Dap160/intersectin binds and activates aPKC to regulate cell polarity and cell cycle progression

Asymmetric cell division has been proposed as a mechanism for maintaining stem cell numbers while allowing the generation of differentiated cell types(Morrison and Kimble, 2006). It is also been proposed that asymmetric cell division of a small number of tumor stem cells may be the origin of several tumor types(Morrison and Kimble, 2006). Asymmetric cell division typically involves establishment of a cell polarity axis, followed by alignment of the mitotic spindle with the polarity axis to produce daughter cells with different molecular composition. One evolutionarily conserved protein involved in both cell polarity and asymmetric cell division is atypical protein kinase C (aPKC; called PKC-3 in C. elegans and aPKCζ, λ/ι in mammals). aPKC is required for C. elegans blastomere cell polarity(Tabuse et al., 1998), Drosophila oocyte polarity (Tian and Deng, 2008); C. elegans, Drosophila and mammalian epithelial polarity (Aranda et al.,2006; Gopalakrishnan et al.,2007; Harris and Peifer,2007; Helfrich et al.,2007; Hutterer et al.,2004; Nagai-Tamai et al.,2002; Rolls et al.,2003; Suzuki et al.,2004; Suzuki et al.,2002; Takahama et al.,2008; Yamanaka et al.,2006; Yamanaka et al.,2003; Yamanaka et al.,2001); and Drosophila planar cell polarity(Djiane et al., 2005); and has been implicated as an oncogene in several human tumors(Fields et al., 2007; Regala et al., 2005; Yi et al., 2008). aPKC is called `atypical' because unlike classical PKCs or novel PKCs, it is not activated by Ca²⁺ or diacylglycerol, but rather by a relatively poorly understood mechanism of protein-protein interactions.

Drosophila larval brain neural progenitors (neuroblasts) are a powerful system with which to study the role of asymmetric cell division and stem cell self-renewal and proliferation. aPKC is required for both cell polarity and stem cell self-renewal in Drosophila neuroblasts(Lee et al., 2006; Rolls et al., 2003). aPKC colocalizes with the polarity proteins Par-6, Cdc42 and Bazooka (Baz; Par3 in mammals) at the apical cortex of mitotic neuroblasts, and is segregated into the self-renewing neuroblast at each cell division. The apical cortical crescent of aPKC ensures exclusion of the differentiation factors Miranda,Prospero, Brain tumor (Brat) and Numb from the apical cortex, resulting in their restriction to the basal cortex and ultimate partitioning into the smaller ganglion mother cell (Doe,2008; Knoblich,2008), which typically divides once to form two postmitotic neurons. In aPKC mutants neuroblasts, cell polarity is disturbed such that Miranda and Numb basal proteins become distributed uniformly on the neuroblast cell cortex, both at metaphase and at telophase, resulting in a molecularly symmetric cell division and a decrease in neuroblast numbers(Lee et al., 2006; Rolls et al., 2003). aPKC has been proposed to regulate neuroblast cortical polarity by phosphorylating and inhibiting cortical localization of the Lethal giant larvae (Lgl) protein(Betschinger et al., 2003),which is required for basal targeting of Miranda and Numb proteins(Ohshiro et al., 2000; Peng et al., 2000). Overexpression of a membrane-targeted aPKC, but not a kinase dead version,leads to reduced basal protein localization and the formation of supernumerary neuroblasts (Lee et al.,2006), revealing the importance of aPKC kinase activity for promoting neuroblast cell polarity and self-renewal. Despite the central role of aPKC kinase activity in establishing neuroblast cell polarity, relatively little is known about how aPKC activity is regulated. Recently, we showed that the small G-protein Cdc42 is colocalized with aPKC and can stimulate aPKC activity by relieving Par-6 inhibition of aPKC(Atwood et al., 2007). However,the stimulation of activity is modest and the polarity phenotypes of cdc42 mutant are not fully penetrant, which suggests that additional aPKC activators may exist.

