Wee1 was concentrated in the perinuclear region but undetectable in neurites.

In contrast to Wee1, SadA/B were barely detectable during the earliest stages of neuronal differentiation in unpolarized neurons between 0.5 and 1.5 d.i.v. (stage 1 and 2; Fig. 1, supplementary material Fig. S1E). Both Sad kinases could be detected in hippocampal neurons starting at 2 d.i.v. (stage 3) and showed a punctate staining pattern in the soma and in axonal and dendritic processes at 3 d.i.v. consistent with published results (Kishi et al., 2005). The specificity of the anti-SadA and -SadB antibody was confirmed by western blotting and by staining neurons prepared from Sada^−/− and Sadb^−/− mutants (supplementary material Fig. S1B-D). Thus, Wee1 and SadA/B show an overlapping expression pattern. Whereas Wee1 is already present during the earliest stages of neuronal development and is downregulated in stage 3 neurons, expression of SadA/B becomes detectable at stage 2 and remains present at later stages. Only stage 3 neurons (2-3 d.i.v.) transiently coexpress Sad kinases and Wee1 for a short period of time.

To investigate the function of Wee1 in neurons, we suppressed endogenous Wee1 by RNA interference (RNAi) using expression vectors for small hairpin RNAs (shRNAs). The efficiency of the constructs was confirmed by transfection of HEK293T cells with vectors for Wee1 and the shRNAs and by transfection of hippocampal neurons with the Wee1 RNAi constructs (supplementary material Fig. S2A,B). When neurons were transfected with the RNAi construct at 0.5 days in vitro (d.i.v.), neuronal polarity was not affected at 3 d.i.v. (Fig. 2A,B, supplementary material Fig. S2C-E). Neurons developed a single Tau-1-positive axon with a slightly increased length. However, the microtubules in these neurons were disorganized and failed to extend minor neurites. A second shRNA targeted at a different sequence in the Wee1 transcript showed the same effect (supplementary material Fig. S2A,F,G). Expression of an RNAi-resistant Wee1 construct rescued the loss of endogenous Wee1, confirming the specificity of the RNAi vector (supplementary material Fig. S2H-I). As an additional control, we used an shRNA with two mismatches to the target sequence (Wee1 shRNAmut). Expression of this control shRNA at 0.5 or 3 d.i.v. had no effect (supplementary material Fig. S2C,H). Consistent with the absence of Wee1 in polarized neurons, transfection of hippocampal neurons with the Wee1 shRNA construct at 3 d.i.v. did not affect neuronal differentiation (Fig. 2B).

To test if the downregulation of Wee1 at 4 d.i.v. is important for the normal differentiation of neurons, we transfected hippocampal neurons with an expression vector for Wee1 at 0.5 (stage 1) and 3 d.i.v. (stage 3). If the absence of Wee1 is essential for neuronal differentiation, its expression should induce a gain-of-function phenotype only at 5 d.i.v. but not at 3 d.i.v. when endogenous Wee1 is still present. Transfection at stage 1 did not have an effect on neuronal morphology when neurons were analyzed at 3 d.i.v. (Fig. 2B, supplementary material Fig. S3). However, Wee1 expression disrupted neuronal polarity when neurons were transfected at stage 3 and analyzed at 5 d.i.v. (Fig. 2B-D). At 5 d.i.v., the majority of transfected neurons (58.7±2.3%; n=69) formed multiple neurites that were positive for axonal (Tau-1) and in many cases also dendritic (MAP2) markers after Wee1 overexpression (Fig. 2B,C). These neurons extended multiple neurites of similar length that were significantly longer than dendrites and were classified as indeterminate neurites because they showed properties of both axons and dendrites (Fig. 2C,D, supplementary material Fig. S3B) (Schwamborn et al., 2006). A little more than a third (39.7±0.5%) of the neurons extended several Tau-1-positive neurites that were longer than 100 μm (control: 115.2±2.2 μm, n=52; Wee1: 109.2±1.2 μm), 19.0±1.8% of the neurons formed multiple Tau-1 RNAi-positive neurites of similar length and shorter than 100 μm (88.7±1.5 μm). This phenotype was similar to that observed for neurons deficient for SadA and B (Kishi et al., 2005). Expression of Myt1, which is closely related to Wee1, had no effect on neuronal polarity (supplementary material Fig. S2C,D). Thus, Wee1 is required only at early stages of neuronal differentiation. It becomes dispensable after neuronal polarity is established but disrupts polarity when expression is maintained in polarized neurons.

