Clathrin heavy chain plays multiple roles in polarizing the Drosophila oocyte downstream of Bic-D

Studies in different organisms portray Bic-D homologs as factors that link the dynein-dynactin minus end-directed microtubule (MT) motor complex to different cargoes, moving them to particular cellular regions (Claussen and Suter, 2005; Vazquez-Pianzola and Suter, 2012). Drosophila Bic-D (also known as BicD) directly binds Egalitarian (Egl), which engages with Dynein light chain (Dlc; also know as Ctp) and specific mRNA localization signals (Navarro et al., 2004; Dienstbier et al., 2009). This complex is thought to associate with further proteins that regulate translation and stability of transported mRNAs, and the resulting larger complex translocates with the help of motors along the MT cytoskeleton. The Bic-D-dependent transport machinery is used repeatedly during Drosophila development. In oogenesis, Bic-D is needed for the transport of osk, bcd and grk mRNAs from the nurse cells into the oocyte and then to specific compartments within the oocyte (Suter and Steward, 1991; Ran et al., 1994; Clark et al., 2007). Correct localization of these factors within the oocyte is crucial for specifying anteroposterior and dorsoventral axes of oocytes and embryos. Later in the life cycle, the mRNA localization machinery delivers specific mRNAs to the apical side of the syncytial embryo (Bullock and Ish-Horowicz, 2001) and promotes localization of inscuteable mRNAs in neuroblasts (Hughes et al., 2004).

Drosophila Bic-D binds mRNA cargoes not only through the cargo-binding adaptor Egl, but also through the Fragile-X Mental Retardation Protein (Fmr1) (Bianco et al., 2010). Mammalian BicD2 regulates centrosome and nuclear positioning during mitotic entry (Splinter et al., 2010) and mammalian Bic-D isoforms bind Rab6 to control COPI-independent Golgi-endoplasmic reticulum transport (Matanis et al., 2002). Furthermore, Rab6B-BicD1 interaction regulates retrograde membrane transport in human neurites (Wanschers et al., 2007). Fly and worm Bic-D genes are also involved in nuclear migration in photoreceptors, oocytes and hypodermal precursor cells (Swan et al., 1999; Swan and Suter, 1996; Fridolfsson et al., 2010), and Bic-D also dynamically regulates transport of lipid droplets (Larsen et al., 2008). Given the involvement of Bic-D in the localization of surprisingly diverse cargoes, we searched for adaptor proteins that mediate cargo binding as well as for novel cargo molecules.

Using two different monoclonal anti-Bic-D antibodies that recognize different epitopes, we immunopurified Bic-D complexes and identified Chc as a complex component in both approaches through mass spectrometry (Fig. 1A). Beads alone and beads coupled to control antibodies did not precipitate Chc. Myc-tagged Chc expressed under UAS control and a Flag-Tetracysteine (4C)-tagged Chc expressed from its endogenous promoter (Kasprowicz et al., 2008) were also immunoprecipitated with both anti-Bic-D antibodies from embryonic (Fig. 1B) and ovarian (Fig. 1C) extracts. Li et al. showed that Chc interacts directly with Bic-D and is its major interactor in the nervous system (Li et al., 2010). In contrast to this, Bic-D interacts with Egl, Chc, Pabp and others in embryos (Fig. 1A) (Vazquez-Pianzola et al., 2011; P.V.P. and B.S. unpublished data). Thus, in non-neuronal tissues, such as young embryos and ovaries, Chc appears to be one of many partners of Bic-D.

Anti-Bic-D antibody staining of ovaries that were fixed in a healthy state does not reveal aggregated structures, suggesting that native complexes are small particles that cannot be resolved with normal confocal microscopy. Colocalization studies can therefore only reveal whether Bic-D and Chc colocalized in the same cellular compartment. Indeed, a fraction of differently tagged (V5, Myc, Flag and GFP) or un-tagged Chc colocalizes with Bic-D in specific regions of the oocyte (Fig. 1D-H; data not shown). The germ line Chc signal became enriched in the oocyte from early stages onwards (Fig. 1D,F) and further signal enrichment became apparent at the posterior cortex of the oocyte from stage 6 onwards (Fig. 1E-G). During stage 6, Chc shows a partial colocalization with Bic-D and Egl at the posterior cortex (Fig. 1H; supplementary material Fig. S1). From stage 8-9 onwards, Chc signals localized to the oocyte cortex with higher enrichment at the posterior pole (Fig. 1E,G). Another partial colocalization between Bic-D/Egl and Chc is evident at the anterior cortex in stage 8 oocytes (Fig. 1H; supplementary material Fig. S1B).