Despite the importance of aPKC in cell polarity and growth control, little is known about how aPKC activity is regulated. Here, we have sought to identify aPKC-interacting proteins using a biochemical approach, and then assay their function in Drosophila neuroblasts. We performed immunoprecipitation experiments coupled to mass spectrometry analysis (IP/MS)using aPKC as the bait protein and identified Dap160 (dynamin associated protein-160), a member of the intersectin protein family(Adams et al., 2000). Dap160 was originally identified by its ability to interact with the endocytic protein Dynamin in Drosophila head extracts(Roos and Kelly, 1998). dap160 mutants have nerve terminals with reduced levels of endocytic proteins (Koh et al., 2004; Marie et al., 2004),consistent with a role in endocytosis. However, dap160 mutants were also isolated in a genetic screen for modifiers of the Sevenless receptor tyrosine kinase (Roos and Kelly,1998), showing that Dap160 also regulates signal transduction,similar to mammalian intersectin(Malacombe et al., 2006; Martin et al., 2006; Tong et al., 2000a; Tong et al., 2000b). Here, we show that Dap160 binds aPKC, increases its kinase activity, and that both Dap160 and aPKC are required for neuroblast cell polarity and cell cycle progression.

The wild-type fly stock was yellow white (y w). The dap160^Q24, Df(2)10523^#35 and UAS-Dap160 stocks were gifts from Graeme Davis (UCSF). dap160^Δ1 was a gift from Hugo Bellen (Koh et al., 2004). aPKC^k0643 flies lacking the early embryonic lethal background mutation have been described previously(Rolls et al., 2003). Analysis of cell polarity was carried out in stage 15-16 dap160^Q24/Df(2)10523 trans-heterozygotes embryos as previously described (Marie et al.,2004). The hypomorphic dap160^Q24/dap160^Δ1hetero-allelic combination was used to assay larval brain neuroblast numbers. The shi^ts2 chromosome was a gift from Mani Ramaswami(Arizona). To generate positively marked MARCM clones, we recombined FRT40 onto the dap160^Q24 chromosome using standard techniques, and used the previously described FRTG13 aPKC^k06403 chromosome(Rolls et al., 2003). dap160 neuroblast mutant clones were generated by mating dap160^Q24 FRT40/CyO, actGFP to y w hsFLP; FRT40A tubP-GAL80/CyO ActGFP; tubP-GAL4 UAS-mCD8::GFP/TM6 Tb Hu(Bello et al., 2008). aPKC^k06403 neuroblast mutant clones were generated by mating FRTG13 aPKC^k06403/CyO to y w hsFLP; FRTG13 tubP-GAL80/CyO actGFP; tubP-GAL4 UAS-mCD8::GFP/TM6 Tb Hu (C. Cabernard and C.Q.D., unpublished). Clones were induced at 24-28 hours after larval hatching for 1 hour at 37°C and aged for 4 days at 25°C.

GST-tagged Dap160 was engineered by polymerase chain reaction from a P-spaceneedle-Dap160-dsred construct (a gift from Graeme Davis) followed by subcloning into pGEX-4T1 vector (Pharmacia). The construct was verified by DNA sequencing. GST-Dap160 was expressed in BL21 cells overnight at 25°C,adsorbed onto glutathione agarose (Sigma), washed three times with binding buffer [10 mM HEPES (pH 7.5), 100 mM NaCl, 1 mM DTT, 0.1% Tween-20]. We co-incubated GST-Dap160 with Par-6 or aPKC proteins at room temperature for 15 minutes followed by five washes in binding buffer. Interactions were tested by eluting proteins in SDS sample buffer, SDS-PAGE and western blotting. Par-6 and aPKC proteins were a gift from Scott Atwood (Prehoda laboratory, OR). For immunoprecipitation experiments, a 12-hour collection of y w embryos was homogenized in lysis buffer [50 mM HEPES (pH 7.5), 150 mM NaCl, 0.1%Tween-20, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol, supplemented with protease inhibitor tablets; Roche] to produce 1 ml of lysate. Lysates were pre-cleared with protein agarose-A beads for 1 hour at 4°C and subsequently divided equally (500 μl each) in two Eppendhorf tubes and incubated with 2 μl of either anti-βgal or anti-aPKC antibodies for 4 hours at 4°C. Lysates were then incubated with protein agarose A beads for 1 hour at room temperature. For pulldowns, beads were precipitated and washed three times in modified lysis buffer containing 1 mM NaCl. Samples were sent to the Fred Hutchinson Cancer Research Center proteomic facility (Seattle, WA) for mass-spectrometry analysis.

aPKC activity assays were carried out as described previously(Atwood et al., 2007). Briefly 0.16 μM aPKC was co-incubated with 50 μM peptide substrate, no Par-6 or 1.14 μM Par-6, no Dap160 or increasing molar concentration of Dap160 protein (0.34, 0.68, 1.02 μM). Phosphorylated peptides were resolved by gel electrophoresis, quantified using ImageJ and normalized to total peptide input.

