PIP5K-dependent production of PIP2 sustains microtubule organization to establish polarized transport in the Drosophila oocyte

Polarity is a central feature of cell function and development. The establishment and maintenance of cell polarity rely on restricting the distribution of proteins to particular cortical and cytoplasmic domains. In many cases, this is achieved through asymmetric transport organized along a polarized microtubule (MT) cytoskeleton. Polarization of the MT network is controlled by the positioning of the microtubule-organizing center (MTOC) and by the interaction of the MT array with the cell cortex, particularly the cortical actin cytoskeleton. It is clear that interactions between the plasma membrane and the cytoskeleton are central to the organization of polarized transport, but the molecular mechanisms underlying these processes are poorly understood.

The Drosophila oocyte, in which axis-determining mRNAs are localized to various cortical regions, has been successfully used to study the specification of cell polarity and the establishment of polarized transport(St Johnston, 2005). During mid-oogenesis, at least two perpendicular subsets of MTs are formed,reflecting the dorsoventral (DV) and anteroposterior (AP) axes of the oocyte(Januschke et al., 2006; MacDougall et al., 2003). The axis-determining mRNAs are localized in a MT-dependent fashion(Riechmann and Ephrussi,2001): bicoid (bcd) mRNA is localized to the anterior cortex of the oocyte, oskar (osk) mRNA to the posterior pole, and gurken (grk) mRNA to an anterodorsal cap near the oocyte nucleus. In order to identify new factors involved in the polarized transport of mRNAs in the Drosophila oocyte, we carried out a germline mosaic screen for mutations on chromosome arm 2R that disrupt AP and/or DV axis formation. In the course of this screen, we identified a type I phosphatidylinositol 4-phosphate 5-kinase (PIP5K), Skittles (Sktl), as an essential factor for oocyte polarization.

Type I PIP5K synthesizes phosphatidylinositol 4,5 bisphosphate (PIP2) from phosphatidylinositol 4-phosphate. The membrane phospholipid PIP2 is involved in the control of cell polarity and plays a role in various cellular activities (Doughman et al.,2003). PIP2 regulates the membrane localization and activity of many cellular proteins via its specific interaction with phosphoinositide-binding domains (Downes et al., 2005). Although accumulated data suggest that PIP2 is an important regulator of actin-based cellular processes, in vivo analysis of type I PIP5Kα during development has been lacking. In Drosophila, Sktl has been shown to be required for chromatin-mediated gene regulation (Cheng and Shearn,2004) and is important in germline development(Hassan et al., 1998), but its cytoplasmic function remains to be determined.

In this work, we demonstrate that the PIP5K Sktl controls the PIP2 level at the plasma membrane. We show that sktl mutations disrupt the maintenance of oocyte polarity and cause defects in actin and MT organization. We provide evidence that PIP2 synthesis by Sktl is required to activate the actin-associated protein Moesin at the cortex. Moreover, our observations indicate that Sktl activity is required for cortical recruitment of the PAR proteins Bazooka (Baz), Par-1, Lethal (2) giant larvae [Lgl; L(2)gl - FlyBase]to the cell membrane. This study suggests that PIP2, by regulating several proteins, could mediate interactions between the plasma membrane, PAR proteins and the cytoskeleton that are essential for cell polarization.

The sktl^2.3 allele was generated in a mutagenesis screen (25 mM EMS) and was identified by sequencing the sktl gene amplified from heterozygous mutant adult DNA extracts. The sktl^M5 allele was generated in an EMS allelic screen for sktl and is lethal for all other sktl alleles. To identify the genomic location of the locus mutated in l(2)2.3, we used the Exelixis deficiencies collection (Parks et al.,2004). A lethal complementation analysis revealed that Df(2R)Exel6070 fails to complement l(2)2.3. Overlapping deficiencies,Df(2R)Exel7164 and Df(2R)Exel7166, complement the allele. This enabled us to locate the mutation within a 150 kb fragment containing thirteen genes. l(2)2.3 is lethal over sktl^Δ20. sktl^Δ20,sktl^Δ5 and sktl^Δ15 have been described(Hassan et al., 1998). Germline clones were generated by the FLP/FRT technique(Chou et al., 1993). nanosGAL4 VP16 (nosgal4) (Van Doren et al.,1998) and α4tubulin67c GAL4 (tubgal4)(Januschke et al., 2002) were used to express the transgenes during oogenesis. Other fly lines: GFP-Stau(Schuldt et al., 1998),Kin-lacZ (Clark et al., 1994),Dmn-GFP (Januschke et al.,2002), UAS-Baz-GFP and UAS-Baz^S151A,S1085A-GFP(Benton and St Johnston,2003b), UAS-GFP-Par-1 (N1S)(Doerflinger et al., 2006),UAS-GFP-Lkb1 (Martin and St Johnston,2003) and UAS-Lgl-GFP (Tian and Deng, 2008).