As Clathrin activity in endocytosis involves also Clathrin light chain (Clc), we also studied Clc localization (Fig. 1I; supplementary material Fig. S1). Anti-Clc antibody staining also revealed Clc enrichment in the oocyte from early stages onwards. Clc concentrates at the posterior cortex, partially colocalizing with Bic-D in stage 6-7 oocytes. However, at stage 8, Clc is more uniformly distributed throughout the oocyte cortex and no clear posterior enrichment is observed from stage 8 onwards (Fig. 1I; supplementary material Fig. S1). Co-staining for Chc and Clc showed that, although both proteins are enriched in the oocyte and specifically at the oocyte cortex, they do not completely colocalize during development (supplementary material Fig. S1). This finding could suggest that Chc may play roles in oocyte development independently of Clc.

Bic-D and Egl seem to localize normally in Chc⁴ mutant ovarioles. The converse experiment, testing the function of Bic-D in later processes of oogenesis, is complicated by the fact that loss of Bic-D prevents oocyte formation (Ran et al., 1994). To deplete the germ line of Bic-D protein after oocyte determination, we used Bic-D^mom flies in which Bic-D is provided solely from an inducible promoter that can be turned off once oocyte fate is established (Swan and Suter, 1996). Around 4 days after shutting down Bic-D, these ovaries contain mid-oogenesis germ line devoid of Bic-D. Once Bic-D protein became undetectable, Chc protein accumulation in the oocyte and at the oocyte cortex became reduced as well (Fig. 2A), even though Chc levels were not affected (Fig. 2B). Therefore, in the female germ line Bic-D acts upstream in localizing Chc.

The dominant mutant Bic-D alleles affect the localization of the cargo osk mRNA (Ephrussi et al., 1991; Kim-Ha et al., 1991). We therefore tested whether they also affect Chc localization. In stage 10 oocytes heterozygous for the dominant Bic-D alleles Bic-D^IIIE48 or Bic-D^71.34 or Bic-D^null, Chc accumulated strongly at the posterior in most cases (79%, n=44 in Bic-D^71.34/+ or Bic-D^null/+; 89%, n=38 in Bic-D^IIIE48/+ or Bic-D^null/+). Thus, only 10-20% of oocytes with one normal copy of Bic-D showed reduced accumulation of Chc at the posterior. By contrast, posterior accumulation of Chc was reduced in 49% (BicD^IIIE48/Bic-D^null; n=49) and 44% (Bic-D^71.34/Bic-D^null; n=99) of the mutant stage 9-10 oocytes (Fig. 2C). Interestingly, the observed effects are not due to an abolished interaction between Chc and the dominant mutant Bic-D protein because Chc is efficiently precipitated with Bic-D from Bic-D^IIIE48/Bic-D^null and Bic-D^71.34/Bic-D^null mutant extracts (not shown). Furthermore, Chc protein levels are also normal in these ovaries (Fig. 2D,E). Because a significant portion of dominant Bic-D mutant protein accumulates strongly at the anterior cortex, it appears that there may not be sufficient functional Bic-D available to transport Chc cargo to the posterior cortex.

Because Bic-D transports mRNAs into the oocyte, we analyzed the distribution of Chc mRNA in wild-type and Bic-D^mom ovaries (Fig. 3A,B). Once the female germ line was depleted of Bic-D, Chc mRNA accumulation in the oocyte also became reduced, even though total levels of Chc mRNA remained normal (Fig. 3C). Bic-D thus functions to localize Chc mRNA into the oocyte, rather than in stabilizing the transcript. Similar to the mRNA levels, ovarian Chc protein levels were also normal in Bic-D^mom mutants (Fig. 2B). Reduced oocyte accumulation of Chc protein in Bic-D-depleted egg chambers therefore seems to be due to the lack of Chc mRNA transport. To test whether the Bic-D/Egl localization machinery associates with Chc mRNA, we immunopurified Bic-D/Egl::GFP complexes. This procedure resulted in enrichment of Chc mRNAs to a similar level as the enrichment of the known Bic-D/Egl targets ftz, hairy and osk (Fig. 3D). By contrast, Kr, a non-localizing mRNA, and the housekeeping mRNAs rp49 (also know as RpL32) and Tub67C were much less enriched in these immunoprecipitates.