An affinity-purified anti-Dap160 antibody was raised against the C-terminal peptide sequence GPF VTS GKP AKA NGT TKK (Alpha Diagnostic). Guinea pig or chicken anti-Dap160 was used at 1:100 (this study); guinea pig or rat anti-Miranda at 1:500 (Doe laboratory); rabbit anti-aPKC at 1:1000 (Sigma),guinea pig anti-Numb at 1:1000 (a gift from Jim Skeath, Missouri); rabbit anti-Scrib at 1:2500 (Doe laboratory); rat monoclonal anti-Dpn (Doe laboratory) at 1:1; mouse anti-Pros monoclonal (purified MR1A at 1:1000; Doe laboratory); rabbit anti-phosphohistone H3at 1:1000 (Sigma, St Louis, MO);rabbit anti-GFP at 1:1000 (Sigma, St Louis, MO); rat anti-Par-6 at 1:250(Atwood et al., 2007); mouse anti-β galactosidase at 1:500 (Promega); mouse anti-α tubulin (at 1:1500, Sigma); guinea pig anti Sec15 at 1:1500 (Hugo Bellen); rabbit anti-Rab11 at 1-1000 (Donald Ready); and rabbit anti-Dap160 at 1:500(Marie et al., 2004; Roos and Kelly, 1998). Secondary antibodies were obtained from Molecular Probes (Eugene, OR). Embryos were fixed and stained as described previously(Siegrist and Doe, 2005),except that fixing was carried out in 9% paraformaldehyde for 15 minutes. shi^ts2 embryos were shifted to restrictive temperature(37°C) for 30 minutes and subsequently fixed and stained. Larval brains were dissected, fixed and stained as described previously(Siller et al., 2005), and analyzed with a BioRad Radiance 2000 or Leica TCS SP laser scanning confocal microscope using a 60×1.4 NA oil immersion objective. Images were processed with Illustrator software (Adobe).

For live imaging we used the GFP protein-trap line G147(GFP::Jupiter) that expresses a microtubule-associated GFP fusion protein(Morin et al., 2001), worniu-GAL4, UAS-GFP::Miranda, UAS-CHERRY::Jupiter (C. Cabernard and C.Q.D., unpublished) and worniu-GAL4; UAS-Dap160/TM6B Tb. Live imaging was performed as previously described(Siller et al., 2005), except that temporal resolution was either 15 seconds for 2-hour imaging intervals or 3 minutes for overnight imaging sessions. The 4D data sets were processed in ImageJ (NIH) and Imaris (Bitplane, Switzerland) software.

To test the significance in neuroblast number variation between wild-type, dap160 mutant and Dap160-overexpressing brains we used a t-test (two-tailed distribution and two samples, equal variance). To test the significance in NEBD to AO length variation between wild type, dap160 and aPKC mutants we used a t-test(two-tailed distribution and two samples, unequal variance). To test the significance of the differences in polarity protein localization between wild type and dap160 mutant neuroblasts, we performed a z-test using the percent normal crescent for each polarity protein comparing wild type and dap160 mutants; the difference in percent normal between wild type and dap160 mutants was significant with a one-tail confidence level greater than 99.6 % for all polarity proteins analyzed.

To identify novel components in aPKC pathways of neuroblast cell polarity and neuroblast self-renewal, we sought to identify aPKC/Par-6 interacting proteins by IP/MS using aPKC as our bait protein. IP eluates containing complex mixtures of proteins from a non-specific pull-down (anti-green fluorescent protein; GFP) or anti-aPKC pull down were analyzed by Multiple Dimensions Protein Identification Technology (MUDPIT). These experiments returned Dap160 as a potential aPKC-interacting protein with the highest score of tryptic products hits. Dap160 contains two Eps15 homology domains (EH), a coiled-coil domain (CC) and four Src homology 3 domains (SH3)(Fig. 1A). These domains are conserved in the vertebrate Dap160 homologue intersectin(Fig. 1A), although Dap160 lacks the DH, PH, C2 domains present in the long isoform of intersectin and consequently is not predicted to have GTPase activity(Fig. 1A). Dap160 has two alternative splice variants: a long isoform, which migrates at 160 kDa and a shorter isoform migrating at 120 kDa (Fig. 1A,B).

