1. Cell Biology
Download icon

Dynein-mediated transport and membrane trafficking control PAR3 polarised distribution

  1. Julie Jouette
  2. Antoine Guichet  Is a corresponding author
  3. Sandra B Claret  Is a corresponding author
  1. Institut Jacques Monod, CNRS, UMR 7592, Paris Diderot University, Sorbonne Paris Cité, France
Research Article
  • Cited 0
  • Views 934
  • Annotations
Cite this article as: eLife 2019;8:e40212 doi: 10.7554/eLife.40212

Abstract

The scaffold protein PAR3 and the kinase PAR1 are essential proteins that control cell polarity. Their precise opposite localisations define plasma membrane domains with specific functions. PAR3 and PAR1 are mutually inhibited by direct or indirect phosphorylations, but their fates once phosphorylated are poorly known. Through precise spatiotemporal quantification of PAR3 localisation in the Drosophila oocyte, we identify several mechanisms responsible for its anterior cortex accumulation and its posterior exclusion. We show that PAR3 posterior plasma membrane exclusion depends on PAR1 and an endocytic mechanism relying on RAB5 and PI(4,5)P2. In a second phase, microtubules and the dynein motor, in connection with vesicular trafficking involving RAB11 and IKK-related kinase, IKKε, are required for PAR3 transport towards the anterior cortex. Altogether, our results point to a connection between membrane trafficking and dynein-mediated transport to sustain PAR3 asymmetry.

https://doi.org/10.7554/eLife.40212.001

Introduction

Cell polarity is a fundamental process involved in diverse processes crucial for cell function and development. Establishment and maintenance of cell polarity are under the control of a set of polarity proteins highly conserved through metazoans. This network of proteins, some of which are PAR proteins, is set up through regulations involving recruitment or repulsion(Assémat et al., 2008; Goldstein and Macara, 2007; Laprise and Tepass, 2011; Nelson, 2003; Rodriguez-Boulan and Macara, 2014; St Johnston and Ahringer, 2010). The precise localisation of these polarity complexes defines plasma membrane domains with specific functions (Rodriguez-Boulan and Macara, 2014; St Johnston and Ahringer, 2010). Accordingly, the polarity complexes are essential for establishment of neural growth cone polarity (Yu et al., 2006), apico-basal epithelial polarity (Kuchinke et al., 1998; Nance et al., 2003), or antero-posterior polarity in Drosophila oocytes (Cox et al., 2001b; Tomancak et al., 2000) and C. elegans embryos (Kemphues, 2000).

Two major polarity modules control establishment and maintenance of cell polarity: one module composed of PAR3 (also named Bazooka [BAZ] in Drosophila), PAR6, and aPKC proteins, and a second module composed of PAR1, LGL, and SLMB proteins (Goldstein and Macara, 2007; Rodriguez-Boulan and Macara, 2014; St Johnston and Ahringer, 2010). Localisation of these two modules is mutually exclusive and mainly involves interplay between physical interactions and phosphorylations (Coopman and Djiane, 2016). aPKC phosphorylates PAR1 to exclude it from the PAR3 cortical domain. Conversely, PAR3 is phosphorylated by the kinase PAR1, recognised by the 14.3.3 proteins and excluded from the PAR1 cortical domain (Benton and St Johnston, 2003; Morais-de-Sá et al., 2010). PAR3 is also maintained at the plasma membrane through physical interactions with membrane lipids, phosphoinositides (Krahn et al., 2010a). We reported in particular that PI(4,5)P2 (phosphatidyl inositol 4,5 diphosphate) controls PAR3 apical targeting in epithelial cells (Claret et al., 2014). In addition, the cytoskeleton seems to be important for proper localisation of PAR3, and it is noteworthy that the apical actin network is required to position PAR3 apically as well as microtubules (MTs) via dynein-dependent transport in Drosophila embryos (Harris and Peifer, 2005; McKinley and Harris, 2012).

In Drosophila oocytes, PAR3, at the anterior cortical domain, and PAR1, at the posterior, specify the polarity axes by controlling the MT organisation (Cox et al., 2001a; Doerflinger et al., 2003), and thus the localisation of determinants such as bicoid, oskar, and gurken mRNAs, crucial for subsequent future embryo development (St Johnston, 2005). Generating mutually exclusive cortical domains is especially important for polarisation of this large single-cell system. However, throughout oogenesis, the asymmetric localisation of PAR proteins is dynamic and PAR1 and PAR3 domains are not always mutually exclusive. During early oogenesis, localisation of PAR3 and PAR1 is independent (Huynh et al., 2001), while at mid-oogenesis (stage 7–8), PAR3 and PAR1 overlap at the posterior plasma membrane. Both show mutually exclusive localisations at stage 9 (Doerflinger et al., 2010).

During stages 8–10, which correspond to the critical period for localisation of the polarity axes determinants (Steinhauer and Kalderon, 2006), the oocyte undergoes a rapid threefold size increase while subcellular localisation of PAR modules has to remain strictly conserved. Although the mutual antagonism between PAR3/aPKC/PAR6 and PAR1/LGL/SLMB modules plays an important role in setting up PAR3 restriction (Morais-de-Sá et al., 2014; Tian and Deng, 2008), the molecular mechanisms involved in these processes remain elusive and may not be sufficient to sustain PAR3 asymmetry. Indeed, computer modelling has highlighted that mutual antagonism between the two complexes is not enough to maintain asymmetry between the polarity determinants (Fletcher et al., 2012). Moreover, the fate of PAR3, once phosphorylated by PAR1, and how it is redirected from the posterior to the anterior cortex is unknown. PAR3 could diffuse laterally in contact with plasma membrane until it reaches the anterior, or it could be dispersed from the posterior membrane in the cytoplasm as has been suggested for the Drosophila embryo (McKinley and Harris, 2012) and then recycled. Our previous findings have highlighted the role of the PIP5Kinase Skittles (SKTL) and its product PI(4,5)P2 in PAR3 localisation in oocytes and epithelial cells (Claret et al., 2014; Gervais et al., 2008). PI(4,5)P2, among other functions (Tan et al., 2015; Zimmermann et al., 2005), is crucial to recruit the endocytic machinery at the plasma membrane during the first step of endocytosis (Compagnon et al., 2009; Posor et al., 2015). We hypothesise that SKTL, by controlling endocytosis and/or vesicular trafficking could regulate indirectly PAR3 localisation.

Here, with use of quantitative analysis, we examined the precise evolution of PAR3 distribution during stages 8–10 of oogenesis. We show that PAR3 is excluded from the posterior cortex much later than establishment of PAR1 to the posterior and that actin cytoskeleton and endocytosis are important for this process. Subsequently, MTs and dynein motor are required for PAR3 transport to the anterior plasma membrane. This transport is connected with vesicular trafficking, cytoplasmic PAR3 being associated with PI(4,5)P2-enriched endosomes. We provide evidence of physical interactions between PAR3, SKTL, and the dynein light intermediate chain DLIC, which could explain transport of PAR3 by dynein directly on vesicles. Finally, we found that IKK-related kinase, IKKε in the correct localisation of PAR3. Knockdown of IKKε leads to an accumulation of PAR3 near the minus ends of MTs. Altogether our results point to a connection between membrane trafficking and dynein-mediated transport to sustain PAR3 asymmetry in the Drosophila oocyte.

Results

Fine tuning of PAR3 distribution along the anterior posterior axis

To characterise the evolution of PAR3 distribution during oocyte development, we developed a quantification method to monitor the precise variation of PAR3 distribution along the plasma membrane during late oogenesis (Figure 1A). The oocyte plasma membrane was therefore subdivided into three regions: the anterior plasma membrane (APM), which corresponds to the membrane in contact with the nurse cells, the posterior plasma membrane (PPM), which corresponds to the posterior domain of the oocyte where PAR1 and Staufen are localised (Figure 1—figure supplement 3A–B), and the lateral plasma membrane (LPM), which corresponds to the plasma membrane between the two previous regions (Figure 1A). The APM domain corresponds to juxtaposition of oocyte and nurse cell plasma membranes. Therefore, for each oocyte individually, we quantified the mean fluorescent intensity of oocyte adjacent nurse cells. We then removed, from the anterior signal, the mean fluorescent intensity of the nurse cell plasma membrane to precisely quantify the signal coming only from the APM of this oocyte. Afterwards, the intensity profile along the plasma membranes, as well as the global intensity by plasma membrane and cytoplasm domains, were compiled and the signal density was analysed (see Materials and methods for details).

Figure 1 with 3 supplements see all
Dynamic PAR3 distribution along the oocyte anterior posterior axis.