Because Bic-D is involved in localizing Chc mRNA and protein, it became important to assess the dynamics of Chc::GFP transport in the presence and absence of Bic-D in the female germ line (Fig. 4). As seen for Chc in fixed samples, Chc::GFP accumulates at the oocyte cortex in the wild type, but much less so in Bic-D^mom oocytes (Fig. 4A). Chc::GFP spots move in strictly linear directions in mutant and wild-type oocytes, suggesting that Chc::GFP is transported along MT tracks. Clusters of similar size and shape to the moving spots are located in the vicinity of the cortex in Bic-D^mom oocytes (Fig. 4B) but not in wild-type oocytes, in which the strong accumulation of cortical Chc::GFP leads to a more homogenous and brighter signal (Fig. 4B). Particles moved frequently and with similar directionality along the oocyte cortex in wild type and mutants (Fig. 4C,D), but average particle speed was significantly higher in mutants (Fig. 4E). It therefore seems that lack of Bic-D affects movement of Chc on MTs and that this defect prevents cortical accumulation of Chc protein. The increased particle speed observed upon loss of Bic-D may point towards an inhibitory function of Bic-D and dynein-mediated transport towards kinesin-based transport. Interestingly, the particle speed for Rab6, another Bic-D cargo, increases also after inhibition of the dynein-dynactin interaction by expressing p50-dynamitin (Januschke et al., 2007). Furthermore, overexpression of Chc::GFP in the hypomorphic and the dominant Bic-D mutant background enhanced the Bic-D-dependent osk mislocalization phenotypes. Different cargoes competing for limited Bic-D could lead to such an enhancement of the hypomorphic phenotype. Alternatively, Chc overexpression could modify the cortex in a way that improves anchoring of ectopic osk mRNA (supplementary material Fig. S2).

Although Chc¹, Chc² and Chc³ are lethal alleles, Chc⁴ is a non-lethal mutant allele that causes a remarkable reduction in endocytic capacity and a wide range of phenotypes, including sterility (Bazinet et al., 1993; Peralta et al., 2009). In homozygous, hemizygous and transheterozygous (Chc¹/Chc⁴) Chc⁴ females, oogenesis arrests mostly at stage 8, but escaper follicles develop into eggs. This Chc⁴ oogenesis arrest phenotype is rescued by different Chc⁺ constructs, revealing also that differently tagged transgenes had Chc⁺ rescuing activity. A single genomic copy of Chc⁺ with a Flag-tag (4C-Chc) (Kasprowicz et al., 2008) rescued sterility, poor viability and locomotive defects of Chc¹/Chc⁴ females, and two copies of this transgene also rescued Chc¹ lethality. pUASP-Chc-V5, pUASP-Chc::eGFP (Li et al., 2010) and pUASP-myc-Chc, driven by an Actin-Gal4 driver, were able to rescue the sub-lethality and sterility of Chc⁴/Chc¹ females, but not the Chc¹ lethality. The reduced rescue efficiency could be due to low or non-ubiquitous expression of the Gal4 driver or the pUASP vector. Chc⁴ mutant egg chambers showed an overall reduced size, smaller oocytes and smaller, but normally localized, oocyte nuclei (Fig. 5A). Receptor-mediated endocytic yolk uptake into the oocyte depends on Clathrin and is initiated normally around stage 8 (Tanaka et al., 2011). Chc⁴ mutant oocytes and mutant larval Garland cells showed reduced endocytosis (Peralta et al., 2009) (data not shown).