The polyubiquitin-PH-PLCδ-GFP construct was cloned from the pEGFPN1-PH-PLCδ plasmid into a pUbi-mod poly vector (a gift from Y. Belaiche, Curie Institute, Paris, France). The cDNA of sktl from DGRC(LP03320) was amplified, sequenced and subcloned into pUASP, pUASP:GFP,pUASP:RFP and pTub (a gift from P. Roth, Temasek Life Sciences Laboratory,Singapore) vectors to prepare Drosophila transgenes.

Immunolocalization and in situ hybridization were performed using standard protocols (Tautz and Pfeifle,1989; Wilkie and Davis,2001). Antibodies were used at the following dilutions:anti-β-galactosidase (Promega), 1:200; anti-Osk(Hachet and Ephrussi, 2001),1:2000; anti-Grk (DSHB), 1:200; anti-Stau(St Johnston et al., 1991),1:5000; anti-phospho-ERM (Cell Signaling Technology), 1:100; anti-Spn-F(CG12114) (Abdu et al., 2006),1:10; anti-DPLP (Cp309) (Kawaguchi and Zheng, 2004), 1:500; and anti-Baz(Wodarz et al., 2000), 1:2000. Western blots were probed with anti-phospho-ERM at 1:5000, anti-Moesin at 1:30,000, and anti-α-Tubulin at 1:2500. F-actin was visualized after staining with Alexa Fluor 680-phalloidin (Molecular Probes). DAPI and propidium iodide were used for DNA detection. Texas Red labeled Lycopersicon esculentum (LE) lectin (Vector Laboratories) was used at 150 μg/ml and Wheat germ agglutinin (WGA) at 1/100 (Molecular Probes). Anti-Khc (AKIN02-A, Cytoskeleton) was used for MT detection. Cold-shock experiments were performed as described(Januschke et al., 2006). Images were obtained with a Leica SP2 AOBS microscope.

We carried out a germline mosaic screen for mutations on chromosome arm 2R that disrupt AP and DV axis formation in the Drosophila oocyte. To visualize the AP axis, we used, as previously described(Martin et al., 2003), the protein Staufen (Stau) tagged with green fluorescent protein (GFP), which moves to the posterior of the oocyte with osk mRNA(Fig. 1A)(Bolivar et al., 2001; St Johnston and Nusslein-Volhard,1992). To visualize DV axis formation, we scored the positioning of the nucleus relative to the anterodorsal corner of the oocyte(Fig. 1A, asterisk). One homozygous mutant line affecting both nuclear positioning and Stau localization was identified in the screen and named l(2)2.3. In 57% of such egg chambers, Stau accumulated in the center of the oocyte(Fig. 1B), whereas in the remainder Stau was diffusely distributed in the cytoplasm (not shown)(n=118). The nucleus was randomly positioned within the oocyte at stages 9-10 (Fig. 1B,asterisk). Eggs deposited by l(2)2.3 mutant females were ventralized [loss of dorsal appendages 51%, fused appendages 43%, wild type (WT) 6%; n=116] (Fig. 1D), but always presented an aeropyle at the posterior(Fig. 1D, inset). Hence, these results indicate that l(2)2.3 affects AP and DV axis formation without impairing the early signaling step from the germline to the posterior follicle cells (Gonzalez-Reyes et al.,1995; Roth et al.,1995). Finally, posterior follicle cell clones mutant for l(2)2.3,marked by the loss of GFP, did not affect oocyte polarization, as revealed by the posterior localization of Stau (Fig. 1E), whereas mutant clones restricted to the germline impaired both nucleus positioning (Fig. 1F, asterisk) and Stau localization(Fig. 1F). This demonstrates that the l(2)2.3 phenotype is strictly germline dependent.