(A) Description of the quantification Fiji Macro. After selecting three points in the oocyte (yellow stars) and delimitation of oocyte perimeter, the macro allows us to obtain different data about protein repartition in oocytes: the intensity profile of the plasma membrane (magenta); the mean fluorescent intensity and the length of each of the plasma membrane domainsare automatically generated (anterior/APM in red; lateral/LPM in green; posterior/PPM in blue) and the mean signal intensity inside the cytoplasm (cyan). (B–E) Distribution of PAR3 between stages 8 and 10 in representative examples. (B) Localisation of PAR3-GFP, expressed in the germline under control of the maternal driver Tub67c-GAL4, from stage 8 to stage 10. The brackets indicate oocytes. (C) Representative intensity profiles of plasma membrane distribution in a single oocyte (APM in red, PPM in blue). Green arrows highlight PAR3 posterior accumulation, and red arrows PAR3 posterior exclusion. (D) Raw quantity of PAR3 in each plasma membrane domain from stage 8 to stage 10 (APM in red; LPM in green; PPM in blue). (E) To avoid size/expression fluctuations of oocytes between the different mutant genotypes, PAR3 distribution has been normalised by the length of the membrane and the total oocyte signal (density/total). In this case, we cannot compare the level between the different stages but only the asymmetrical distribution of PAR3 between the different domains. (F) Evolution of PAR3 asymmetry from stage 8 to stage 10. The asymmetry ratio (APM/PPM of PAR3 density) highlights the increase of PAR3 polarity in oocytes from stage 8 to stage 10. Stage 8, n = 8; stage 9A, n = 9; stage 9B, n = 15; stage 10, n = 14. Mann-Whitney test, NS: not significant; **: p<0.01 ; ***: p<0.001; ****: p<0.0001. Error bars indicate SEM. The scale bars represent 30 µm in (B) and in all following figures.

https://doi.org/10.7554/eLife.40212.002

To identify and to characterise PAR3 regulation mechanisms, we first used a Drosophila strain expressing a Bac-encoded PAR3-GFP under its endogenous regulatory regions, in the absence of endogenous PAR3 (Figure 1—figure supplement 1A, (Besson et al., 2015)). However, accumulation of PAR3 at the apex of somatic follicular cells around the oocyte (Figure 1—figure supplement 1B) masks the potential localisation of PAR3 at the plasma membrane of oocyte, with the two membranes being very close. To circumvent this difficulty, we chose to follow PAR3 with a GFP tag, expressed only in the oocyte and its associated nurse cells (Benton and St Johnston, 2003). In this context, as previously reported (Doerflinger et al., 2010), PAR3 mostly accumulates at the anterior plasma membrane (Figure 1B–C). With our quantitative tool, we measured PAR3 signals coming from the different plasma membrane subdomains during late oogenesis (Figure 1D). We observe a clear increase of PAR3 quantity at the APM (in red) from stage 8 to stage 10, and in parallel an exclusion of PAR3 on the PPM (in blue). Thereafter, to avoid fluctuations associated with membrane growth during oogenesis or with experimental procedures, we normalised the raw data to the total amount of PAR3 signal in each oocyte and related to the membrane length (cf. materiel and methods, and Figure 1—figure supplement 2). With these density results, although we cannot compare the quantities between different stages, we can follow the evolution of asymmetry in a stage. As expected, PAR3 density is highest at the oocyte APM (Figure 1E). However, PAR3 density repartition presents a striking dynamic in both PPM and LPM. At stage 8 and early stage 9 (stage 9A, Figure 1—figure supplement 1C), PAR3 is denser at the PPM than at the LPM (Figure 1C, green arrows). Then at late stage 9 (stage 9B, Figure 1—figure supplement 1C), PAR3 is progressively excluded from the PPM (Figure 1C, red arrows). These changes in PAR3 distribution reflect establishment of two distinct plasma membrane domains and are highlighted by the asymmetric ratio of PAR3 measured by the anterior to posterior density ratio (Figure 1F). The beginning of PAR3 posterior exclusion in the middle of stage 9 correlates with the localisation switch of other factors such as Staufen (Figure 1—figure supplement 3B) toward the posterior pole of the oocyte and correlated to the MT network reorganisation (Januschke et al., 2006). This result is surprising as PAR1 is assumed to exclude PAR3 from the plasma membrane, yet PAR1 is already localised at the posterior pole from at least stage 7, long before PAR3 exclusion (Doerflinger et al., 2010). This may indicate that other processes participate in the disappearance of PAR3 from the PPM and the LPM. As the LPM appeared to follow the PPM comportment, thereafter we focused on APM accumulation and on PPM exclusion, two mechanisms important for establishment of antero-posterior polarity.

Posterior exclusion and anterior accumulation are two PAR3 localisation processes that can be decoupled/separated

To understand how PAR3 is excluded from the posterior domain and enriched at the anterior domain in the oocyte, we investigated further cytoskeleton involvement in this process. With latrunculin drug treatment, PAR3 is still predominantly at the APM as under control conditions (Figure 2A and C, Figure 2—figure supplement 1). However, its posterior exclusion, which normally occurs at stage 9B, is not observed (Figure 2B), indicating that the actin network is required for PAR3 posterior exclusion. We next addressed the potential MT requirement for PAR3 antero/posterior distribution in the oocyte. In the presence of colchicin, a drug that depolymerises MT, PAR3-polarised distribution along the antero/posterior axis is lost and tends toward isotropy (Figure 2A and C, Figure 2—figure supplement 1). Compared to control, PAR3 APM localisation is strongly reduced (Figure 2C). However, PAR3 is still excluded from the PPM although to a lesser extent than the control (Figure 2A–B). This confirms the importance of MTs in PAR3-polarised localisation, except at the posterior region where exclusion is still present. Hence it appears that in the oocyte, posterior exclusion and anterior accumulation can be uncoupled. We also note that both MT and actin disassembly increase cytoplasmic accumulation of PAR3 in dotted structures (Figure 2D). We verified that these dotted structures are also detected in the absence of PAR3 overexpression with a PAR3-Protein trap strain (Januschke and Gonzalez, 2010). In this condition, we can observe some dotted structures in the oocyte cytoplasm (Figure 2—figure supplement 2A). Furthermore, when microtubules are depolymerised by colchicin, PAR3 accumulates in numerous dotted structures in the cytoplasm (Figure 2—figure supplement 2B–D).

Figure 2 with 2 supplements see all
Cytoskeleton involvement in PAR3 polarity.

(A–D) Role of cytoskeleton on PAR3 distribution. Flies are nourished with latrunculin (cyan) for 48 h, colchicin (yellow) for 24 h, or only yeast paste (control). (A) Representative distribution of PAR3-GFP in stage 9B oocytes. Note the PPM exclusion of PAR3 (arrow) or the strong APM accumulation (arrowhead). (B) The PAR3 posterior exclusion ratio (ratio LPM/PPM density) is represented between stage 8 and stage 10 oocytes. Under the value of 1, there is a posterior accumulation of PAR3, and above 1 a posterior exclusion. (C) Antero-posterior asymmetry (ratio APM/PPM density at stage 9B) is strongly affected by colchicin, but not by latrunculin. (D) The two drugs lead to an increase in the cytoplasmic fraction of PAR3 in stage 9B oocytes. Mann-Whitney test, NS: not significant; *p<0.05 ; ***p<0.001 ; ****p<0.0001. Error bars indicate SEM. Control (stage 8, n = 8; stage 9A, n = 9; stage 9B, n = 15; stage 10, n = 14);+latrunculin (stage 8, n = 5; stage 9A, n = 12; stage 9B, n = 11; stage 10, n = 9);+colchicin (stage 8, n = 8; stage 9A, n = 9; stage 9B, n = 11; stage 10, n = 6).

https://doi.org/10.7554/eLife.40212.008

By producing PI(4,5)P2, SKTL controls anterior accumulation and posterior exclusion of PAR3

PAR3 is a cytoplasmic protein that can interact with membranes by direct interaction with phospholipids, in particular PI(4,5)P2 (Claret et al., 2014; McKinley et al., 2012; Wu et al., 2007), and/or by interaction with membrane-associated proteins such as those of the adherens junction (Coopman and Djiane, 2016).

We have previously shown that the PIP5Kinase SKTL, by providing the phosphoinositide PI(4,5)P2, is crucial to maintaining PAR3 at the adherens junctions in epithelial cells (Claret et al., 2014). In oocytes, PAR3 is associated with the plasma membrane (Gervais et al., 2008) but also forms some particles in the cytoplasm. These particles are also associated in part with PI(4,5)P2-containing membranes (Figure 3G) and the lipid kinase SKTL, which produces PI(4,5)P2 (Figure 3H). By immunostaining and by colocalisation using Mander’s overlap coefficient, we identified that half of PAR3 vesicles (50.05% ± 0.08 SEM, n = 8) are PIP(4,5)P2 -positive (Figure 3G).

By producing PI(4,5)P2, SKTL controls PAR3 APM accumulation and PPM exclusion.