In Chc⁴ mutants, maternal mRNAs seemed to be imported normally from the nurse cells into the oocyte and grk, orb, Bic-D and Chc mRNAs showed mostly normal localization (Fig. 5B). By contrast, osk mRNA accumulated in oocytes, but it was found uniformly distributed in Chc⁴ mutant oocytes or ectopically localized in the middle of stage 9-10 oocytes unless a rescue transgene was provided (Fig. 5C). osk mRNA localization at the anterior and posterior cortex of stage 8 oocytes was mostly not observed in mutant egg chambers. Osk protein levels remain low in Chc⁴ stage 9 egg chambers, and 12 out of 20 egg chambers did not show Osk expression at the posterior cortex compared with only three out of 20 in the wild type (data not shown). Germ line clones of the lethal allele Chc^GF23 (Yan et al., 2009) showed morphological defects, but, nevertheless, grk mRNA localization was mostly normal (Fig. 6A). By contrast, osk mRNA failed to enrich at the oocyte posterior from stage 6 onwards (Fig. 6B), further supporting the view that Chc is needed more specifically for proper localization of osk mRNA.

Mutations in Rab11 and Rabenosyn-5 (Rbsn-5), two genes involved in endocytosis, cause defects in osk mRNA localization (Jankovics et al., 2001; Dollar et al., 2002; Tanaka and Nakamura, 2008). However, as opposed to Rab11 and Rbsn-5, Chc⁴ affects osk mRNA localization earlier than stage 9. We therefore tested whether problems in MT organization in Chc⁴ egg chambers could lead to osk mRNA mislocalization (Fig. 7). Indeed, Kinesin heavy chain (Khc), a marker for stable MT plus ends, is enriched at the posterior cortex only in wild-type oocytes and in rescued mutants, but not in Chc⁴ mutants (Fig. 7A). Posterior enrichment of EB1, a marker for growing MT plus ends, was also affected in stage 6-9 Chc⁴ mutant egg chambers (Fig. 7B). Furthermore, α-Tubulin staining showed that MTs are enriched at the posterior oocyte cortex at stage 6, before they re-focus at the anterior and anterodorsal cortex by stage 8 (Fig. 7C). In Chc⁴ mutant egg chambers, α-Tubulin shows either a dotted pattern or no staining at the posterior cortex in stage 6. Also, stage 8 oocytes do not show anterior α-Tubulin accumulation, indicating again problems in focusing the MT network. In agreement, focusing of MTs was also not observed in germline clones of the Chc^GF23 allele (supplementary material Fig. S3). Such early functions in MT organization were not described for other genes involved in the endocytic pathway (Jankovics et al., 2001; Dollar et al., 2002; Tanaka and Nakamura, 2008). Revisiting Rbsn-5 and Rab5 mutant phenotypes in our laboratory, we confirmed normal localization of Khc in early oogenesis and that Khc localization was only affected after stage 9 (supplementary material Fig. S4). Therefore, Rbsn-5 and Rab5 (involved in the formation of early endosomes) mutations that interfere with endocytosis, affect MT organization only after stage 9 of oogenesis. By contrast, Chc is involved already earlier, around stage 6, in the organization of MT polarity and the MT network. Because bulk endocytosis starts only after MT reorganization at stage 8 and because the other endocytic genes tested are not important for MT organization prior to stage 9, it appears that Chc may play an endocytosis-independent role in MT organization around stage 6. Interestingly, in germ lines lacking Bic-D, Khc is also delocalized and detached from the posterior cortex (Fig. 7D) and α-Tubulin staining revealed abnormal MT organization (data not shown). Even though the defects observed in Chc mutants are clearly stronger than the ones observed in the hypomorphic Bic-D^mom situation, this result indicates that, like Chc, Bic-D is also involved in controlling MT organization.