To analyze the regulation of Wee1 by Sad kinases, we generated constitutively active (SadAca) and inactive mutants of SadA (SadAki). SadAca mimics the phosphorylation by LKB1, which activates Sad kinases (Barnes et al., 2007; Lizcano et al., 2004). Using tau-GFP as a substrate, we confirmed that SadAki is inactive and SadAca is constitutively active, displaying an activity similar to that after coexpression of SadA and LKB1 (Fig. 3A). SadA and SadB interact with Wee1 in neurons as demonstrated by immunoprecipitation from lysates of hippocampal neurons (Fig. 3B). The activity of Wee1 is regulated by inhibitory and stimulatory phosphorylation at multiple sites including serine123 and S642 (Harvey et al., 2005; Kellogg, 2003; Kim and Ferrell, 2007). Upon entry into M phase, Wee1 is inhibited by hyperphosphorylation at several serine residues including S123 (Kellogg, 2003; Kim and Ferrell, 2007). Following phosphorylation, Wee1 is ubiquitinated and degraded through the ubiquitin/proteasome system (Watanabe et al., 2005; Watanabe et al., 2004; Watanabe et al., 1995). As a consequence of S642 phosphorylation, both a stimulation of Wee1 kinase activity and the inhibition of its function were reported (Katayama et al., 2005; Lee et al., 2001; Lu et al., 2004; Rothblum-Oviatt et al., 2001; Wang et al., 2000). To confirm that SadA phosphorylates Wee1, SadAca and Wee1 were coexpressed in HEK293T cells and phosphorylated Wee1 immunoprecipitated with an anti-phospho-serine antibody (Fig. 3C). This experiment showed that SadAca phosphorylates Wee1 whereas mutation of S642 to alanine (Wee1S642A) abrogated Wee1 phosphorylation. To investigate if SadA phosphorylates Wee1 at S642, both proteins were coexpressed in HEK293T cells and analyzed with an antibody specific for Wee1 phosphorylation at S642 (Fig. 3D). Expression of Wee1 by itself or together with inactive SadAki resulted in a weak phosphorylation of Wee1 by endogenous kinases. Coexpression of SadAca strongly increased Wee1 phosphorylation whereas no signals were detectable for Wee1S642A.

To investigate if the phosphorylation of Wee1 depends on Sad kinases in vivo, we analyzed Wee1 phosphorylation in SadA/B-deficient mouse mutants (Kishi et al., 2005). Because only a very small number of Sada^−/−;Sadb^−/− embryos could be recovered at embryonic day (E)18, Sada^−/−;Sadb^+/− neurons were used for these experiments. The phenotype of neurons derived from Sada^−/−;Sadb^−/− and Sada^−/−;Sadb^+/− embryos was indistinguishable (supplementary material Fig. S4A-C), indicating that one copy of Sadb is not sufficient for neuronal differentiation in the complete absence of Sada. 70.4±1.3% of Sada^−/−;Sadb^+/− neurons formed multiple neurites that were longer than minor neurites (wild type: 10.4±0.7 μm; supplementary material Fig. S3) and slightly shorter than axons (wild type: 101.2±4.1 μm; Sada^−/−;Sadb^+/−: 88.3±3.7 μm). Many of these neurites were positive for both axonal (Tau-1) and dendritic (MAP2) markers as described previously (Kishi et al., 2005). Because they lacked clear axonal or dendritic identity, these neurites were classified as indeterminate neurites.