(A–F) Distribution of PAR3 in response to SKTL activity. PAR3-GFP is expressed in germinal cells at stage 9B under control conditions, or with overexpression (OE) of Myc-SKTL, Myc-SKTLDNRQ, or in context sktl2.3/sktl∆5. (A–C) Representative distribution of PAR3-GFP in oocytes in these different genetic contexts. (D) Quantification of PAR3 density at each plasma membrane domain in sktl mutant or SKTL overexpressed (OE) contexts in stage 9B oocytes. Error bars indicate SEM. (E) Antero-posterior asymmetry of PAR3 (ratio APM/PPM density) at stage 9B in control or sktl mutant. (F) Quantification of PAR3 posterior exclusion ratio in sktl mutant or SKTL overexpressed contexts at stage 9B. For (D–F): control (stage 9B, n = 10); sktl2.3/sktl∆5 (stage 9B, n = 8); Myc-SKTL (stage 9B, n = 8); Myc-SKTLDNRQ (stage 9B, n = 10). Mann-Whitney test, NS: not significant; *p<0.05 ; **p<0.01 ; ***p<0.001 ; ****p<0.0001. (G) PAR3-GFP (green, G’) expressed in the germline is present occasionally on vesicles containing PI(4,5)P2 visualised with PHPLC RFP (magenta). Arrowheads show the vesicles that are associated with PAR3 and PI(4,5)P2. G’, G’’ and G’’’ are magnifications of G (white frame). (H) Colocalisation of PAR3 (green, H’), PI(4,5)P2 visualised with PHPLC RFP (red, H’’’), and SKTL visualised with myc tag (white, H’’). Note that coexpression of PAR3 and SKTL increase the cytoplasmic dotted localisation of PAR3. Scale bars indicate 30 µm.

https://doi.org/10.7554/eLife.40212.016

We then investigated SKTL requirements on PAR3 polarised distribution along the oocyte anterior posterior axis focusing both on the amount of SKTL and on its kinase activity. In the absence of SKTL (sktl2.3/sktl∆5), PAR3 is accumulates more at the PPM than at the APM (Figure 3A, D and E). Conversely, when SKTL is overexpressed (OE), PAR3 is more excluded from the PPM than under control conditions (Figure 3B, D and E). Thus, SKTL seems to have a preponderant function to exclude PAR3 from the PPM and to increase PAR3 density at the APM. To monitor whether SKTL kinase activity is required for this process, we performed the same experiment with a kinase dead form, SKTLDNRQ whose mutation in mammalian homolog leads to a dominant negative effect (Coppolino et al., 2002). In such a case, SKTLDNRQ overexpression does not enhance PAR3 exclusion from the PPM (Figure 3C, D and E). PAR3 distribution along the anterior posterior axis tends to be isotropic (Figure 3C and D). We can conclude that SKTL kinase activity, hence the production of PI(4,5)P2, is essential to regulate PAR3 distribution.

SKTL can bypass PAR1-dependent posterior exclusion

Upon phosphorylation, PAR1 excludes PAR3 from the posterior domain of the oocyte (Benton and St Johnston, 2003). Accordingly, with our quantification method, we found that PAR3 was not excluded from the PPM upon PAR1 knockdown by RNAi (Figure 4B and D and Figure 4—figure supplement 1). Likewise, a PAR3-AA mutant form, non phosphorylable by PAR1 (Benton and St Johnston, 2003), is not excluded from the PPM (Figure 4C–D). Thus, PAR1 through its role in PAR3 phosphorylation is required to exclude PAR3 from the PPM. As SKTL also controls the posterior exclusion of PAR3, we combined the knockdown of PAR1, which normally presents no PAR3 exclusion, with the overexpression of SKTL, which increases the exclusion. In this case, we observed that SKTL can still exclude PAR3 even in the absence of PAR1 (Figure 4E). We then confirmed this observation using the PAR3-AA mutant form (Figure 4E). In this case too, SKTL overexpression can still exclude PAR3 from the posterior membrane. These results suggest that SKTL leads to PAR3 exclusion independently of PAR3 phosphorylation by PAR1.

Figure 4 with 1 supplement see all
SKTL-dependent PAR3 posterior exclusion bypasses regulation by PAR1.

(A–C) Representative distribution of PAR3-GFP in stage 9B oocyte in a control situation (A), in RNAi PAR1 context (B), or when PAR3 phosphorylation sites by PAR1 are mutated (C). In the control genotype, PAR3 is excluded from PPM (arrow), unlike the other two genotypes. N indicates the oocyte nucleus position. The scale bars represent 30 µm. (D) PAR3 posterior exclusion in response to PAR1 at stage 9B. In germinal cells, PAR3AA-GFP, a mutant form, non phosphorylable by PAR1 or PAR3-GFP is expressed with nothing, with par1, or with mCherry knock-down contexts. The posterior exclusion ratios in stage 9B oocytes are represented. PAR3 (stage 9B, n = 15); PAR3-AA (stage 9B, n = 6); PAR3 RNAi mCherry (stage 9B, n = 10); PAR3 RNAi PAR1 (stage 9B, n = 10). (E) SKTL effect on PAR3 posterior exclusion is observed in combination with a PAR1 activity decrease (in green) or with the PAR3-AA non phosphorylable form (in red). PAR3, RNAi PAR1, RNAi mCherry (stage 9B, n = 10); PAR3, RNAi PAR1, mycSKTL (stage 9B, n = 11) PAR3 RNAi PAR1 (stage 9B, n = 10); PAR3, RNAi mCherry, mycSKTL (stage 9B, n = 10); PAR3-AA, RNAi mCherry (stage 9B, n = 10); PAR3-AA, mycSKTL (stage 9B, n = 10). Mann-Whitney test, NS: not significant; *p<0.05 ; **p<0.01 ; ***p<0.001 ; ****p<0.0001.

https://doi.org/10.7554/eLife.40212.020

RAB5-dependent endocytosis is important for PAR3 PPM exclusion

To understand how SKTL could induce the PPM exclusion of PAR3, we investigate the implication of endocytosis in this process. PI(4,5)P2 has numerous functions in the cell, one of which is a role in the first steps of endocytosis (for review Posor et al., 2015). In Drosophila oocyte, PIP(4,5)P2 is required for endocytic-vesicle formation and the small GTPase RAB5 is required for maturation of these early endocytic vesicles (Compagnon et al., 2009).

As we previously described, in addition to a plasma membrane localisation, PAR3 is also detected in dotted structures. By immunostaining and by colocalisation using Mander’s overlap coefficient, we identified that half of PAR3 vesicles (50.2% ± 0.05 SEM, n = 10) are RAB5-positive, an early endosome marker (Figure 5A). Moreover, when RAB5 activity is impaired by expression of a dominant negative form, RAB5S43N (Entchev and González-Gaitán, 2002), as revealed by lipophilic dye uptake, the endocytosis is strongly reduced at the posterior of the oocyte (Figure 5—figure supplement 1A–B) and PAR3 distribution is affected (Figure 5D and E). The same result is observed using a RAB5 knockdown (Figure 5C and E, RAB5 RNAi). PAR3 accumulation at the APM is lost and its density significantly increases at the PPM, leading to loss of its PPM exclusion (Figure 5E and F). Thus, proper PAR3 distribution along the antero-posterior axis appears to rely on endocytosis being removed from the PPM and being enriched at the APM.

Figure 5 with 1 supplement see all
PAR3 asymmetry depends on RAB5.

(A) PAR3-GFP (A’), (A’’’ green) expressed in germline is present occasionally in RAB5-positive early endosomes (A’’), (A’’’) magenta). A’, A’’ and A’’’ are magnifications of A (white frame). Arrowheads show the vesicles that are associated with PAR3 and RAB5. (B–F) PAR3 distribution in response to RAB5 activity impairment. PAR3-GFP is expressed in germinal cells at stage 9B in RAB5 RNAi, RAB5DN(S43N) or in mCherry knockdown (control) contexts. (B–D) Representative images of PAR3 distribution under control conditions (B), in RAB5 RNAi (C), or in RAB5DN(S43N) (D). Scale bars indicate 30 µm. (E) Quantification of PAR3 density at each plasma membrane domain. Error bars indicate SEM. (F) Quantification of PAR3 posterior exclusion in different RAB5 mutant contexts. Control (stage 9B, n = 10); RAB5 RNAi (stage 9B, n = 12); RAB5DN(S43N) (stage 9B, n = 6). Mann-Whitney test, ns: not significant; *p<0.05 ; **p<0.01.

https://doi.org/10.7554/eLife.40212.024

RAB11-dependent recycling pathway is required for PAR3 APM enrichment

To determine the nature of all the PAR3 dotted structures that can be detected in the wild type context, we immunostained different compartments of the cell and measured their colocalisation with PAR3 (Figure 6A).

Figure 6 with 2 supplements see all
Role of RAB11 on PAR3 asymmetrical localisation.