With the help of three overlapping deficiencies, we mapped l(2)2.3 to a 150 kb interval (see Materials and methods). We tested a series of lethal mutations in the genes of this region and found that l(2)2.3 is an allele of sktl. Whereas sktl^Δ20,which is reported to be a null allele(Hassan et al., 1998), is lethal in trans to l(2)2.3, sktl^Δ5 and sktl^Δ15 are female-sterile in trans to l(2)2.3. sktl^Δ5/l(2)2.3 females laid ventralized eggs and their oocytes had defects in the localization of both the nucleus and Stau (data not shown).

Molecular characterization revealed that the l(2)2.3 mutation comprises a small deletion (276 bp) in the unique intron of sktl, as do sktl^Δ5 and sktl^Δ15. l(2)2.3 results in a large reduction in sktl transcript levels in nurse cells and oocytes(Fig. 1, compare H with G),confirming that l(2)2.3 is a hypomorphic allele of sktl similar to sktl^Δ5 and sktl^Δ15. Finally, the lethality and oogenesis phenotypes were rescued by ubiquitous expression of a sktl cDNA transgene, confirming that these defects are caused by mutations in this gene. Germline clones, mutant for sktl^Δ20 and sktl^M5 (see Materials and methods), arrested oogenesis at stage 4 (see Fig. S1A in the supplementary material).

Early oogenesis can be separated into different steps: determination,polarization along the AP axis and, subsequently, the maintenance of this polarity (for a review, see Huynh and St Johnston, 2004). The early arrest found in null alleles of sktl is not due to oocyte determination defects. Indeed, Orb, a suitable marker for oocyte determination, was restricted as normal to one cell in the cyst (see Fig. S1I in the supplementary material). Furthermore, Orb,and the centrosomes as revealed by γ-Tubulin, were correctly translocated to the posterior of the oocyte suggesting that early AP polarization takes place normally in the mutant egg chambers (see Fig. S1G,I in the supplementary material). However, starting from stage 2, posterior components such as Orb started to lose their posterior restriction (see Fig. S1I, white arrowhead, in the supplementary material) and were eventually found mislocalized throughout the cyst. Thus, Sktl is not required for the determination of the oocyte during its polarization in region 3 of the germarium. However, Sktl appears to be required for the maintenance of oocyte polarity. We therefore conducted our analysis with the hypomorphic sktl alleles from which we were able to obtain vitellogenic oocytes,enabling us to study the function of Sktl during the later stages of oocyte polarization (mid-oogenesis).

We investigated whether Sktl is required for polarized transport, i.e. for asymmetric mRNA localization. To score for mRNA localization defects along the DV axis, we analyzed grk mRNA. In sktl^2.3germline clones, up until stage 8, grk mRNA accumulated above the nucleus in the anterodorsal corner of the oocyte(Fig. 2C), as in the WT(Fig. 2A). However, during stage 9, grk mRNA became mislocalized into the cytoplasm, where it accumulated around the mispositioned nucleus(Fig. 2D) (85%, n=64). Likewise, when analyzing mRNA localization along the AP axis, we found that bcd mRNA was localized correctly to the anterior margin up until stage 8 (Fig. 2G), but became mislocalized in the oocyte during stage 9, forming a ring around the mispositioned nucleus (Fig. 2H)(88%, n=43). In the WT, osk mRNA transport toward the posterior pole of the oocyte is achieved during stage 9, and Osk is subsequently translated at the posterior(Markussen et al., 1995; Rongo et al., 1995). During stage 8, when bcd and grk have reached their final location(Fig. 2A,E), osk mRNA accumulates temporarily in the center of the WT oocyte(Fig. 2I), before it reaches the posterior pole (Fig. 2J). In sktl mutant oocytes, osk mRNA localized normally to the center of the cytoplasm during stage 8(Fig. 2K). However, during the subsequent stages, it failed to reach the posterior and remained in the middle of the oocyte (Fig. 2L) (98%, n=57). Moreover, Osk protein was never translated in these mutants(see Fig. S2B in the supplementary material). These results indicate that Sktl is required for mRNA transport along the AP and DV axes of the oocyte.