Clathrin plays a major role in the formation of coated vesicles during endocytosis. As Bic-D interacts with Chc and is needed for proper localization of Chc in oocytes (Figs 1, 2), we analyzed the levels of endocytosis in BicD^mom egg chambers by visualizing FM4-64 uptake and autofluorescence of yolk granules (Fig. 8A,B). Wild-type stage 8-9 egg chambers show an accumulation of FM4-64 in the follicle cells and at the oocyte cortex with a higher accumulation at the posterior (Fig. 8A). Similarly, yolk granules accumulate at high amounts and are clearly enriched at the posterior in wild-type oocytes (Fig. 8A). However, in Bic-D^mom egg chambers in which Bic-D has become undetectable (labeled ‘day 4’), FM4-64 dye uptake and yolk granule autofluorescence are reduced and mostly lack enrichment at the posterior cortex (Fig. 8A,B). Therefore, Bic-D is needed to support oocyte endocytosis. Elevated levels of posterior endocytosis have been linked to the establishment of posterior identity (Vanzo et al., 2007; Tanaka and Nakamura, 2008). To find out whether the Bic-D-mediated elevated accumulation of Chc at the posterior cortex can be linked to the elevated endocytosis at the posterior pole, we took advantage of a dominant Bic-D allele that prevents this elevated accumulation of Chc at the posterior pole. In Bic-D^71.34/Bic-D^null oocytes, FM4-64 dye uptake is also reduced (Fig. 8A). In addition, these oocytes show reduce yolk granule autofluorescence and, although they still show more yolk granules towards the posterior, these granules are smaller and not clearly enriched at the very posterior cortex where they concentrate in wild-type oocytes (Fig. 8A,B). Interestingly, in synapses lacking functional Chc small vesicles fail to be formed and these synapses contain larger vesicles (Kasprowicz et al., 2008) This is consistent with Bic-D-dependent posterior localization of its interaction partner Chc being an essential step in upregulating posterior endocytosis and patterning. Similarly, the absence of Bic-D and the presence of the Bic-D^71.34 mutation interfere with posterior oocyte localization of Rab11 and Rbsn-5 (Fig. 8C,D).

Chc and Bic-D are present in the same complexes in embryos and ovaries, where their functions show considerable overlap. As opposed to the nervous system, in which Li et al. describe Chc as the major Bic-D-associated protein in Drosophila heads (Li et al., 2010), we found that Chc is one of many Bic-D partners in ovaries and embryos. The magnitude of these differences suggests that they are not solely caused by the different purification protocols employed. Instead, they seem to reflect differences in tissue-specific complex formation and additional activities of Bic-D during oogenesis and embryogenesis that are mediated by different adaptor and cargo proteins.

Chc is the main structural component of clathrin-coated pits and vesicles, and a central player in clathrin-mediated endocytosis (Brodsky, 2012; Faini et al., 2013). The association of Bic-D with Chc led us to discover that Bic-D is needed for endocytosis in the oocyte through two different mechanisms. On one hand, Bic-D seems to localize Chc mRNA into the oocyte. But in addition, Bic-D also appears to transport Chc protein to the posterior oocyte cortex. In Bic-D dominant mutants Chc mRNA localization is normal, but Chc protein is reduced at the posterior and much Bic-D protein remains anteriorly, pointing to insufficient protein transport towards posterior. Direct observation of moving particles also show that lack of Bic-D alters the dynamics of Chc::GFP transport in oocytes.

Li and co-workers proposed that Bic-D is needed for localization of Chc to periactive zones through a role in synaptic vesicle recycling (Li et al., 2010). A Chc recycling mechanism may be at work at the oocyte cortex, but our data are also consistent with a function of Bic-D in transporting Chc to the oocyte and then to its cortex. It seems possible that both mechanisms are at work in the female germ line, but further studies are needed to clarify their individual contribution and to find out whether they occur simultaneously or successively.

From oogenesis stage 8 on, Chc also functions in posterior localization of osk mRNA. However, as opposed to Bic-D and Egl, Chc does not get recruited to localizing mRNAs injected into syncytial embryos (data not shown), suggesting a more indirect role in mRNA transport. At stage 6, slightly before osk mRNA localization defects become visible, Chc is involved in organizing the MTs that are needed for setting up anteroposterior oocyte polarity. By contrast, Rbsn-5 and Rab11, encoding components of the endocytic and endocytic recycling pathway, respectively, are involved in osk mRNA localization only at stage 9 (Jankovics et al., 2001; Dollar et al., 2002; Tanaka and Nakamura, 2008) and Rbsn-5 and Rab-5 null mutants show MT polarity defects also only after stage 9 (Tanaka and Nakamura, 2008) (supplementary material Fig. S4). Therefore, although many endocytosis-related genes play roles later in oogenesis for MT maintenance, Chc is needed for the establishment of MT organization already around oogenesis stage 6. Furthermore, establishment of MT polarity by Chc appears to be endocytosis independent because Rbsn-5^null mutant egg chambers (showing strong endocytosis defects) and Rab-5^null mutants (also required for vesicle fusion in early endocytosis) do not reveal a requirement for this process (Tanaka and Nakamura, 2008; Morrison et al., 2008) (supplementary material Fig. S4).