(A) Colocalisation of PAR3 (PAR3-GFP maternally expressed) with vesicular trafficking markers. To quantify the colocalisation, we measured the Pearson’s coefficient on stage 9B oocytes and results are presented in the box plot. As a control, we used the value of colocalisation obtained with the same images but after rotation of one by 90 degrees. The stars represent the p-value with the control. For HRS, KDEL, and SYX16: n = 7; for PHPLC: n = 8; for RAB5: n = 10; for RAB11: n = 13. (B) PAR3-GFP (B’), green) expressed in germline is present occasionally in RAB11-positive recycling endosomes (B’’), (B’’’) magenta). B’, B’’, and B’’’ are magnifications of B (white frame). Arrowheads show the vesicles associated with PAR3 and RAB11. The arrow points to the vesicle associated only with PAR3. (C,D) RAB11 effect on asymmetrical distribution of PAR3. The antero-posterior asymmetry (C) and the posterior exclusion (D) of PAR3 at stage 9B have been evaluated in rab11P2148 germline clones. Control (stage 9B, n = 10); rab11P2148 (stage 9B, n = 10). Mann-Whitney test, NS: not significant; *p<0.05; ***p<0.001 ; ****p<0.0001. Error bars indicate SEM. (E,F) Representative images of PAR3 distribution in rab11P2148 clones. The rab11P2148 mutant cells are indicated by the absence of nuclear RFP staining (nls-RFP). (E) The heterozygote egg chamber (rab11P2148/+) is used as a control. (F) rab11P2148 mutant oocyte. Note that PAR3 is always excluded from PPM (arrow). Scale bars indicate 30 µm.

https://doi.org/10.7554/eLife.40212.028

PAR3-associated vesicles are enriched in PI(4,5)P2, and colocalise with the RAB5-positive compartment (early endosome), but also the RAB11-positive compartment (26.8% (±0.06 SEM, n = 13) of cytoplasmic PAR3 colocalises with RAB11-recycling endosomes Figure 6A and B). There is no colocalisation with endoplasmic reticulum (KDEL), Golgi compartment (SYX16), or late endosome (HRS) markers (Figure 6A and Figure 6—figure supplement 1). Taken together, these results indicate that PAR3 can be associated with RAB5 and RAB11 endosomes. However, endosomes could be a mosaic of several RAB domains present in continuity on the same membrane vesicle (Sönnichsen et al., 2000; Wandinger-Ness and Zerial, 2014). Here we do not know if PAR3 is associated with a unique endosome supporting RAB5 and RAB11 or with a specific compartment containing only one RAB.

To identify whether RAB11 is involved in asymmetric distribution of PAR3, we knocked down RAB11 in the oocyte through germline clones with rab11p2148 allele, which precludes oocyte development (Jankovics et al., 2001) and monitors PAR3 distribution (Figure 6F). There was no significant difference on PAR3 PPM exclusion between control (Figure 6D–E) and rab11p2148 mutant clones (Figure 6D and F). However, we noted a RAB11 effect on PAR3 APM enrichment (Figure 6C). Thus RAB11 seems to be more important for APM accumulation than for the PPM exclusion.

Finally, we monitored whether PAR3, which comes from the nurse cells through the ring canals, is also connected to membrane traffic. We noted that PAR3 accumulates in the vicinity of the ring canals and that it is associated with PI(4,5)P2 membrane and with RAB11 endosomes (Figure 6—figure supplement 2). This suggests that neo-synthesised PAR3 is also transported in the oocyte in association with endosomes.

DLIC, PAR3, and SKTL are interacting partners

To uncover new interacting partners for SKTL upon polarity establishment of PAR3, we performed a proteomic screen by immunoprecipitation and then mass spectrometry analysis. We identified, after immunoprecipitation of SKTL-GFP, a peptide of the dynein light intermediate chain (DLIC), which is absent from the control. DLIC is an essential subunit of the MT-based dynein motor (Reck-Peterson et al., 2018). This interaction between SKTL and DLIC was confirmed by western blot with anti-DLIC antibody (Figure 7E).

Figure 7 with 2 supplements see all
Dynein regulates PAR3 asymmetry.

(A–C) Distribution of PAR3 in response to dynein activity decrease. PAR3-GFP is expressed in germinal cells at stage 9B in control (osk-GAL4; UASp PAR3-GFP) or dhc64 knockdown contexts (osk-GAL4; UASp RNAi dhc64; UASp PAR3-GFP). (A) Representative distribution of PAR3 in oocyte (note the dotted accumulation of PAR3 under the plasma membrane in the insert). The scale bars represent 30 µm. (B) Quantification of PAR3 density at each plasma membrane domain. (C) Quantification of PAR3 posterior exclusion (ratio LPM/PPM) in dynein mutant context. For (A–C): control (stage 9B, n = 10); RNAi dhc64c (stage 9B, n = 10). For (B, C), Mann-Whitney tests; NS: not significant; *p<0.05 ; **p<0.01; ***p<0.001; ****p<0.0001. Error bars indicate SEM. (D) Co-immunoprecipitation (IP) of PAR3 by DLIC-GFP in ovarian extracts. Ovaries were dissected from Pubi-DLIC-GFP flies or control Pubi-PDI-GFP flies. The ovarian extract was incubated on magnetic beads coupled with antibody anti-GFP. PAR3 was revealed by anti-PAR3 antibody. (E) SKTL interacts with both PAR3 and DLIC. CoIP of PAR3-HA or DLIC with GFP-SKTL in S2 cell extracts. Cells were transfected with GFP-SKTL or GFP-SNAP (as control) and with PAR3-HA. The cell extracts were incubated on magnetic beads with anti-GFP antibody. PAR3 was revealed by anti-HA antibody and DLIC by anti-DLIC antibody.

https://doi.org/10.7554/eLife.40212.034

Interestingly, LIC2, DLIC mammalian homologue, interacts with PAR3 through N-terminal dimerisation and PDZ1 domains of PAR3 (Schmoranzer et al., 2009). Thus, we investigated whether this interaction is conserved in Drosophila. We immunoprecipitated DLIC-GFP in ovarian extract and identified a weak association with endogenous PAR3 (Figure 7D). Interestingly, it seems that we coimmunoprecipitate predominantly a high molecular weight PAR3 form (Figure 7—figure supplement 2). This weight could correspond to an oligomeric form of PAR3 (Kullmann and Krahn, 2018).

As DLIC interacts with PAR3 and SKTL, we wondered whether an interaction between PAR3 and SKTL also occurs. Through immunoprecipitation, we found a physical interaction between PAR3 and SKTL (Figure 7E). Hence, as described for other PI(4,5)P2 effectors (Choi et al., 2015), PAR3 seems to form a tripartite complex with PI(4,5)P2 and SKTL, the enzyme that produces it (Figure 3H).

Dynein regulates transport of PAR3 to the anterior plasma membrane but has only a modest impact on posterior exclusion

We have shown that MTs are important to asymmetric localisation of PAR3 and that PAR3 interacts with DLIC. In oocytes (stage 8–10), the MT network forms a gradient following the antero/posterior axis (Januschke et al., 2006). As MTs are more abundant and more nucleated (minus ends) at the anterior pole of oocytes (Khuc Trong et al., 2015; Parton et al., 2011), dynein, as a minus end directed motor, could be necessary for PAR3 accumulation at the APM. To investigate this possibility, we therefore knocked down dynein by expression of ShRNAs directed against the dynein heavy chain isoform (DHC64c) in the oocyte (Sanghavi et al., 2013). Under these conditions, PAR3 accumulation is lost at the APM and its level significantly increases at the LPM (Figure 7A–B). However, PAR3 is still excluded from PPM (Figure 7C), consistent with the weak MT requirement for this process (Figure 2B). Taken together, these data indicate that PAR3 could be transported in complex with the dynein motor on MTs toward the APM.

We noted that, upon Dhc64C inactivation, PAR3 is not evenly distributed along the plasma membrane but is retained in the dotted structures as shown in Figure 7A and Figure 7—figure supplement 1A. PAR3 is still associated with PI(4,5)P2 vesicles when MTs are impaired in the presence of colchicin (Figure 7—figure supplement 1B). Thus, PAR3 transport to the APM depends on MTs, but the association of PAR3 with vesicles does not.