In the oocyte, the anterodorsal positioning of the nucleus involves at least two distinct steps. First, during the transition from stage 6 to 7, the oocyte nucleus migrates from a posterior to an anterodorsal position in the cytoplasm. Then, the nucleus is anchored at the anterodorsal cortex. This attachment is a MT-based process involving at least two distinct mechanisms for anchoring to the anterior and the lateral cortex, respectively(Guichet et al., 2001; Januschke et al., 2002; Koch and Spitzer, 1983). In sktl^2.3 mutant oocytes, there was no defect in oocyte nucleus migration during stage 7 (data not shown), and at stage 8 the nucleus was always correctly localized (Fig. 2C,K, asterisks). However, in 80% of stage 9 (n=54) and 96% of stage 10 (n=48) oocytes, the nucleus was mislocalized(Fig. 2M). Optical cross-sections of sktl^2.3 mutant oocytes revealed that the mislocated nucleus was still tightly associated with the lateral cortex(Fig. 2N). These results indicate that Sktl is required for the anterodorsal maintenance of the nucleus and, more specifically, for its anterior anchorage.

Previous work has indicated that the nucleus appears to have an influence on MT organization (Guichet et al.,2001; Januschke et al.,2006). We investigated whether the inactivation of Sktl affects MT organization. In sktl^2.3 germline clones, we observed that up to stage 8, the MT network is organized as in controls(Fig. 3A,B). However, from stage 9 onwards, the anterior-to-posterior MT organization seemed to be lost in the sktl^2.3 germline clones. Indeed, the MT array was organized around the mislocalized nucleus(Fig. 3D), in contrast to the MT organization in the WT (Fig. 3C). We next analyzed the distribution of various markers for MT polarity. MT plus ends can be marked by a fusion of the Kinesin heavy chain with β-galactosidase (Khc-β-gal)(Clark et al., 1994). In WT oocytes, Khc-β-gal localizes to the posterior cortex during stage 9,revealing that the MT plus ends are pointing towards the posterior(Fig. 3E). In sktl^2.3 germline clones, Khc-β-gal was never detected at the posterior and, in 23% of the egg chambers (n=79), it accumulated in anterior regions of the cytoplasm(Fig. 3F). Likewise, with a different marker for MT plus ends, Dynamitin-GFP (Dmn-GFP),(Duncan and Warrior, 2002; Januschke et al., 2002), we found that the posterior localization of Dmn-GFP seen in WT was always lost in sktl^2.3 mutant oocytes, where it accumulated at the anterior in 33% of the cases studied (n=35)(Fig. 3, G versus H). This is in agreement with the results obtained with Khc-β-gal(Fig. 3F). Hence, it appears that MT plus ends can be ectopically located close to the anterior of the oocyte, indicating that sktl mutation prevents the correct orientation of the MTs.

Interestingly, Dmn is also a suitable marker for revealing the MT minus ends associated with the nucleus of the oocyte(Duncan and Warrior, 2002; Januschke et al., 2002)(Fig. 3G, red arrowhead). In sktl^2.3 germline clones, Dmn-GFP was found around the mislocalized nucleus (Fig. 3H,red arrowhead). In addition, MT minus ends in the WT are also distributed along the anterior margin of the oocyte(Clark et al., 1997; Theurkauf et al., 1993). These MT minus end extremities are especially highlighted by Spindle-F (Spn-F)(Fig. 3K)(Abdu et al., 2006) and can also be revealed by the localization of bcd mRNA(Fig. 2E,F)(Cha et al., 2001). In sktl^2.3 mutant oocytes, Spn-F and bcd mRNA were distributed along the anterior margin up to stage 8(Fig. 3L; Fig. 2G). However, from stage 9 onwards, Spn-F formed a cortical ring at the position of the misplaced nucleus(Fig. 3M). This localization is very reminiscent of the localization of bcd mRNA in sktl^2.3 mutant oocytes(Fig. 2H). Taken together,these observations suggest that MT minus ends are present as a cortical ring around the mispositioned nucleus.