Interestingly, mammalian Chc (CHC17; also known as CLTC) associates with mitotic spindles and forms a complex with the MT-associated proteins TACC3 and ch-TOG (also known as CKAP5) (Royle et al., 2005; Booth et al., 2011). This complex stabilizes kinetochore fibers by mediating short-distance interactions between adjacent MTs (Booth et al., 2011). Similar to Chc, Drosophila ch-TOG, encoded by mini spindles (msps), is also required for organization and structure of oocyte MTs and for the mRNA localization of bicoid (bcd) mRNA (Moon and Hazelrigg, 2004). However, the hypomorphic msps mutants did not show defects in osk mRNA localization. Stronger reduction of msps activity may help in determining whether the hypomorphic msps mutants still provide sufficient msps activity to localize osk mRNA or whether the roles of Chc in MT organization and osk mRNA localization can be separated. Chc-coated vesicles also form at the trans-Golgi network (TGN). Interestingly, non-centrosomal MTs nucleating at the Golgi have been found to be indispensable for establishing proper cell polarity (Ori-McKenney et al., 2012; Zhu and Kaverina, 2013). Further studies should therefore address whether Chc organizes oocyte MT polarity by nucleating MT at the TGN.

Incorporating our results into a model, we show how endocytosis, pole plasm assembly and anchoring of the pole plasm are tightly linked processes involving feedback loop regulation (Fig. 9). Bic-D contributes to osk mRNA localization and pole plasm assembly in two ways. First, from earlier studies we learned that the dynein adaptor Bic-D acts directly by transporting osk mRNA from the nurse cells into the oocyte (Suter and Steward, 1991; Swan and Suter, 1996; Clark et al., 2007). Here, we found that Bic-D controls the endocytic pathway also by transporting Chc mRNA into the oocyte and by localizing Chc protein to the oocyte cortex and enriching it at the posterior cortex. Bic-D is additionally involved in polarizing the endocytic pathway by directly or indirectly (e.g. through localization of Chc or osk mRNA/Osk protein) localizing the endocytic proteins Rab11 and Rbsn-5. Thus, through these mechanisms, Bic-D promotes endocytosis at the posterior, which then stimulates pole plasm anchoring and osk mRNA localization at the posterior. Chc is also needed for establishing the MT cytoskeleton during the middle stages of oogenesis and it may be needed for the maintenance of this A-P polarity. In summary, we have found a novel and central role for Chc and Bic-D proteins in controlling pole plasm assembly by affecting MT polarity, endocytosis and osk mRNA localization.

Gal4 lines, Chc⁴ and Chc¹ stocks were obtained from the Bloomington Drosophila Stock Center. The transgenic genomic rescue constructs pCHC3w⁺ and 4C-CHC [FLAG-Tetracysteine tag (4C)] were from C. Bazinet and P. Verstreken (Bazinet et al., 1993; Kasprowicz et al., 2008), and UAS-Chc::eGFP. Tub-Egl::eGFP flies were from S. Bullock (Dienstbier et al., 2009; Li et al., 2010). Bic-D^71.34 and Bic-D^IIIE48, the two dominant alleles, the null allele Bic-D^r5 and Bic-D^mom were described previously (Suter et al., 1989; Ran et al., 1994; Swan and Suter, 1996) except that Df(2L)Exel7068 (Exelixis) was used as Bic-D-deficiency. Chc^GF23 (from T. Schüpbach) (Yan et al., 2009), Rab5² (Lu and Bilder, 2005) and Rbsn-5^C241 (from A. Nakamura) (Tanaka and Nakamura, 2008) were used to generate mutant germ line clones with hsFLP; Ubi-GFP(S65T)nls FRT40A, His2Av-GFP, hsFLP, FRT19A/FM7a or Ubi-GFP, hsFLP, FRT19A/FM7. Clones were induced twice per day at 37°C for 2 h for 2 consecutive days. Flies were dissected 2-4 days after hatching. The Chc⁴ chromosome from the Bloomington Drosophila Stock Center was cleaned up by recombination with the multiply marked X chromosome c v m w y sd os, resulting in a lethal-free Chc⁴ chromosome marked with sd. Production of homozygous Chc⁴ females was described previously (Bazinet et al., 1993). pUASP-Chc-V5-K10-attB and pUASP-myc-chc-K10-attB transgenic flies were generated with the phiC31 integrase transgenesis system and the y w;+; attP-64A line as host (Bischof et al., 2007; Koch et al., 2009).