Connection between posterior exclusion and anterior accumulation of PAR3

Our results indicate that PAR3 oocyte distribution relies on two separate processes: an exclusion from the posterior plasma membrane and an accumulation at the anterior plasma membrane. Thus, we investigated whether the two processes were connected. We monitored PAR3 dynamics in the oocyte through fluorescence recovery after photobleaching (FRAP) of PAR3-GFP. As a substantial proportion of the PAR3 that accumulates at the anterior part of the oocyte could come from synthesis and subsequent transport from the nurse cells through the ring canals (Figure 6—figure supplement 2 and Doerflinger et al., 2010), we wanted to dissociate this PAR3 arrival from the potential relocalisation of PAR3 in the oocyte from the posterior. To do so, we chose to photobleach PAR3-GFP localised in the nurse cells and at the APM (Figure 8A), at stage 9A (Video 1). We then followed evolution of PAR3-GFP fluorescence both at the APM and at the PPM (Figure 8A). In this way, we monitored only the PAR3 relocalisation process within the oocyte. We observed that after photobleaching, both by quantity and through normalised fluorescent measurement, PAR3 accumulates progressively at the anterior while being excluded from the posterior (Figure 8B and C). This indicates that PAR3, already present in oocytes, can be transported to the APM. The difference in quantity between the PAR3 which accumulates at the anterior and that which is excluded from the posterior, could be explained by the presence of PAR3 in transit into the cytoplasm.

PAR3 recovery after APM photobleaching.

(A) PAR3-GFP expressed in ovarian follicle (Tub67c-GAL4; UASp PAR3-GFP) is photobleached in all the nurse cells and at the APM (yellow area, A1). The fluorescence recovery was followed for around 1400 s (A2). (B) PAR3 quantity in each domain (anterior and posterior domains) was quantified using the same method as previously for three ovarian follicles and raised to 0 after bleaching. It can be seen that, after photobleaching, PAR3 accumulates progressively at the anterior while it is excluded from the posterior. (C) PAR3 quantity of each zone before FRAP was normalised to 1, and recovery of fluorescence was observed. (D) The same experiment as in (A) was performed on ovarian follicle incubated with colcemid. The quantification is shown in graphs in (C) and (D). In (B) and (C), the error bar represents SEM. The scale bars represent 20 µm.

https://doi.org/10.7554/eLife.40212.039
Video 1
Recovery of fluorescence after photobleaching of PAR3 at the APM domain.

Representative FRAP experiment on PAR3-GFP (Tub67c-GAL4; UASp PAR3-GFP) in Drosophila egg chamber.

https://doi.org/10.7554/eLife.40212.042

We next addressed whether this transport in the oocyte relies on the MT cytoskeleton, as predicted from our results. We repeated PAR3 FRAP experiments in the presence of colchicin (Figure 8D). We observed that after photobleaching, the anterior accumulation and the posterior exclusion of PAR3 were significantly reduced (Figure 8C). Thus, this result shows that the MT network is important for anterior relocalisation of PAR3 fraction from the posterior in the oocyte.

IKKε/IK2 controls asymmetrical localisation of PAR3

We have shown that PAR3 localisation depends on the MT network and its associated dynein motor. In the cytoplasm, PAR3 localised on RAB11-positive vesicles and RAB11 is important for PAR3 APM accumulation. Our results suggest that PAR3 is transported in association with MTs and recycling endosome vesicles to the APM. Interestingly, in the proteomic screen for SKTL partners, we also identified the IKKε/IK2 kinase. IKKε has been shown to regulate RAB11-associated cargo transport with dynein in developing bristles (Otani et al., 2011). Furthermore, IKKε phosphorylates a RAB11 cofactor, NUF, to release the cargo from the dynein motor at the MT minus end. Interestingly, in oocytes, IKKε is specifically enriched at the APM (Dubin-Bar et al., 2008). We then monitored whether IKKε is required for PAR3 polarised distribution. Upon IKKε RNAi-mediated knockdown in the oocyte, PAR3 becomes isotropic all along the plasma membrane, without clear accumulation at the APM (Figure 9B). Thereafter, PAR3 piles up in the cytoplasm (Figure 9C) in large circular structures surrounded by actin mesh that we called actin clumps (ACs) (Figure 9A and Figure 9—figure supplement 1A). We further noted that aPKC, a partner of PAR3, is also present on ACs (Figure 9—figure supplement 1C). Interestingly, ACs accumulate near MT minus ends close to the oocyte nucleus as revealed by Nod-LacZ transgene (Clark et al., 1997) (Figure 9—figure supplement 2B–C).

Figure 9 with 2 supplements see all
IKKε regulates PAR3 microtubule unloading and APM accumulation.

(A–C) Distribution of PAR3 in response to IKKε knockdown. (A) IKKε knockdown affects the localisation of PAR3 and leads to an accumulation of circular actin clumps (ACs) enriched in PAR3. (B) Quantification of PAR3 density at each plasma membrane domain from stage 9A to stage 10 oocytes in WT or IKKε knockdown contexts (Tub67c-GAL4; UASp RNAi ikkε; UASp PAR3-GFP). (C) Quantification of PAR3-GFP distribution in the cytoplasm related to the whole oocyte intensity at stage 9B in WT or in IKKε knockdown contexts. For (B, C), Control (stage 9A, n = 9; stage 9B, n = 15; stage 10, n = 14); RNAi IKKε (stage 9A, n = 15; stage 9B, n = 10; stage 10, n = 9). (C) Mann-Whitney test. Error bars indicate SEM. **** indicates p<0.0001. (D, E) Actin clumps (D’, E’, green) contain RAB11 (magenta, D’’), but no RAB5 (E’’, red) is around these actin clumps. Actin is visualised after staining with phalloidin. (D’–D’’’) and (E’–E’’’), are magnifications of D and E (white frame). (F, G) In an IKKε knockdown oocyte, ACs contain PI(4,5)P2, visualised with PHPLC GFP probe, (F) and Myc-SKTL, visualised with Myc tag (G). (H, I) PI(4,5)P2 or SKTL are involved in formation of the ACs. ACs are reduced in SKTLDNRQ context (H) and disappearin sktl2.3/sktl∆5 context (I). (J, K) MTs are necessary for actin ring formation. Flies were nourished with colchicin for 24 h (K) or only yeast paste (J). Actin (green) is visualised after staining with phalloidin and the nuclear membranes after staining with WGA (red). The oocyte nucleus position is indicated by an ‘N’ in (G), (J), and (K).

https://doi.org/10.7554/eLife.40212.043

We then investigated the nature of those structures and their association with endocytic vesicles. ACs colocalise with RAB11 (Figure 9D) and NUF positive-vesicles (Figure 9—figure supplement 1B). RAB5 vesicles are not enriched in the ACs, but are all around them (Figure 9E). Furthermore, there is no colocalisation with Syntaxin16 highlighting the Golgi and lysosome compartment (Akbar et al., 2009) (Figure 9—figure supplement 1D).

These structures also contain SKTL (Figure 9G) and PI(4,5)P2 (Figure 9F), suggesting that PAR3, is still associated with PI(4,5)P2/SKTL membrane during its transport. ACs are strongly reduced when SKTL kinase dead form is expressed (SKTLDNRQ, Figure 9H), and disappear completely in sktl mutant (sktl2.3/sktl∆5, Figure 9I). Hence, in the absence of IKKε, SKTL activity and PI(4,5)P2 are necessary for AC accumulation.

Furthermore, in the absence of IKKε, the dynein subunit DLIC, which interacts with PAR3 and SKTL, is held back on structures similar to ACs (Figure 9—figure supplement 2A) at the MT minus ends (Figure 9—figure supplement 2B–C). Importantly, in the absence of MTs, upon colchicin treatment, ACs disappear, indicating that their accumulation depends on the MT network (Figure 9K compared to Figure 9J). Thus, in oocyte, IKKε seems to regulate the connection between PAR3 associated to RAB11-positive vesicles and MT minus ends.

Discussion

Our results shed light on a dual step process that sustains PAR3 asymmetry in the Drosophila oocyte (Figure 10). The first step occurs at the middle of stage 9 and is responsible for PAR3 exclusion from the PPM. It involves PAR1, the actin cytoskeleton, PI(4,5)P2 and RAB5-dependent endocytosis. The second step relates to PAR3 accumulation at the oocyte anterior side. It brings into play MT-associated transport with the minus end directed motor, dynein in association with vesicular trafficking. Both steps rely on PAR3 interactions with PI(4,5)P2 endosomal vesicles.

Speculative model of PAR3 localisation regulation.