Since the nucleus may function as a MT nucleation center in the oocyte(Januschke et al., 2006), we investigated whether the misplaced nucleus is still associated, as in WT, with centrosomal components such as DPLP (Cp309 - FlyBase)(Januschke et al., 2006). In both WT (Fig. 3I) and sktl mutant (Fig. 3J)oocytes, DPLP was localized around the nucleus, and as a bright dot in the vicinity of the nucleus possibly corresponding to the centrosome. In order to test the MT-nucleating capacity of the mispositioned nucleus, we used a cold-induced MT disassembly assay(Januschke et al., 2006). As in the WT, during the initial period of recovery at 25°C after complete depolymerization through cold-shock treatment(Fig. 3P), MT polymerization only took place in the immediate vicinity of the mispositioned oocyte nucleus(Fig. 3Q). Therefore, as in WT oocytes, MT nucleation in the sktl mutant is asymmetric and mainly restricted to the area surrounding the nucleus. However, we cannot exclude the possibility that we might have missed some MT nucleation activity at the cortex as a result of the experimental set-up. These results confirm the previously described role of the nucleus as the main active MTOC in the oocyte(Januschke et al., 2006). To conclude, in sktl mutants, the MT array is reorganized around the delocalized nucleus with a reverse posterior-to-anterior orientation and this is very likely to be responsible for mRNA mislocalization.

sktl encodes a putative ortholog of a type I PIP5K(Knirr et al., 1997) that catalyzes the phosphorylation of phosphatidylinositol 4-phosphate to generate the phosphatidylinositol 4,5 bisphosphate (PIP2), a major component of the plasma membrane. We examined the localization of a GFP-tagged Sktl protein by expressing a UASp-GFP-sktl transgene under the control of the germline-specific driver nanosGal4 (nosgal4). This combination rescues the phenotypes of sktl^2.3/sktl^2.3 and sktl^Δ5/sktl^2.3mutant oocytes, indicating that the GFP-sktl transgene is functional. From the early stages of oogenesis until the development of mature egg chambers, Sktl-GFP accumulated at the cortex just below the plasma membrane of the oocyte and nurse cells, where it colocalized with the actin cytoskeleton(Fig. 4A-D). A fraction of Sktl-GFP was detected as cytoplasmic particles(Fig. 4A). The localization of Sktl at the plasma membrane is consistent with its involvement in PIP2 synthesis (Oude Weernink et al.,2004).

In order to monitor the distribution and the level of PIP2, we generated a PLCδPH-GFP transgene containing GFP fused to the PIP2-specific pleckstrin-homology domain of phospholipase Cδ(Balla et al., 1998). During oogenesis, the PIP2 reporter was specifically distributed along the plasma membrane in both the germline and follicle cells(Fig. 4E,I, arrow and arrowhead). Moreover, PIP2 reporter colocalized with the glycosamyl-modified proteins present in the oocyte plasma membrane(Fig. 4L-N) and with the cortical actin cytoskeleton (Fig. 4E-H). Sktl (Fig. 4J) and PIP2 (Fig. 4I) colocalized along the oocyte cortex(Fig. 4K, arrow in the inset).

To address the potential PIP5K activity of Sktl at the oocyte plasma membrane, we analyzed the distribution of the PIP2 reporter in sktlmutant germline clones (Fig. 4O-Q). We observed a severe decrease in the level of PIP2 reporter all along the oocyte cortex (Fig. 4, compare arrows in the insets, O with L and Q with N). Our results strongly implicate Sktl in the control of PIP2 levels at the plasma membrane, suggesting that Sktl is required for PIP2 synthesis in the Drosophila oocyte.