Immunoprecipitation with an anti-phospho-serine antibody showed that the amount of phosphorylated Wee1 was reduced in lysates of brains from Sada^−/−;Sadb^+/− mutants whereas the total amount of Wee1 was slightly increased (Fig. 3E). After immunoprecipitation of Wee1 from wild type or Sada^−/−;Sadb^+/− mutants, a strong decrease in the phosphorylation of Wee1 at S642 was observed in the embryonic brain of the mutants with a phosphospecific antibody (Fig. 3F). The residual phosphorylation may result from other kinases that regulate Wee1 in non-neuronal cells present in the brain (Kellogg, 2003). These results show that Sad kinases are required for the phosphorylation of Wee1 in the embryonic brain.

Neuronal polarity is disrupted after Wee1 overexpression in stage 3 neurons, the stage when the expression of endogenous Wee1 is strongly reduced. This result suggests that Wee1 has to be downregulated after neurons extended a single axon. Its persistent expression interferes with the establishment of neuronal polarity. To test the possibility that Sad kinases restrict the expression of Wee1 in neurons, we investigated the consequences of Wee1 phosphorylation by SadA. We first coexpressed Wee1 and SadAca or inactive SadAki in HEK293T cells and analyzed the phosphorylation of Wee1 at S123 that is indicative of inactive Wee1 (Watanabe et al., 2005). Expression of SadAca indeed resulted in the inhibition of Wee1 as revealed by an increased phosphorylation at S123 (Fig. 4A). To investigate if Sad kinases are required for Wee1 inhibition in vivo, we analyzed the inhibitory phosphorylation of Wee1 at S123 in lysates from the embryonic brain of Sada^−/−;Sadb^+/− mutant mice (Fig. 4B). Compared with control Sadb^+/− mice, phosphorylation of Wee1 at S123 was reduced in Sada^−/−;Sadb^+/− brains. No difference was detectable in lysates from embryonic liver, indicating that the regulation of Wee1 by Sad kinases is specific for neural tissue, consistent with the tissue-specific expression of Sad kinases (Kishi et al., 2005). Immunofluorescence staining revealed that Wee1 was absent at 5 d.i.v. in Sadb^+/− neurons but persisted in Sada^−/−;Sadb^+/− neurons (Fig. 4C, supplementary material Fig. S4D). These results show that Wee1 is a target of Sad kinases and that its downregulation at 4 d.i.v. depends on Sad kinases.

Our results suggest that Sad kinases restrict Wee1 expression after the establishment of polarity, probably by initiating its degradation. The persistence of Wee1 in Sada^−/−;Sadb^+/− neurons could interfere with their normal differentiation and lead to the formation of indeterminate neurites with both axonal and dendritic characteristics. If the phenotype of SadA/B-deficient neurons results from the failure to restrict Wee1 activity, the suppression of Wee1 should compensate for the loss of Sad kinases. To test this possibility, we suppressed endogenous Wee1 in neurons prepared from Sada^−/−;Sadb^+/− embryos by RNAi. After expression of the inactive control construct Wee1-shRNAmut, 68.0±2.2% (n=49) of the Sada^−/−;Sadb^+/− neurons form multiple neurites that are positive for Tau-1 staining as observed in untransfected SadA/B-deficient neurons (70.4±1.3%; Fig. 5, supplementary material Fig. S4). After transfection with a vector for an shRNA directed against Wee1 at 3 d.i.v., the majority of neurons (87.5±1.5%; n=68) developed a single Tau-1-positive axon and several minor neurites. Thus, the suppression of Wee1 rescued the loss of Sad kinases in mutant neurons and restored normal neuronal polarity, indicating that the Sad kinases act upstream of Wee1 to restrict its activity at stage 3.