(A) In Drosophila oocytes, PAR3 asymmetrical localisation proceeds in at least two steps. The first step (C) occurs at the posterior plasma membrane and leads to the PAR3 exclusion starting from stage 9B. This step implicates PAR3 phosphorylation by PAR1 (C2) and RAB5-dependent endocytosis (C1). PIP5K, SKTL, and its product PI(4,5)P2 (orange) are crucial for this process, as well as the actin cytoskeleton (blue gradient). As PIP(4,5)P2 is critical for the endocytosis onset but also for actin cytoskeleton regulation, the link between all these protagonists remains hypothetical. Furthermore, as the polarised localisation of PAR1 depends on actin, we cannot exclude a role for PAR1 on endocytosis directly or indirectly by PAR3 phosphorylation. The second step (B) takes place at the anterior cortex where PAR3 has to be strongly enriched. Dynein-dependent transport brings PAR3 to the MT minus ends likely with recycling endosomal cargo (RE). Then PAR3 cargo would be released from the dynein through an IKKε-dependent process. However, how PAR3 reaches the cortex is unknown. PI(4,5)P2 is also important for the PAR3 endosomal sorting and we can speculate that an association with vesicles is required for anterior cortex targeting. Finally, while we have shown that the posterior fraction of PAR3 provides, through MTs, the anterior pool of PAR3, the neo-synthesized fraction of PAR3 from the nurse cell could also contribute to the anterior pool. As the MT network presents only a slight bias of minus end directed transport in the oocyte, it is thus possible that PAR3 is transported step by step along the lateral cortex before reaching the anterior membrane.

https://doi.org/10.7554/eLife.40212.049

The precise role of the actin cytoskeleton in exclusion of PAR3 from the PPMremains to be clarified. An attractive hypothesis is that the posterior actin network could act on PAR3 removal from the plasma membrane, possibly by direct regulation of the endocytosis process (Figure 10C1). In support of this, a connection between actin and a specific Oskar-dependent endocytosis pathway has already been highlighted at the oocyte posterior pole (Tanaka and Nakamura, 2011; Vanzo and Ephrussi, 2002; Vazquez-Pianzola et al., 2014). However, the actin requirement could also be connected with the role of actin in PAR1 posterior localisation (Doerflinger et al., 2006). Indeed, in the presence of latrunculin, PAR1 is not focused on the PPM but its localisation become isotropic. Consequently, posterior PAR3 would be less phosphorylated by PAR1 and so less excluded.

How is the posterior to anterior redistribution of PAR3 achieved in the oocyte? Recent analyses on MT-associated transport of mRNA particles in the oocyte have shown that the MT network is polarised with a slight bias for MT plus ends towards the posterior (Trovisco et al., 2016; Zimyanin et al., 2008). As PAR3 is associated with the MT minus end directed dynein motor, this MT polarity bias could be used to relocate PAR3 associated with vesicles towards the APM (Figure 10A). However, this anterior redistribution is unlikely to be direct because of the weak MT polarisation bias (Khuc Trong et al., 2015). Hence, it may involve intermediate re-localisation steps along the LPM before being redirected toward the APM, which is in accordance with the observed PAR3 distribution.

In the last few decades, the mutual inhibitions of PAR1 and PAR3 have been well studied at a functional level, and the molecular mechanism relies on some direct or indirect phosphorylations (Figure 10C2). However, the fate of PAR3 after phosphorylation by PAR1 at the posterior is unknown. PAR3 does not seem to be degraded in a SLMB/E3 ubiquitin ligase-dependent manner at the PPM, unlike aPKC or PAR6 (Morais-de-Sá et al., 2014). Using the FRAP approach, we have shown that posterior fractions of PAR3 proteins contribute after re-localisation through a MT-dependent process to the anterior pool of PAR3. It is important to mention that this anterior fraction of PAR3 coming from the posterior of the oocyte is likely to be minor compared to the anterior pool of PAR3, which comes directly from the nurse cells.

Our results indicate here that PAR3 and PAR1 coexist at the PPM until stage 9A, indicating that the relations between them both are more complex and that an additional process may be required to exclude PAR3 from the PPM. Starting from stage 9B, PAR1 excludes PAR3 from the PPM but has little effect on PAR3 anterior accumulation. However, the strict exclusion of PAR3 on this small posterior domain is sufficient to obtain an important anterior/posterior asymmetry.

PAR3 is a cytoplasmic protein, but here we show that it is often found in association with membranes, the plasma membrane or endocytic membrane (early/recycling endosome). Interestingly, in mammalian epithelial cells PAR3 associates with exocyst components (Ahmed and Macara, 2017). This association is present in the oocyte, but is also observed in nurse cells, particularly at the front of the ring canals where PAR3 accumulates in PI(4,5)P2 and RAB-positive compartments. How is PAR3, a cytosolic protein, transported with endocytic membranes? PAR3 could be associated to a cargo bound to vesicles, which would then be transported to the anterior. In parallel, PAR3 could also control its own association with the vesicles. Indeed, PAR3 interacts with SKTL, which, in return, produces PI(4,5)P2 that stabilises PAR3 to the membrane (Claret et al., 2014; Krahn et al., 2010b). This further points to the importance of SKTL in PAR3 APM accumulation and PPM exclusion. As PI(4,5)P2 is a well-known regulator of the first steps of endocytosis (Posor et al., 2015), we hypothesise that SKTL acts on PAR3 PPM exclusion by regulating the formation of vesicles required for future PAR3 anterior targeting. Our results also provide evidence for a role of PI(4,5)P2/SKTL in recycling endosome sorting. PI(4,5)P2 has been described as present both at the cell surface and on the distal portions of the tubular endosome (Brown et al., 2001). PI(4,5)P2 is required for the recruitment to the membrane of many proteins involved in vesicle formation, fusion, and actin polymerisation (Posor et al., 2015). This suggests that PI(4,5)P2 and SKTL are important on membranes, including those of recycling endosomes, for the budding and the sorting of specific elements of PAR3.

Here, we show that IKKε participates in establishment of PAR3 asymmetry in the oocyte. Upon IKKε knockdown, PAR3 accumulates with aPKC on RAB11 positive-vesicle aggregates, suggesting that PAR3 transport at the MT minus ends is arrested. In polarised bristles, IKKε was previously shown to take the RAB11 endocytic vesicles down to the MT minus ends (Otani et al., 2011). The mechanism could be similar in oocytes (Figure 10B). In IKKε knockdown, vesicle aggregates accumulate at the MT minus ends with dynein subunit DLIC. This accumulation depends on MTs, but also on PIP(4,5)P2/SKTL. These vesicles enriched for PI(4,5)P2, SKTL and able to nucleate actin present common characteristics with the anterior plasma membrane. They could constitute a pre-built platform in the cytoplasm before being targeted to the appropriate position at the plasma membrane. This idea of a preformed platform has already been suggested in the tracheal system, where PAR3 associates with recycling endosomes containing E-cadherin during adherens junction rearrangements (Le Droguen et al., 2015). Moreover, during de novo apical lumen formation in MDCK cells and in human pluripotent stem cells, there is an accumulation of apical components in RAB11 positive-vesicles that are subsequently delivered to the apical plasma membrane (Overeem et al., 2015; Taniguchi et al., 2017). The first sign of apical domain formation in these cells is the relocation of the polarity protein PAR3.

Materials and methods

Fiji macro and quantification methods

To quantify membrane and cytoplasm repartition of proteins in oocytes, we developed a macro on Fiji (cf. Oocyte Analysis macro). In each oocyte, we selected three points (two on both sides of the anterior membrane and one in the middle of the posterior membrane). Four sections of its plasma membrane were then automatically generated (anterior; lateral 1; lateral 2; posterior) independently of the oocyte stage (see Figure 1—figure supplement 3 for details). After delimitation of the plasma membrane with the plot profile tool, we obtained its intensity profile. After delimitation of each plasma membrane domain with the polygon tool, we obtained the mean fluorescent intensity and the length of these domains. The anterior signal corresponds to the signal from the APM of the oocyte and the signal from the plasma membrane of the neighbouring nurse cells. Therefore, for each oocyte individually, we quantified the mean fluorescent intensity of a simple and a double plasma membrane of adjacent nurse cells. We then removed, from the anterior signal, the mean fluorescent intensity of the nurse cell plasma membrane to precisely quantify the signal coming only from the APM of this oocyte (Figure 1-Figure supplement 3C). Finally, after delimitation of the cytoplasm, we obtained its mean signal intensity. For each oocyte individually, the measured intensity signals were normalised in two steps. First, all the raw quantities (grey level intensity) were divided by the total intensity in the oocyte (APM + PPM + LPM + cytoplasm). Second, the plasma membrane values were expressed in density, that is the length of the corresponding membranes that divides them (see intermediate quantification in Figure 1—figure supplement 2). This process was performed on each oocyte individually, and then the results were pooled by genotypes and/or conditions and/or developmental stages. In our representation of the density, only the plasma membrane domains are indicated and not the cytoplasm. So the density of the different cortical zones does not add up to 1. The quantity corresponds to the grey level intensity measured on the image. The density represents the same values divided by the length of corresponding membranes. Between stage 9A and a stage 10, the oocyte plasma membrane overgrows. The density shows how PAR3 are concentrated at the membrane by avoiding the oocyte and plasma membrane size variation. The posterior exclusion ratio is the ratio of LPM density on PPM density. The asymmetry ratio is the ratio of APM density on PPM density.