Next, we sought to verify that Dap160 and aPKC are part of the same protein complex in vivo. An affinity-purified peptide antibody raised against Dap160 C terminus, which recognizes a 160 kDa band in wild-type larval lysate but not in dap160 mutant lysate (Fig. 1B), was used to perform aPKC/Dap160 pull-down experiments. We found that Dap160 and aPKC reproducibly co-immunoprecipitate(Fig. 1C), thus indicating that Dap160 and aPKC are present in the same protein complex. We obtained similar results with a second, independently generated Dap160 antibody(Marie et al., 2004; Roos and Kelly, 1998)(Fig. 2M).

We tested whether Dap160-aPKC directly interact using an in vitro protein interaction assay. We incubated full-length N-terminal GST-tagged Dap160 with full-length N-terminal His-tagged aPKC and found that Dap160 can bind aPKC(Fig. 1D). Par-6 and aPKC form a complex in Drosophila (Atwood et al., 2007), so we tested for direct interactions between Dap160 and Par-6. We found that full-length Dap160 also directly binds Par-6(Fig. 1E). We conclude that Dap160 directly interacts with both aPKC and Par-6.

We next tested whether Dap160 increased aPKC activity (similar to the Par-6 binding protein Cdc42) or repressed aPKC activity (similar to the aPKC binding protein Par-6) (Atwood et al.,2007). We co-incubated aPKC protein and a proven fluorescent peptide substrate (Atwood et al.,2007) with increasing amounts of Dap160 protein in the in the presence of Par-6. By measuring the amount of phosphorylated substrate relative to total peptide input, we found that Dap160 increased aPKC activity in a dose-dependent manner (Fig. 1F). To test whether Dap160 stimulation of aPKC is dependent on Par-6, we compared the ability of Dap160 to stimulate aPKC in the presence or absence of Par-6. We found that Dap160 directly stimulates aPKC and that Par-6 could almost completely suppress the ability of Dap160 to activate aPKC(Fig. 1G). Note that aPKC basal activity is variable between experiments (compare Fig. 1F with Fig. 1G) but we found that Dap160 was always able to increase aPKC basal activity. We conclude that the Dap160/aPKC or Dap160/Par-6/aPKC complex is more active than the aPKC/Par-6 complex.

To determine whether Dap160 has the potential to activate aPKC in neuroblasts, we examined Dap160 localization during neuroblast asymmetric cell division. In wild-type neuroblasts, Dap160 protein is cytoplasmic at interphase, but becomes enriched at the apical cortex during mitosis(Fig. 2A-D); this is identical to aPKC localization (Fig. 2G-K) (Rolls et al.,2003). Two different antibodies produced similar staining in wild-type neuroblasts, and showed no apical Dap160 enrichment in dap160 mutants (Fig. 2B,N), confirming specificity of the antibodies. In larval neuroblasts, Dap160 was undetectable at the cortex at mitosis(Fig. 2E), possibly owing to low protein levels and/or reduced antibody detection sensitivity. Therefore,we used a UAS-dap160 transgene to overexpress Dap160 in larval neuroblasts, and observed enrichment of Dap160 at the apical cortex (90%, n=12; Fig. 2F,L)although weaker ectopic cortical patches could also be observed (66%, n=12; Fig. 4K). This provides further support for the specificity of our antibodies for Dap160 protein. We conclude that Dap160 and aPKC are cytoplasmic in interphase neuroblasts, and can be colocalized at the apical cortex of mitotic neuroblasts.