How might Sktl regulate the positioning of the nucleus and the localization of mRNAs in the oocyte? The mutually antagonistic interactions between the PAR proteins Baz and aPKC at the anterior, and between Par-1 and Lgl at the posterior, are required for mRNA localization and MT organization in the oocyte during mid-oogenesis (Benton and St Johnston, 2003a; Martin and St Johnston, 2003; Shulman et al., 2000; Tian and Deng,2008; Tomancak et al.,2000). Furthermore, some elements of these PAR complexes, such as Lgl and Par-1, are also involved in the anterodorsal positioning of the nucleus (Doerflinger et al.,2006; Tian and Deng,2008). Interestingly, it has recently been shown that in MDCK cells, Par-3 recruitment to the plasma membrane requires PIP2(Wu et al., 2007). Thus, we examined whether Sktl activity is required for Baz localization. In the oocyte, Baz-GFP localizes strongly at the anterior and lateral cortex, but is excluded from the posterior (Fig. 5A) (Benton and St Johnston,2003a; Benton and St Johnston,2003b). In sktl^2.3/sktl^Δ5 mutant oocytes, Baz-GFP was less enriched along the anterolateral cortex, but became distributed in the cytoplasm (Fig. 5B) (52%, n=63). Furthermore, the requirement for Sktl to control Baz localization is not restricted to the oocyte because in follicle cell clones mutant for sktl^M5, the apical restriction of Baz was lost and the protein became ectopically distributed in the cytoplasm(Fig. 5C,D). Thus, Sktl activity is required to recruit Baz at the membrane in the oocyte and epithelial follicle cells. Because the deficit of PIP2 might have a global effect on the oocyte cortex, we tested whether Lkb1 (the DrosophilaPAR-4 homolog), which is known to be bound to the plasma membrane through a prenylation motif (Martin and St Johnston,2003), was still localized in the absence of Sktl. We found that in sktl^2.3 germline clones, GFP-Lkb1 localized to the oocyte cortex as in WT (Fig. 5E,F). This indicates that the delocalization of Baz observed in sktl mutants is due to the decrease in PIP2 level, rather than to a complete disorganization of the cortex.

Previous analysis revealed that the restricted localization of Baz is essential for oocyte polarity. Indeed, overexpression of a mutated form of Baz that cannot be phosphorylated by Par-1 (Baz^S151A,S1085A-GFP),induced a delocalization of Baz all along the cortex and some accumulation in the cytoplasm associated with penetrant oocyte polarity phenotypes(Benton and St Johnston,2003b). Interestingly, we found that the overexpression of this mutated form of Baz affected the positioning of the nucleus, as in sktl mutants (Fig. 5G). Thus, we questioned whether Sktl and Baz might act together to polarize the oocyte. Using a sktl^2.3/sktl^Δ15hypomorphic combination, in which the delocalization of the oocyte nucleus occurs in only 31% of cases examined (Table 1), we found that the nucleus was mislocalized in 63% when the gene dosage of baz is lowered(Table 1). Thus, sktland baz interact genetically. Taken together, these results suggest that the targeting of Baz to the cortex, as mediated by Sktl, could be required for the positioning of the nucleus and for MT organization.

Cross-regulatory interactions between the aPKC-Baz complex and Lgl and Par-1 are crucial in order to establish complementary cortical domains in polarized cells. It has been shown that aPKC phosphorylation of Lgl restricts Lgl activity and Par-1 enrichment to the posterior of the oocyte(Tian and Deng, 2008)(Fig. 5H,J). Thus, we analyzed the localization of the fusion proteins Lgl-GFP and GFP-Par-1 (N1S isoform) in sktl^2.3/sktl^Δ5oocytes. Interestingly, Lgl and Par-1 were no longer restricted to the posterior but became distributed uniformly along the cortex and started to accumulate in the cytoplasm (Fig. 5I,K) (Lgl mislocalization, 100%, n=30; Par-1 mislocalization, 100%, n=8). These results indicate that Sktl is required for the maintenance of complementary compartments at the oocyte cortex along the AP axis, and by this means might regulate oocyte polarity during mid-oogenesis.