To quantify the colocalisation of PAR3 with other proteins, the Coloc two plugin (Fiji) was used. Oocytes (between 7 and 13) of several females per genotype were analysed; the cytoplasmic contour was delimited and the quantification was performed only for one plan in this zone. The Pearson correlation coefficient (PC) or the Manders overlap coefficient (MOC) was determined in single confocal images. The PC value ranges from −1 (no correlation) to +1 (complete correlation), with values in between indicating different degrees of partial correlation. The negative control was provided by quantifying Pearson’s coefficient for the same images, but after rotation of one by 90 degrees, a condition in which only random colocalisation is observed. The MOC measures the fraction of PAR3 signal overlapping with red signal and vice versa.

Fly stocks

Mutant sktl2.3 has been described in (Gervais et al., 2008), sktl∆5 in (Hassan et al., 1998), and par-1w3 and par-16323 in (Shulman et al., 2000). The following fly stocks were also used: canton-S as wild type; UASp-PAR3-GFP and UASp-PAR3AA-GFP (Benton and St Johnston, 2003); UASp-GFP-PAR1(N1S) (Doerflinger et al., 2006); Ubi-PHPLC-RFP (Claret et al., 2014); pUbi-DLIC-GFP (Pandey et al., 2007); Pubi-PDI-GFP (Bobinnec et al., 2003); UASp-RAB5DN(S43N) (Pelissier et al., 2003); FRT82B rab11P2148/J2D1 (Jankovics et al., 2001); Bac PAR3-GFP = P[w+, FRT9-2]18E, f, baz[815.8], P{CaryP, PB[BAC Baz-sfGFP2]attP18} (Besson et al., 2015); the knockdown stocks from the Transgenic RNAi Project (TRiP, Bloomington Drosophila stock center) UAS-RNAi par1GL00253; UAS-RNAi dhc64GL00543 (Sanghavi et al., 2013); UAS-RNAi ikkεGL00160; UAS-RNAi rab5HMS00147 (BL34832); UAS-RNAi mCherry (BL35785); PAR3-Trap = P{PTT-GC}bazCC01941 (BL51572); HspFLP; FRT82B RFP; the stocks UASp-Myc-SKTL and UASp-Myc-SKTLDNRQ, that we established. Tub67c-GAL4 (Januschke et al., 2002) and Osk-GAL4 VP16 (Telley et al., 2012) were used to express transgenes in germinal cells. Tub67c-GAL4 was used to express all UAS transgenes, except the knockdown stock UASp-RNAi dhc64GL00543 where Osk-GAL4 VP16 was used. The crosses were kept at 25°C. However, to express two UAS transgenes, the flies were kept at 29°C, 24 h before dissection. rab11P2148 germline clones were generated as described in Compagnon et al. (2009), and selected against RFP.

Drug treatment

Flies were fed with yeast paste, on vinegar agar plates, containing 1 mM of latrunculin B (Sigma), in DMSO, (sigma) or DMSO alone as control for 48 h, or 16 µM of colchicin (Sigma) for 24 h or 48 h. Ovaries were then dissected, fixed, and stained using standard procedures.

Immunohistochemistry

Immunostainings were performed using standard protocols. The following primary antibodies were used: rabbit anti-SKTL at 1:20000 (Claret et al., 2014); mouse anti-Myc 9E10 (Santa Cruz) at 1:250; rabbit anti-RAB11 at 1:8000 (Nakamura); rabbit anti-RAB5 at 1:50 (Marcos Gonzalez-Gaitan); guinea pig anti-HRS at 1:500 (H. Bellen); mouse anti-KDEL at 1:300 (abcam 10C3, RRID: AB_298945); rabbit anti-NUF at 1:500; mouse anti-spNF at 1:50 (Abdu Uri); rabbit anti-Syntaxin16 at 1:1000 (R.Leborgne); rabbit anti-PKC (Santa Cruz) at 1:1000; mouse anti-HTS (Hybridoma, RRID: AB_528289) at 1:10; mouse anti-ß-Gal (Promega) at 1:250; rabbit anti-Staufen at 1:200. F-actin was visualised after staining with rhodamine-phalloidin (RRID: AB_2572408) or Alexa488-phalloidin (RRID: AB_2315147; life technology) at 1:200. Alexa594-WGA (Molecular probes) was used at 1:100 to stain nuclear membrane. Images were obtained with a ZEISS LSM700 confocal microscope and a Leica TCS-SP5 AOBS inverted scanning microscope.

Molecular biology

SKTL kinase dead mutant, SKTLDNRQ, was created using the PCR overlap mutagenesis method of insertion, as has been done on mammalian PIP5Kα (Coppolino et al., 2002), two punctual mutations in SKTL sequence (Asp398 mutated in Asn and Arg564 mutated in Gln).

Using the gateway recombination cloning, SKTL and SKTLDNRQ sequences were tagged with 6 MYC at the N terminal (in the vector pPMW, Murphy Lab). The sequence AttB used by the phiC31 integrase was inserted into the vector pPMW. The constructions were then integrated by transgenesis into the AttP2 site on chromosome III (BestGene).

Co-immunoprecipitation and western blot

Ovary extracts were obtained from wild type and DLIC-GFP transgenic flies by dissecting ovaries (20 flies per genotype) into PBS. Ovaries were placed on ice with lysis buffer (10 mM Tris/Cl pH7.5; 150 mM NaCl; 0.5 mM EDTA; 0.5% NP-40 with Complete Protease Inhibitor cocktail, Roche) and mechanically homogenised using micro pestles in matching tubes. S2-(DGRC (RRID: CVCL_TZ72)) cell lysates were obtained after freeze-thaw lysis of cells transfected with GFP-SKTL with PAR3-HA. Ovaries and cells lysates were then spun at 16000 g for 10 min at 4°C. Using Chromotek standard procedures, the supernatant was collected and GFP-SKTL, PAR3-GFP, or DLIC-GFP were precipitated with 15 µL GFP-Trap_MA (Chromotek) magnetic beads for 2 h at 4°C with rotation. Input and bound fractions were analysed by SDS-PAGE and western blotting, using NuPage, 4–12% Bis-Tris gels (Life technology). DLIC was detected using 1:5000 guinea pig anti-DLIC C-ter (Satoh et al., 2008), PAR3-HA using rabbit 1:5000 Anti-HA (GenWay), PAR3 using 1:500 rabbit anti-PAR3 N-ter (Wodarz et al., 1999), PAR1 using 1:100 rabbit anti-PAR1 (Rb96, J. McDonald), GFP using 1:1000 mouse anti-GFP (Roche), or 1:5000 rabbit anti-GFP, Tubulin using 1:1000 mouse anti-tub (DM1A sigma). For the immunoprecipitation controls, we used PDI-GFP in fly extracts and GFP-SNAP (a kind gift from M. Sanial) in S2 cells extracts.

Proteomic SKTL partner screen

S2 cells (S2-DGRC (RRID: CVCL_TZ72) were transfected with SKTL-GFP vector or not (control) and harvested 3 days later. Protein extraction and immunoprecipitation were performed using Chromotek standard procedures. After trypsin digestion, the samples were analysed by a LTQ Velos Orbitrap (Thermo Fisher Scientific). Among the potential partners identified, we focused on two proteins: DLIC (one peptide identified not present in the control: SGSPGTGGPGGAGNPAGPGR score 80) and IKKε (two peptides identified not present in the control: VMQQQQQEVMAVMR and LLAIEEDQEGR, score 24).

FRAP experiment

Egg chambers were dissected in Schneider’s Insect Medium supplemented with 13% FBS, 0.34 U/ml insulin. Egg chambers were cultured in a 150-µm-deep micro chamber filled with the same medium and sealed on one side with a 0.17 μm coverslip matching the characteristics of the facing objective lens, and a membrane permeable to oxygen on the other side. Imaging was carried out with spinning disk confocal (Axio Observer Z1 (Zeiss), Spinning Head CSU-X1, Camera sCMOS PRIME 95 (photometrics)) using a Plan-Apochromat 40×/1.25 oil objective lens. For each egg chamber, six prebleach images were taken. Then, the region of interest (ROI) corresponding to nurse cells and oocyte APM was scanned 25 times with the appropriate laser line at full laser power (473 nm) to photobleach GFP fluorescence. Fluorescence recovery into the ROI was monitored immediately after the bleach by time-lapse imaging at low-intensity illumination. Time-lapse recordings were processed with Fiji. The FRAP efficiency was controlled with the entire thickness of the sample.

The same experiment was realized on egg chambers treated with colcemid. The egg chambers were incubated for 30 min in Schneider medium, 2% FBS, 0.2 U/ml insulin with 0.82 µM of colcemid, before the onset of the FRAP experiment, and the egg chambers stayed in this medium during the fluorescence recovery.

Endocytosis assay

To confirm the effect of RAB5DN expression on endocytosis, we performed an endocytosis assay using the lipophilic dye FM4-64 (Molecular probes). This dye partitions into membranes where it becomes fluorescent and can be internalised in the cell by endocytosis.