To express UAS-Chc::eGFP in the Bic-D^mom mutant background, the P{matα4-GAL-VP16} driver was recombined with the Bic-D deficiency Df(2L)Exel7068. Flies of the genotype Df(2L)Exel7068, P{matα4-GAL-VP16}/SM1; hs-BicD/TM3Sb were crossed to BicD^r5/SM1; UAS-Chc::eGFP/TM3Sb flies and the progeny of this cross was heat shocked twice per day (37°C, 2 h) from L3 till at least day two of adulthood, before shutting off Bic-D expression. Otherwise, flies were kept at 25°C.

To construct pUASP-myc-Chc-K10-attB and pUASP-Chc-V5-K10-attB, the Chc open reading frame was amplified from LD43101 (BDGP) using primers with XbaI sites for cloning in-frame into pUASP-myc-K10-attB and pMT-V5-HisA (Life Technologies), respectively. From the resulting pMT-Chc-V5 template, the Chc-V5 region was amplified with primers containing NotI and KpnI sites to facilitate cloning into the pUASP-K10-attB vector.

Immunoprecipitations (IPs) were performed according to Vazquez-Pianzola et al. (Vazquez-Pianzola et al., 2011) with embryonic extracts from 0- to 8-hour-old translin^null mutant embryos and from same age embryos laid by Nullo-Gal4/SM1; pUASP-myc-Chc-K10 attB/TM3Sb females. For IPs, 70 pairs of ovaries were used per ovarian IP, whereas seven pairs were collected for one western blot lane. These were boiled for 2 min in 20 µl of SDS sample buffer, vortexed for 15 s, boiled for another 8 min and loaded on a SDS-PAGE gel. Western blots were performed using the mouse monoclonal anti-myc 9E10 [1:5, Developmental Studies Hybridoma Bank (DSHB)], anti-Egl (1:5000) (Mach and Lehmann, 1997), mouse anti-Bic-D antibodies (1B11 plus 4C2, 1:10) (Suter and Steward, 1991), mouse anti-α-Tubulin AA4.3 (1:2000, DSHB), rabbit anti-mammalian Chc (1:500) (Hirst et al., 2009) and rabbit anti-Flag (1:320; Sigma, F7425). Horseradish peroxidase-conjugated secondary antibodies were from GE Healthcare (NA934V and NA931V).

Whole-mount immunostainings were performed with mouse anti-Bic-D (1B11 plus 4C2, 1:10) (Suter and Steward, 1991), rat anti-Chc (1:35) (Wingen et al., 2009) or rabbit anti-Chc (1:200) (Kametaka et al., 2010), rabbit anti-Osk (1:500) (Markussen et al., 1995), rabbit anti-Egl (1:5000) (Mach and Lehmann, 1997), rabbit anti-Staufen (1:2000) (St Johnston et al., 1991), mouse anti-myc 9E10 (1:5, DSHB), mouse anti-Flag (1:200, Sigma, F3165), rabbit anti-Flag (1:200, Sigma, F7425), rabbit anti-Rab11 (1:8000) (Tanaka and Nakamura, 2008), rabbit anti-Rbsn-5 (1:5000) (Tanaka and Nakamura, 2008), rabbit anti-Khc (1:150, Cytoskeleton), mouse anti-α-tubulin DM1A (1:500, Sigma, T9026), rabbit anti-EB1 (1:500) (Rogers et al., 2002) and rabbit anti-Clc (1:500) (Heerssen et al., 2008). Both anti-Flag antibodies stained the oocyte nucleus in wild-type ovaries. To reduce this background cross-reactivity, antibodies were preadsorbed on wild-type embryos. Secondary antibodies were Cy3-conjugated anti-mouse (712-165-150), Cy5-conjugated anti-rabbit (111-175-144), Cy5-conjugated anti-mouse (115-175-146) (Jackson ImmunoResearch) and A488-conjugated goat anti-rabbit (A11008, Molecular Probes). Nuclei and F-actin were stained for 20 min with 2.5 μg/ml Hoechst 33258 and 1 unit/ml rhodamine-conjugated phalloidin (Molecular Probes), respectively, during the final washing steps. Imaging was with a Leica TCS-SP2 confocal microscope. Mouse anti-α-tubulin DM1A staining of ovaries is described by Sato et al. (Sato et al., 2011) and EB1 staining by Sanghavi et al. (Sanghavi et al., 2012).