To investigate the role of Wee1 phosphorylation, we tested the ability of RNAi-resistant non-phosphorylatable Wee1S642A and the phospho-mimic Wee1S642E to rescue the suppression of endogenous Wee1 by RNAi. Coexpression of Wee1 and the anti-Wee1 shRNA in HEK293T cells confirmed that the mutants were resistant to the shRNA construct (supplementary material Fig. S5). All three forms of Wee1 were able to replace endogenous Wee1 after transfection of neurons at 0 d.i.v. and analysis at 2 d.i.v., showing that phosphorylation of S642 is not essential until stage 3 (Fig. 6).

To examine the function of Wee1 S642 phosphorylation in stage 3 neurons, we replaced the endogenous protein with exogenous mutant Wee1 in wild-type and Sada^−/−;Sadb^−/− neurons. Suppression of Wee1 by RNAi and co-transfection of RNAi-resistant Wee1 expression vectors allows to test if these Wee1 mutants can substitute for endogenous Wee1. Neurons were transfected with the vectors for an shRNA directed against Wee1 and wild-type Wee1, Wee1S642A or Wee1S642E at 2 d.i.v. and analyzed at 5 d.i.v. (Fig. 7A,B). Suppression of Wee1 rescued the phenotype of Sada^−/−;Sadb^−/− neurons. Expression of Wee1 or Wee1S642A reversed Wee1 knockdown and resulted in the formation of multiple Tau-1-positive neurites in SadA/B-deficient neurons. By contrast, Wee1S642E did not substitute for endogenous Wee1 and had no effect on neuronal morphology. The majority of Wee1S642E-expressing Sada^−/−;Sadb^−/− neurons displayed normal polarity and extended a single axon. Expression of Wee1 mutants in wild-type neurons gave the same result. These results show that Wee1S642E is as active as wild-type Wee1 in 2 d.i.v. but not in 5 d.i.v. neurons and that the phospho-mimic mutation S642E can substitute for Wee1 phosphorylation by Sad kinases in SadA/B-deficient neurons.

The downregulation in stage 3 neurons suggests that SadA/B initiate the degradation of Wee1. To investigate this possibility, we analyzed the expression level of exogenous Wee1 in transfected wild-type neurons. Wee1S642A and Wee1S642E were expressed at comparable levels in HEK293T cells (supplementary material Fig. S6). Transfected neurons expressed comparable levels of RNAi-resistant Wee1, Wee1S642A and Wee1S642E at 2 d.i.v., showing that wild-type and mutant proteins do not differ in their stability at this stage (Fig. 7C). By contrast, only low levels of Wee1S642E could be detected at 4 d.i.v. compared with Wee1 and Wee1S642A, indicating that the stability of the phospho-mimic Wee1S642E mutant changes in a stage-dependent manner. Thus, phosphorylation of Wee1 at S642 is required in neurons to restrict its function to early stages of differentiation and is essential for its downregulation in stage 3 neurons.

Here we show that the cell-cycle checkpoint kinase Wee1 is required initially for the formation of minor neurites but not for the extension of axons. However, Wee1 has to be downregulated in polarized stage 3 neurons to prevent a disruption of neuronal polarity. This downregulation is initiated by the Sad kinases. The persistence of Wee1 in SadA/B-deficient neurons results in the formation of multiple neurites that display both axonal and dendritic characteristics instead of a single axon. Thus, the SadA/B kinases perform a dual role in neurons. Initially, they promote the establishment of neuronal polarity by regulating microtubule-associated proteins (Barnes et al., 2007). Subsequently, they are required in polarized neurons to limit Wee1 expression.