Interestingly, in a hypomorphic combination of sktl^2.3and sktl^Δ15 alleles, the delocalization of the oocyte nucleus occurred in only 31% of the oocytes, as compared with 90% in sktl^2.3/sktl^2.3(Table 2), and the mislocalization of bcd and grk mRNAs was similarly decreased. However, osk mRNA was still found diffusely distributed in the cytoplasm of 62% of the oocytes (n=47)(Table 2). This suggests that Sktl might also control osk mRNA by a mechanism independent of nucleus positioning. Previous work has shown that the anchorage of osk mRNA to the posterior oocyte cortex is actin dependent(St Johnston, 2005). We addressed the relationship between Sktl and actin organization. In WT oocytes,the microfilaments form a continuous layer along the entire cortex and the orientation of the actin bundles is parallel to the plasma membrane(Robinson and Cooley, 1997)(Fig. 6C-E). Sktl colocalizes with actin (Fig. 4C,D). In the sktl^2.3 mutant, the organization of actin filaments appeared to be normal during early oogenesis (data not shown), but from stage 8 onwards we detected two distinct types of actin defect. In 36% of oocytes(n=118), the cortical actin network was disrupted at the border between the anterior margin and the lateral cortex(Fig. 6H-J). In 15%, the actin microfilaments in the oocyte were loosely bound to the lateral cortex,including to the posterior pole, and they delaminated into the cytoplasm(Fig. 6M-O). It is important to emphasize that the continuity of the plasma membrane was unaffected in the sktl mutant, as revealed by LE lectin labeling(Fig. 6G,L). Thus, impairing Sktl function leads to disorganization of the microfilament scaffold along the oocyte cortex and to detachment of cortical actin. Mutations in several actin-related genes disrupt mRNA localization by inducing premature MT-based cytoplasmic streaming (Emmons et al.,1995; Manseau et al.,1996; Theurkauf,1994). However, in sktl mutant oocytes, we did not observe the early movement (stages 7-9) of yolk particles that is characteristic of premature cytoplasmic streaming (data not shown).

Several proteins known to modulate the actin cytoskeleton are regulated by PIP2 (Niggli, 2005). Among these, Moesin, which links the actin cytoskeleton to the plasma membrane, is required for osk mRNA localization without affecting cytoplasmic streaming (Jankovics et al.,2002; Polesello et al.,2002). Previous ex-vivo studies have shown that Moesin activation is promoted by PIP2 binding to its FERM domain(Barret et al., 2000; Niggli et al., 1995). We evaluated whether Sktl function is required for Moesin activation by examining the level of Moesin phosphorylation in sktl mutants using an antibody specific to phosphorylated ERM proteins(Polesello et al., 2002). In the WT, we detected the phosphorylated form of Moesin (P-Moe) all along the cortex of the oocyte and nurse cells (Fig. 7A). In sktl^2.3 germline clones, P-Moe distribution at the cortex was lost (Fig. 7B). Accordingly, an immunoblot analysis showed that P-Moe was dramatically reduced in sktl^2.3/sktl^Δ5 egg chambers (Fig. 7C), whereas the level of total Moesin remained unchanged(Fig. 7D). Taken together,these results indicate that Sktl is required for the activation of Moesin in the oocyte and could explain the defects in actin organization in the sktl mutant.

Binding interactions between the plasma membrane and the cytoskeleton are important for the establishment of cell polarity. PIP2 is essential for the maintenance of these interactions by influencing the activity of several proteins that regulate the architecture of the cytoskeleton. This study provides in vivo evidence of a function for PIP5KIα in cell polarization.

Our results indicate that the type I PIP5K, Sktl, is essential for PIP2 synthesis in the Drosophila oocyte. PIP2 directly controls the localization and activity of many proteins via its interaction with phosphoinositide-binding domains. Among them, the activation of ERM proteins such as Moesin, results in the unmasking of their functional binding sites. Our study indicates that in vivo, PIP2 provided by a PIP5K is required for Moesin phosphorylation, supporting results obtained previously with cellular systems (Fievet et al., 2004; Lacalle et al., 2007). Since PIP2 is also required for PIP3 synthesis, it is possible that Sktl also affects PIP3 production. It will be interesting to investigate the requirements of PIP3Ks during middle oogenesis and to compare them with those of Sktl.