Wild-type mitotic embryonic neuroblasts have apical aPKC, Par-6 and Baz crescents (Fig. 3A-C), and basal Miranda/Numb cortical crescents (Fig. 3D-E). In dap160 mutant embryonic neuroblasts, aPKC showed ectopic cortical patches beyond the apical cortical domain (71%, n=41 neuroblasts from 13 embryos; Fig. 3F); similarly, Par-6 showed ectopic cortical patches beyond the apical cortical domain (50%, n=12 from 6 embryos; Fig. 3G). By contrast, Baz was mostly apical but occasionally cytoplasmic (22%, n=96 from 35 embryos; Fig. 3H). Cortical aPKC protein may be inactive or less active in dap160 mutant neuroblasts, because Miranda and Numb proteins showed ectopic cortical localization beyond the basal cortical domain (Miranda, 36%, n=55 from 24 embryos; Numb, 38%, n=18 from 11 embryos; Fig. 3I,J), similar to a weak aPKC mutant phenotype(Rolls et al., 2003). dap160 mutant neuroblasts never showed cytoplasmic Miranda, but rather we observed that aPKC overlapped with Miranda at the cortex (15%, n=13 from five embryos; Fig. 3F,I,J), which is never observed in wild-type neuroblasts(Lee et al., 2006; Rolls et al., 2003). Conversely, Dap160 overexpression in larval neuroblasts results in cortical Dap160 and ectopic cortical patches of aPKC(Fig. 4J,L), as well as depletion of cortical Miranda in some neuroblasts (19%, n=21 from 12 larvae; Fig. 4K), suggesting that the ectopic aPKC is at least partially active.

We further investigated Miranda localization by time-lapse analysis of larval neuroblasts expressing GFP::Miranda and the microtubule-binding protein Cherry::Jupiter (C. Cabernard and C.Q.D., unpublished). In wild-type neuroblasts, GFP::Miranda forms basal crescents during mitosis (100%, n=37; Fig. 4M). In neuroblasts with overexpression of Dap160, we could observe divisions where Miranda was cytoplasmic (8%, n=40 divisions from 19 neuroblasts; Fig. 4N), which we never observed in wild type. We conclude that Dap160 positively regulates aPKC localization and activity, and is required for aPKC-mediated neuroblast cortical polarity.

An important function of aPKC is to maintain larval neuroblast pool size: aPKC mutants have fewer proliferating larval brain neuroblasts and overexpression of a membrane-tethered aPKC increases the number of larval brain neuroblasts (Lee et al.,2006). Here, we test whether decreasing or increasing Dap160 levels has a similar effect on brain neuroblast numbers. We find that wild-type third larval instar brain lobes contain 96±5 (n=5)Deadpan-positive (Dpn⁺) neuroblasts, whereas similarly staged dap160 mutants contain only 72.6±6 (n=5); P=0.0002 Dpn⁺ neuroblasts(Fig. 4A). We conclude that both Dap160 and aPKC are required to maintain the normal number of proliferating neuroblasts in the larval brain. The loss of neuroblasts in the dap160 mutant brain could be due to neuroblast death, differentiation or the failure of neuroblasts to exit from quiescence during larval stages(see Discussion).

We next generated dap160 or aPKC mutant single neuroblast clones to determine whether Dap160 (or aPKC) is required within the neuroblast and its progeny for maintaining neuroblast identity. We generated single neuroblast clones in first instar larvae, allowed them to develop for 96 hours and scored them in late third instar larvae (see Materials and methods). Without exception, we found that all dap160 mutant clones contained a single large Dpn⁺ neuroblast (n>50 clones in 16 larvae; Fig. 4D). Similarly, all aPKC single neuroblast mutant clones retained one Dpn⁺neuroblast (n=35 clones in 12 larvae; Fig. 4E). The dap160mutant neuroblast clones showed aPKC and Miranda neuroblast polarity phenotypes similar but weaker than observed in dap160 mutant embryos(compare Fig. 4H,I with Fig. 3). These results may be due to masking of the dap160 phenotype by residual Dap160 protein in the mutant clone, or due to a role of Dap160 outside the neuroblast lineage in maintaining neuroblast numbers (see Discussion). Nevertheless, we can conclude that Dap160 is required within the neuroblast or its progeny for proper neuroblast cell polarity.