Digoxigenin-labeled RNA probes were synthesized from linearized pBS-oskar, ESTs LD32255 and LD43101 (BDGP), pBS-OrbE4 cDNA (Lantz et al., 1992) and pBS-Bic-D. In situ hybridizations experiments (Vazquez-Pianzola et al., 2011) utilized 5% milk for blocking and as antibody dilution buffer. For dual detection of mRNA and GFP signal, the proteinase K step was omitted. Rabbit anti-GFP (1:300, AMS Biotechnology, 210-PS-1GFP) was incubated overnight together with sheep anti-digoxigenin antibodies (Roche, 1 333 089). Donkey anti-sheep A647 (1:200, Jackson ImmunoResearch, 713-606-147) and donkey anti-rabbit A568 (1:1500, Invitrogen, A10042) secondary antibodies were incubated overnight.

RNA from five to ten ovaries was extracted using the RNeasy Kit (Qiagen), DNaseI digestion was performed according to the manual. RNA IPs were performed essentially as described (Easow et al., 2007) with the following modifications. Blocking buffer was supplemented with 1 mg/ml bovine serum albumin and protease inhibitors (Roche). The following were added to the hypotonic buffer: 1 mM DTT, 0.5% Tween-20, protease and RNase inhibitors, 100 U/ml (Biolabs). The wash buffer contained 200 mM KCl. For each IP, 40 μl of protein G sepharose beads (Roche) and 1 ml of monoclonal anti-GFP were used (a gift from A. Marcil). Coated beads were washed with non-hypotonic buffer (hypotonic buffer adjusted to 150 mM KCl and 20% glycerol). For extract preparation, 1 g of 0- to 8-hour-old embryos were homogenized in 1 ml hypotonic buffer. The extract was cleared at 10,000× g for 20 min. For each IP, 600 μl of cleared supernatant was adjusted to 20% glycerol and 150 mM KCl, and supplemented with 2 μl of RNase-free DNase I (20 U/ml, Roche). IP was performed overnight. The immunoprecipitate was washed eight times. Beads were Proteinase K treated, RNA extracted using Trizol (Invitrogen) and resuspended in 25 μl DEPC-treated water, ready for RT-qPCR analysis.

cDNAs were synthesized with the SuperScript II Reverse Transcriptase and oligo dT primers. Primers used for the RT-qPCR analysis amplified fragments between 100 and 150 bp and one primer of a pair spanned an exon junction. The QuantiTect SYBR Green PCR Kit (Qiagen) and a Rotor Gene (Qiagen) were used, and the Pfaffl method was applied to normalize to Tubulin67C mRNA levels and to calculate fold changes (Pfaffl, 2001).

For FM4-64 (Molecular Probes) incorporation assays, see Sommer et al. (Sommer et al., 2005). For live imaging of egg chambers, five to seven females were raised with five to seven males on freshly prepared fly food supplemented with fresh yeast paste. For BicD^mom mutants, fly food was changed every day for 5 days. Ovaries were extracted into modified Schneider's medium (Prasad et al., 2007) and transferred into a drop of 10S Voltalef oil. Ovarioles were stretched by dragging their anterior end from the medium into the oil, immobilizing the ovarioles at the medium-oil interface. Pictures were acquired on a Till Photonics iMIC spinning disc (60× lens, Andor 897 EMCCD) every 500 ms for 200 cycles. Linearly moving particles were selected by hand and particle velocity was calculated using ImageJ.

We thank S. Bullock, C. Bazinet, P. Verstreken, P. Behr, A. M. Nakamura, R. Lehmann., J. Hirst, S. Robinson, S. Waguri, S. L. Rogers, A. Marcil, D. St Johnston, T. Schüpbach, D. Bilder and D. Graeme for antibodies, constructs and fly stocks. We also thank G. Hernandez and S. Bullock for providing most constructive input for this manuscript. Special thanks go to D. Beuchle and B. Schaller for their excellent technical help.