The conclusion that Wee1 function is restricted by Sad kinases is based on the following results. First, Wee1 interacts with Sad kinases in embryonic brain and is phosphorylated by them at S642. S642 phosphorylation is strongly reduced in the embryonic brain of SadA/B-deficient mice. Second, the inhibitory phosphorylation of Wee1 at S123 is increased by expression of SadA and strongly reduced in the embryonic brain of Sada^−/−;Sadb^+/− mutants. Third, Wee1 overexpression at stage 3 results in the same phenotype as the loss of Sad kinases and disrupts neuronal polarity. Fourth, SadA/B-deficient neurons continue to express Wee1. Fifth, the phenotype of SadA/B-deficient neurons can be rescued by suppressing Wee1. Sixth, replacing endogenous Wee1 with phosphorylation site S642 mutants shows that the phospho-mimic mutation S642E can substitute for Wee1 phosphorylation by Sad kinases in SadA/B-deficient neurons. Expression of wild-type Wee1 or Wee1S642A disrupted neuronal polarity whereas Wee1S642E did not. Thus, phosphorylation of Wee1 at S642 results in its inactivation in polarized neurons. Wee1S642E is expressed at similar levels as Wee1 or Wee1S642A at 2 d.i.v. but not at 4 d.i.v., indicating a stage-specific and phosphorylation-dependent regulation of Wee1 stability. Wee1 hyperphosphorylation results in its degradation by the proteasome in proliferating cells (Watanabe et al., 2005; Watanabe et al., 2004). Wee1 degradation may also be responsible for the downregulation of Wee1 at 4 d.i.v. in hippocampal neurons. Taken together, our results show that Sad kinases restrict Wee1 function, which persists in SadA/B-deficient neurons.

Our results as well as previous results (Lu et al., 2004) show that Sad kinases phosphorylate Wee1 at S642. However, conflicting results have been reported for the effect of S642 phosphorylation (Katayama et al., 2005; Lee et al., 2001; Lu et al., 2004; Rothblum-Oviatt et al., 2001; Wang et al., 2000). Both an activation and stabilization and an inhibition have been observed as a consequence of Wee1 phosphorylation at S642. Our results show that phosphorylation of Wee1 at S642 by Sad kinases restricts the function of Wee1 during the establishment of neuronal polarity. Replacing endogenous Wee1 by Wee1S642E reveals that phosphorylation initiates its inactivation only in stage 3-4 but not in stage 1-2 neurons. The stage-specific stability of Wee1S642E suggests that additional cell-type or stage-specific factors such as kinases that hyperphosphorylate Wee1 or adaptors that interact with phosphorylated Wee1 (Watanabe et al., 2005) determine Wee1 stability following the phosphorylation by SadA/B. This may explain the contradictory effects reported for S642 phosphorylation. Future experiments will have to elucidate the molecular details of how the function of Wee1 is regulated in neurons.

Loss of Wee1 did not block axon formation but disrupted microtubule organization and resulted in the absence of minor neurites. The requirement for Wee1 is restricted to early stages of differentiation. In this respect, it differs from other cell-cycle regulators such as the subunit of the anaphase-promoting complex Cdh1 and the cyclin-dependent kinase inhibitor p27kip1 that have a function in postmitotic neurons at later stages (Frank and Tsai, 2009; Juo and Kaplan, 2004; Konishi et al., 2004; Nguyen et al., 2006; van Roessel et al., 2004). The loss of Wee1 prevents the formation of minor neurites without impairing axon extension. Since axons can still form in the absence of Wee1, different pathways appear to be involved in organizing microtubules in axons and minor neurites. The observation that minor neurites show differences in microtubule stability already at stage 2 supports this conclusion (Witte et al., 2008). The phenotype of neurons after suppression of Wee1 suggests a defect in microtubule bundling in minor neurites. Wee1 may act in a pathway similar to that described for Cdk5, which regulates microtubule bundling by phosphorylating doublecortin (Bielas et al., 2007; Tanaka et al., 2004). Wee1 could affect microtubule organization by regulating MAPs such as MAP2 that are specific for minor neurites. MAP2 is necessary for neurite initiation, and expression of the embryonic isoform MAP2c triggers neurite formation (Dehmelt et al., 2003). Unlike tau, MAP2c binds not only to microtubules but also to F-actin, which appears to be essential for neurite extension (Roger et al., 2004). Since MAP2 becomes restricted to dendrites, the different properties of MAPs such as MAP2 and tau may explain why minor neurites but not axons are affected by Wee1 suppression. Thus, Wee1 may specifically regulate MAPs that are required for the initiation and/or maintenance of minor neurites but not axons.