We next determined whether overexpression of Dap160 could affect the number of proliferating neuroblasts. We overexpressed Dap160 using a neuroblast-specific Gal4 driver (worniu-gal4 UAS-dap160) and scored the number of Dpn⁺ larval neuroblasts. We saw no change in the number of larval neuroblasts at first instar (data not shown). At second larval instar, wild-type larvae contain 78±5.6 neuroblasts(n=4; Fig. 4A),whereas larvae overexpressing Dap160 have 106.3±13.9, P=0.03(n=4; Fig. 4A). This result is consistent with Dap160 inducing ectopic neuroblast self-renewal,similar but much weaker than the increase in neuroblast numbers seen following overexpression of membrane-tethered aPKC(Lee et al., 2006). At third instar, wild-type larvae contain ∼96±5 neuroblasts (n=5; Fig. 4A), whereas larvae overexpressing Dap160 have a decline in neuroblast numbers to∼57±10 (n=3; Fig. 4A). This surprising result may be due to neuroblast differentiation, following a graduate accumulation of Miranda/Prospero/Brat differentiation factors in the neuroblasts. This hypothesis is based on our observation that Miranda partitioning into the GMC is defective when assayed in fixed preparations (Fig. 4K)or in live imaging of GFP::Miranda localization(Fig. 4N). Alternatively,prolonged exposure to high Dap160 levels could lead to neuroblast cell death. Taken together, our mutant and misexpression data support a role for both Dap160 and aPKC in maintaining the number of proliferating neuroblasts in the larval brain.

We next tested whether Dap160 and aPKC are required for neuroblast cell cycle progression. We performed time lapse imaging of neuroblast cell cycle progression in both dap160 and aPKC mutant larval neuroblasts expressing the spindle marker GFP::Jupiter. Wild-type neuroblasts take 7.76±2.04 (n=15) minutes to transit from nuclear envelope breakdown (NEBD) to anaphase onset (AO; Fig. 5A), consistent with previous reports (Siller et al.,2006; Siller and Doe,2008; Siller et al.,2005). By contrast, progression through mitosis (NEBD-AO) was delayed in both dap160 mutants and aPKC mutants:13.37±4.4 minutes, n=10; P=0.006 in dap160mutant neuroblasts (Fig. 5B)and 17.84±4.52 minutes, n=11; P=7.6×10^-6 in aPKC mutant neuroblasts(Fig. 5C). In addition, dap160 mutant neuroblasts had a longer interphase length (often over 12 hours; data not shown) compared with an average of ∼2 hours for wild-type neuroblasts (Siller and Doe,2008) (C. Cabernard and C.Q.D., unpublished). We conclude that Dap160 and aPKC promote cell cycle progression in larval neuroblasts.

aPKC plays an important role in regulating cell polarity, neural stem cell identity and neural stem cell proliferation in vertebrates and invertebrates(Costa et al., 2008; Grifoni et al., 2007; Lee et al., 2006; Rolls et al., 2003). Here, we show that Dap160 positively regulates aPKC activity and localization in neuroblasts, and is required for effective aPKC function in establishing neuroblast cell polarity and cell cycle progression.

Par6 directly interacts with aPKC to suppress aPKC activity, while Cdc42 binds Par-6 and modestly upregulates aPKC activity(Atwood et al., 2007; Etienne-Manneville and Hall,2001; Henrique and Schweisguth, 2003; Hirano et al., 2005). Our study shows that Dap160 directly interacts with both aPKC and Par-6, and can stimulate aPKC activity independent of Par-6. However, Par-6 reduces the ability of Dap160 to stimulate aPKC, suggesting that the Dap160/aPKC complex is more active than Dap160/aPKC/Par6 complex,which in turn is more active than the aPKC/Par6 complex. The exact binding sites for the Dap160/aPKC interaction are unknown; good candidates would be the Dap160 SH3 domains and the two SH3-binding motifs (P-X-X-P) in aPKC.

How does Dap160 promote aPKC localization and activity? Because Dap160 is known to bind Dynamin to regulate endocytosis, it is possible that Dap160 may promote apical aPKC localization by clearing aPKC from the basal cortex. Arguing against this model is our finding that a dynamin mutant(shi^ts2) does not affect the apical localization of aPKC or basal localization of Miranda and Numb(Fig. 3K-M; compare with dap160 phenotype shown in Fig. 3F,I,J). We favor a model in which Dap160 regulates aPKC localization via an endocytosis-independent mechanism. In support of endocytosis-independent functions for Dap160, its vertebrate homolog,intersectin, has both endocytosis and signaling functions. The signaling function requires protein domains shared between Dap160 and intersectin, and includes binding and recruiting mammalian son-of-sevenless (Sos) to the plasma membrane where it regulates Ras signaling(Tong et al., 2000a; Tong et al., 2000b).