We showed that Sktl is necessary for the positioning of the oocyte nucleus,for MT organization and for the localization of grk, bcd and osk mRNAs. The nature of the MT regrowth after cold-induced MT disassembly in sktl mutants is consistent with previous observations suggesting that the nucleus functions as the main MT nucleation center in the oocyte, influencing the organization of the MT network and, thereby, mRNA asymmetric transport (Guichet et al.,2001; Januschke et al.,2006). Since MT minus ends are found as a cortical ring around the nucleus, we propose a model whereby, in sktl mutants, MTs nucleated from the nucleus might be translocated to, and anchored at, the surrounding cortex, as previously proposed for anchoring at the anterior margin in the WT(Guichet et al., 2001;Januschke et al., 2008).

How could Sktl control the anchoring of the nucleus? Recent results in the C. elegans embryo indicate that PPK-1, a PIP5K, controls spindle movements by regulating the heterotrimeric G proteins GPR-1/2 and LIN-5, which are similar to Pins (Raps - FlyBase) and Mud (Mushroom body defect),respectively, in Drosophila(Panbianco et al., 2008). Pins requirement has not been reported in the oocyte. However, Mud is distributed around the nucleus (Yu et al.,2006), like the Dynein-Dynactin complex with which it has been reported to control spindle attachment in other systems(Gonczy, 2008). However,although the positioning of the nucleus is Dynein dependent(Januschke et al., 2002), it does not necessarily require Mud (Yu et al., 2006). It would however be interesting to investigate whether Pins and Mud act redundantly with other factors to control the positioning of the nucleus.

Our results also indicate that Sktl regulates PAR polarity proteins along the AP axis. In the absence of Sktl function, Baz, Lgl and Par-1 are mislocalized in the oocyte. In MDCK cells, PIP2 has been shown to bind to Par-3 and to participate in its recruitment at the plasma membrane. In the absence of Sktl, Baz localization is affected in both the oocyte and follicle cells. One hypothesis is that in the absence of a sufficient level of PIP2,Baz is not recruited at the cortex of the oocyte, compromising the equilibrium between anterior and posterior PAR complexes and inducing the mislocalization of Lgl and Par-1. It is also possible that in the absence of Sktl, the defective actin cytoskeleton in the oocyte compromises the localization of Lgl, as occurs in neuroblasts (Betschinger et al., 2005), and of Par-1 (N1S) to the posterior cortex(Doerflinger et al.,2006).

PAR proteins are well-known regulators of MT organization and MT-based transport in polarized cells (Munro,2006). Furthermore, ectopic expression along the entire cortex of Par-1 (N1S) (Doerflinger et al.,2006), Lgl (Tian and Deng,2008) or Baz (this study), as well inactivation of lgl(Tian and Deng, 2008), affect the positioning of the nucleus, as in absence of Sktl. This further suggests that the correct positioning of the PAR proteins along the AP axis is crucial for the anterodorsal anchorage of the nucleus. It is however possible that the defects in the localization of mRNAs observed in sktl mutant oocytes are not only caused by the mispositioning of the nucleus, but also involve a more direct effect of the defective distribution of PAR proteins on MT organization.

It is interesting to note that in the C. elegans embryo, PAR proteins such as PAR-2 and PAR-3 regulate the activity and the distribution of the PIP5K, PPK-1, but that PAR-2 and PAR-3 are not regulated by PPK-1(Panbianco et al., 2008),whereas in the Drosophila oocyte Sktl controls the distribution of the PAR proteins. Hence, phosphoinositide regulators and PAR proteins are closely associated in the control of polarity establishment in different organisms; however, they can act at different levels relative to each other.

We thank S. Delga and R. Naaman for their work on Skittles; C. Moch for the design of pUbi vectors; F. Payre, D. St Johnston, H. Bellen and W. Deng for reagents; J. Compagnon for suggestions; and J. R. Huynh, A. Gautreau, J. Rothman, I. Becam and A. Kropfinger for critical comments on the manuscript. Drosophila embryo injections were carried out by BestGene. L.G. was supported by a fellowship from `Ligue contre le cancer'. This work was supported by grants from the CNRS, Association pour la recherche sur le cancer(ARC; subvention number 4446, 3297), ACI `Jeune chercheur' programme of the Ministere de la Recherche, ANR Blanche (grant Cymempol, Blan06-3-139786).