In summary, we show that Sad kinases initiate the downregulation of Wee1 in polarized stage 3 neurons to limit its activity to early stages of neuronal differentiation. Wee1 impairs neuronal polarity in hippocampal neurons when its expression is maintained at later stages. Sad kinases promote neuronal polarity not only by phosphorylation of microtubule binding proteins (Barnes et al., 2007; Kishi et al., 2005) but also by regulating Wee1 after polarity is established. Thus, the restriction of Wee1 activity is an essential function of Sad kinases and is required for neuronal polarity.

The expression vectors for Wee1 and Myt1 were kindly provided by Helen Piwnica-Worms (Washington University School of Medicine, St Louis, MO). The RNAi-resistant Wee1 mutant was generated by introducing silent mutations into the expression vector for Wee1 (nt 3071-3073: ACT to ACA and nt 3059-3061: TCT to TCA; nucleotide position refers to accession no. NM_003390). Wee1S642A and Wee1S642E were generated by introducing the mutations AGC to GCC and AGC to GAG (nt 3065-3067), respectively. Two mouse homologs of C. elegans Sad1 (SadA and SadB) (Kishi et al., 2005) were amplified by PCR from mouse E12 spinal cord cDNA and cloned into pEGFP-C1 (Clontech). Constitutively active SadA (SadAca), SadBca, kinase-dead SadA (SadAki), and SadBki were generated by introducing the mutations T175E (SadAca), S179D (SadAki), T187E (SadBca) and S191D (SadBki) using the QuickChange site-directed mutagenesis kit (Promega). The N- (aa 1-375) and C- terminal (aa 303-654) domains of SadA and SadB (N: aa 1-389, C: aa 316-777) were amplified by PCR and cloned into the expression vector pEGFP-C1 (Clontech). The vector for tau-EGFP was generated by cloning an oligonucleotide corresponding to amino acids 251-278 of tau into pEGFP-C1 (Clontech).

To suppress Wee1 by RNAi, the plasmids HP_696 (shRNA1) and HP_264818 (shRNA2; Open Biosystems) were used to express an shRNA. To generate control vectors with two exchanges in the target sequence, the oligonucleotide 5′-AGAAT ATGAG CTTCT GCCGC CGACT GTAGA AGCTT GTGCA GTCGG CGGCG GAGGC TCATA TTCTT GCTTT TTT-3′ (Wee1 shRNAmut) was cloned into pSHAG-1 (Paddison et al., 2002).

At 48 hours after the transfection, HEK293T cells were lysed in lysis buffer [2% Triton X-100, 100 mM β-glycerophosphate and complete protease inhibitor cocktail (Roche) in PBS] for 30 minutes at 4°C. Cells were lysed 48 hours after transfection and lysates cleared by centrifugation. For immunoprecipitation, forebrains from E18.5 mouse embryos were homogenized in mRIPA buffer (1% NP-40, 1% sodium deoxycholate, 0.1% SDS, 50 mM HEPES pH 7.4, 150 mM NaCl, 10% glycerol, 1.5 mM MgCl₂) and lysed for 30 minutes at 4°C. Lysates were cleared by centrifugation and mixed 1:1 with wash buffer (0.1% NP-40, 50 mM HEPES pH 7.4, 75 mM NaCl, 10% glycerol). To dephosphorylate proteins, 1000 IU lambda protein phosphatase (Upstate), buffer (Upstate) and 5 mM DTT (Sigma) were added to 500 μl lysate and incubated for 20 minutes at 30°C. For immunoprecipitation and western blots the following antibodies were used: anti-GST (GE; #27-4577-01, 1:5000), anti-α-tubulin (Sigma; #T6199, 1:500), anti-GFP (Babco; #MMS118P, 1:500), anti-myc (Roche; #1-667-149, 1:500), anti-phospho-S262-tau (Millipore, #AB9656 1:500), anti-phospho-serine (Chemicon, #AB1603, 1:500), anti-Wee1 (Upstate; #06-972, 1:500), anti-phospho-S642-Wee1 (CST, #4910, 1:5000) and anti-phospho-Wee1-S123 (Abgent; #AP3284a, 1:500). Western blots were analyzed using the LAS-1000 luminescent image analyzer (Fujifilm) and the WASABI software (Hamamatsu).