Dap160 may promote aPKC localization indirectly, by increasing aPKC kinase activity. aPKC is required to stabilize the Par complex (Baz/Par-3, Par-6,aPKC) in many cell types, including neuroblasts(Atwood et al., 2007), and it is likely that lowered aPKC activity would destabilize anchoring proteins such as Bazooka from the neuroblast apical cortex. In support of this model, we observe a weakening of the Bazooka crescent in dap160 mutant embryonic neuroblasts.

Another possibility is that Dap160 regulates aPKC cortical polarity via vesicle transport. Consistent with this model, Dap160 controls synaptic vesicle transport in Drosophila nerve terminals(Marie et al., 2004), and aPKC(PKC-3), Par-6 and Cdc42 regulate vesicle transport in C. elegansembryos and mammalian cells (Balklava et al., 2007). In fact, Dap160 overexpression in neuroblasts results in enlarged vesicles that are positive for aPKC and the exocyst marker Sec15(Fig. 2O and data not shown),suggesting that Dap160 may direct aPKC-positive vesicles to the apical cortex. One appealing model that awaits testing is that polarized vesicle transport localizes Par proteins to the neuroblast apical cortex, and in turn Par proteins restrict differentiation factors such as Miranda and Numb to the basal cortex.

dap160 mutant embryonic neuroblasts have defects in Baz localization, but dap160 mutant larval neuroblasts show normal Baz localization (data not shown). This may reflect a difference in the mechanism of Baz localization between embryonic and larval neuroblasts, because aPKC mutants also have normal Baz localization in larval neuroblasts(Rolls et al., 2003).

We provide the first evidence that Dap160 and aPKC promote cell cycle progression in neuroblasts. The related vertebrate aPKCζ regulates cell proliferation in Xenopus oocytes, mouse fibroblasts and in human glioblastoma cell lines (Berra et al.,1993; Donson et al.,2000), indicating that aPKC promotes cell cycle progression in many cell types. Similarly, the vertebrate Dap160-related intersectin protein is sufficient to induce oncogenic transformation of rodent fibroblasts(Adams et al., 2000),indicating that intersectin can also promote cell cycle progression. It would be interesting to investigate the relationship between intersectin and PKCζ in this tumor model system.

We find that both dap160 and aPKC mutant larvae have a partial reduction in the number of proliferating neuroblasts (this work; Lee et al., 2006b). This may be due to neuroblasts differentiating or dying. We cannot exclude neuroblast death, although we see no difference in caspase 3 staining (a cell death marker) between wild-type and mutant larval brains(data not shown). Differentiation of the mutant neuroblasts is more likely,based on our observation that the Miranda differentiation factor scaffolding protein frequently fails to be properly segregated into the GMC during neuroblast asymmetric cell division in dap160 or aPKCmutants. Arguing against this model is the fact that dap160 or aPKC single neuroblast mutant clones maintain a single neuroblast in all cases examined. It is possible that the mutant single neuroblast clones may not fully deplete Dap160 or aPKC protein, e.g. owing to a long half-life of the mRNA or protein. Indeed, neuroblasts in dap160 mutant clones have a weaker cortical polarity phenotype than neuroblasts in dap160organismal mutants, consistent with the presence of residual Dap160 protein in the clones. Alternatively, loss of Dap160 or aPKC outside of the neuroblast lineage may lead to the observed decrease in neuroblast numbers in the whole animal mutants, and this would not be seen in the single neuroblast mutant clones.

Misexpression of Dap160 produces a modest increase in the number of proliferating neuroblasts in second instar larvae, whereas misexpression of cortically tethered aPKC results in a dramatic expansion of neuroblast numbers at all larval stages (Lee et al.,2006). It is likely that activity levels of aPKC are limiting in Dap160 overexpression experiments: unknown proteins(s) may oppose Dap160 stimulation of aPKC (similar to Par-6) leading to a weakly active aPKC and thus causing only a modest increase in neuroblast numbers. Surprisingly, we found that prolonged misexpression of Dap160, into the third larval instar,resulted in loss of neuroblasts. This could be due to neuroblast cell death caused by continued exposure of the neuroblast to elevated levels of Dap160 protein.

We thank Graeme Davis, Bruno Marie, Hugo Bellen, Scott Atwood, Donald Ready and Mani Ramaswami for providing reagents. We thank Laurina Manning for technical assistance. We also thank members of the Doe laboratory, Bruce Bowerman and Kenneth Prehoda for comments on the manuscript. This work was supported by HHMI and AHA pre-doctoral fellowship (0615592Z).