Sada^−/−;Sadb^+/− knockout mice were generously provided by Joshua R. Sanes (Harvard University) and genotyped by PCR as described (Kishi et al., 2005) using the primers 5′-TGCCAAGTTCTAATTCCATCAGAAGCT-3′, 5′-TGCCCCTGCTCACCTTAGGTGTCA-3′ and 5′-TGGGAAGGTAAGCAGGGAGGCCAG-3′ (Sada) or 5′-TGCCAAGTTCTAATTCCATCAGAAGCT-3′, 5′-TGTCTCCTATACCTTGATAGGTAGGCA-3′ and 5′-AATGAAGATGGCTTGATAGGCTTACCA-3′ (Sadb), respectively.

Cultures of dissociated hippocampal neurons from rat or mouse embryos were prepared as described previously (Schwamborn et al., 2006; Schwamborn and Püschel, 2004). Briefly, the hippocampus was dissected from E18 embryos, incubated with trypsin (Invitrogen) for 20 minutes at 37°C and dissociated in Dulbecco's minimal essential medium (DMEM). Neurons were plated onto glass coverslips coated with poly-ornithine (Sigma) at a density of 100,000 cells per coverslip and cultured at 37°C and 5% CO₂. Two to four hours after plating the medium was changed to Neurobasal medium with B27 supplement, 0.5 mM glutamine (all Invitrogen) and gentamicin (PAN). Neurons were transfected at 0.5 or 3 d.i.v. (days in vitro) and fixed at 3 or 5 d.i.v., respectively, using 4% paraformaldehyde and 15% sucrose in phosphate-buffered saline (PBS) for 20 minutes on ice or methanol for 10 minutes at −20°C. For immunofluorescence, hippocampal neurons were permeabilized with 0.1% Triton X-100 in PBS, blocked with 10% FCS in PBS (blocking buffer) and incubated with primary and secondary antibody in blocking buffer. Neurites longer than 100 μm and positive for the characteristic Tau-1 staining pattern with a proximal to distal increase in staining intensity were counted as axons (Schwamborn et al., 2006). Processes shorter than 50 μm and positive for MAP2 staining were classified as minor neurites (3 d.i.v.) or dendrites (5 d.i.v.), respectively. Neurites that were positive for Tau-1 and MAP2 staining and did not fulfill the criteria for axons or dendrites were classified as indeterminate neurites.

The following antibodies were used: Tau-1 (Chemicon; #MAB3420, 1:200), anti-MAP2 (Chemicon; #AB5622, 1:1000), anti-Wee1 (Upstate; #06-972, 1:100), anti-α-tubulin (Sigma; #T6199, 1:750) and Alexa-Fluor-conjugated secondary antibodies (Molecular Probes; 1:1000). To generate antibodies specific for SadA and SadB, guinea pigs were immunized with the peptides CPPPSPGGGVGGAAWRS (SadA) CPPAPGLSWGAGLKGQK (SadB) and affinity-purified (Seqlab). The specificity of the SadA and SadB antibodies was confirmed by western blotting (supplementary material Fig. S1) and immunofluorescence using neurons prepared from Sada and Sadb mutants (not shown). To examine immunofluorescence, a Zeiss Axioscope microscope equipped with a SPOT CCD camera and the SPOT 4.1 software (Diagnostic Instruments) was used. The length of neurites was determined using the Spot software.

We are grateful to Helen Piwnica-Worms for plasmids, Joshua R. Sanes for the Sada and Sadb mutants, L. Blume for constructing tau-EGFP, and V. Gerke and C. Klämbt for helpful comments on the manuscript. This work was supported by grants from the Deutsche Forschungsgemeinschaft (SPP 1111 and SFB 629).