Egg chambers were dissected in Schneider’s Insect Medium supplemented with 20% FBS, 10 μg/ml of juvenile hormone, and 0.2 U/ml insulin. Next, they were incubated in the dark with the same medium +20 μM FM4-64 for 45 min at 25°C and then rinsed two times and washed for 15 min at 4°C in medium without FM4-64. The oocytes were immediately observed by confocal microscopy.

Statistical analysis

Data were analysed using the Mann-Whitney test and were represented by box plot (with Tukey variation) thanks to Graphpad Prism 6. All plots are expressed as the mean ± standard error of the mean (SEM).

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
    Reciprocal localization of nod and kinesin fusion proteins indicates microtubule polarity in the Drosophila oocyte, epithelium, neuron and muscle
    1. IE Clark
    2. LY Jan
    3. YN Jan
    (1997)
    Development 124:461–470.
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
    Skittles, a Drosophila phosphatidylinositol 4-phosphate 5-kinase, is required for cell viability, germline development and bristle morphology, but not for neurotransmitter release
    1. BA Hassan
    2. SN Prokopenko
    3. S Breuer
    4. B Zhang
    5. A Paululat
    6. HJ Bellen
    (1998)
    Genetics 150:1527–1537.
  26. 26
  27. 27
    An interaction type of genetic screen reveals a role of the Rab11 gene in oskar mRNA localization in the developing Drosophila melanogaster oocyte
    1. F Jankovics
    2. R Sinka
    3. M Erdélyi
    (2001)
    Genetics 158:1177–1188.
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
    Phosphoinositides in endocytosis
    1. Y Posor
    2. M Eichhorn-Grünig
    3. V Haucke
    (2015)
    Biochimica Et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1851:794–804.
    https://doi.org/10.1016/j.bbalip.2014.09.014
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
    Oskar anchoring restricts pole plasm formation to the posterior of the Drosophila oocyte
    1. NF Vanzo
    2. A Ephrussi
    (2002)
    Development 129:3705–3714.
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75

Decision letter

  1. Matthew Freeman
    Reviewing Editor; University of Oxford, United Kingdom
  2. Anna Akhmanova
    Senior Editor; Utrecht University, Netherlands

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Dynein-mediated transport and membrane trafficking control PAR3 polarised distribution" for consideration by eLife. Your article has been reviewed by a Senior Editor, a Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

You will see from the individual reviews below that there was significant appreciation of the potential interest of this work. Overall, however, the reviewers felt that there were too many significant revisions that would need to be done. There was, however, a feeling that this work might become appropriate for eLife were it to be further developed, and the reviewers did not preclude future submission of a more mature story. It would, however, be treated as a new submission.

Reviewer #1:

In this manuscript, Jouette and colleagues examine the mechanism by which opposing gradients of Par1 and Par3 are established in the Drosophila oocyte. These polarity proteins, and their mutual antagonism are critical to polarity establishment in a number of cell types and a variety of organisms. This is a highly significant topic and an understanding of this mechanism will be of interest to a wide audience.

The main focus of this paper is the mechanism by which Par3 is enriched along the anterior cortex of the oocyte and excluded from the posterior pole. Initial studies suggested that posterior exclusion was predominantly via Par1-mediated phosphorylation of Par3. The authors convincingly show that although this mechanism is important in stage 10 egg chambers, additional factors are also at play. The authors further suggest that Dynein, microtubules and endocytic trafficking are responsible for targeting Par3 to the anterior cortex. This latter point, however, is not convincingly supported by the data presented. In general, there are some very interesting findings in this paper. However, there are also major deficits in interpretation of results and in the controls that are used. Ideally, these issues would be fixed prior to publication.

Essential revisions:

The authors conclude that Par3 interacts with Dlic. The interaction is very hard to appreciate from the data that is presented. The enrichment of the Par3 bands in the "+" lane is barely above background. Secondly, why are there four bands corresponding to Par3? Are these degradation products or splice isoforms? According to FlyBase, the four isoforms of Par3 have sizes of 157, 163, 161 and 159 kD. That is not what we see here. The most prominent band appears to be around 55kD. The Par3-HA band that is shown in Figure 4G is around 250kD.

Figure 3. Par3-GFP is found to accumulate in small vesicles that are positive for π (4,5)P2. This appears convincing. However, the authors go on to state that these same vesicles are also positive for Rab5 and Rab11. The localization of Rab5 and Rab11 to these vesicles is not nearly as convincing. The authors have done a good job of quantifying the localization of Par3 along the membrane. It would be good to also quantify the co-localization of Rab5 and Rab11 with these Par3-GFP vesicles (for instance using a Pearson's coefficient analysis).

The connection between SKTL activity and Par3 localization is convincing. The authors then go on to state that protein interaction partners of SKTL were identified using purification and mass spectrometry. However, the data for this mass spec experiment is not shown. How many peptides of Par3 were found in the SKTL pellet in comparison to the control? The authors then go on to validate this interaction. For this experiment, they used S2 cells. What was the rationale for using S2 cells? All of the rest of the experiments were performed in the ovary. Does this interaction occur in ovarian lysates? An additional point about experimental setup. The negative control for this experiment appears to be "no antibody". A better negative control would be to use cells or extracts expressing GFP alone.

The authors conclude that IKKε regulates transport of recycling endosomes in the oocyte. Although this is possible, the data that is presented do not provide proof of this. What is shown is that depletion of IKKε results in abnormal actin structures at the anterior of the oocyte that also contain Rab11. The images presented are from fixed samples that are stained with antibody. The authors suggest that these vesicles are formed due to blocking of an unloading step. While this is certainly possible, the data only addresses localization.

The authors suggest that when the anterior (high)-posterior (low) distribution of Par3 is altered, oocyte polarity is affected and the oocyte nucleus is mis-localized. However, from Figure 2 it appears that Latrunculin treatment affects this Par3 AP distribution to a greater extent than Colchicin treatment. Yet, the oocyte nucleus is localized under Latrunculin conditions but delocalized under Colchicin treatment conditions. This does not seem consistent.

Reviewer #2:

In this study, Julie and coworkers quantify the plasma membrane distribution of Par3 in the Drosophila oocyte as it matures. During this process, Par3 becomes excluded from the posterior (PPM) and accumulates at the anterior plasma membrane (APM). In addition to the well-known effects of Par1, the authors report that exclusion from the PPM depends on Rab5 and PIP2-dependent endocytosis. They also argue that microtubules and dynein are required for transport to the APM, where it is unloaded through an IKK-related kinase. Interestingly, they show that Par3 can bind to Sktl, a PI4P5-kinase.

This is an interesting study and the careful quantification of Par3 localization is a useful approach. However, I found the work overall to be rather diffuse and the interpretations were highly over-stated. At a general level, it is not clear to me how the authors know that Par3 is in fact being transported from the PPM to the APM – each event (loss from the PPM and accumulation at the APM) could be independent of one another, and this possibility is not excluded in this study. Indeed, there does not seem to be any net gain of Par3 at the APM from stages 8 – 10 (Figure 1). Par3 association with membranes is highly dynamic, through a process likely controlled by phospho-inositide binding, oligomerization, and phosphorylation. It might diffuse through the cytoplasm and simply bind to any membrane with high levels of PIP2. In addition, new Par3 protein could be delivered from the nurse cells.

Essential revisions:

1) The studies of the role of actin and microtubules are rather superficial – for instance the actin data are based only on latrunculin treatment, and the authors assume, based only on the similar phenotype induced by loss of Par1 or expression of a non-phosphorylatable Par3, that the actin network is mediating the effects of Par1 on Par3. This is not a valid conclusion – the same phenotype could be generated through 2 independent mechanisms.

2) Figure 2J (co-IP of Par3 with DLIC-GFP) is not convincing. The negative control lane input is lower than the positive. Also, a better negative control is needed (another GFP fusion protein); the DLIC-GFP input needs to be shown; and some evidence that the indicated bands actually represent Par3 needs to be provided. And in Figure 2—figure supplement 2B an input control is needed.

3) Figure 3. The authors claim that Par3 particles are associated with PIP2-containing membranes and that Par3 is removed from the PPM by endocytosis on Rab5/PIP2 endosomes, but in Figure 3, most of the Par3 in cytoplasmic spots is NOT associated with Rab5 OR PIP2; and the single Rab5+ vesicle is much larger than any of the PIP2 positive spots. Overall, these data do not agree with the stated hypothesis. The staining for Rab5 and Rab11 is also very diffuse – it would be better to use FP-tagged Rabs for this type of experiment.

4) In Figure 3—figure supplement 1, how do we know the antibodies are specific? – the HRS antibody does not seem to detect vesicles.; and the syntaxin16 antibody decorates the plasma membrane but no detectable trans-Golgi.

5) The interaction between SKTL and Par3 is based only on over-expressed proteins. The blot needs the GFP-SKTl input and a better (GFP fusion) negative control. Also, I could not find data to support the contention that Par3 forms a tripartite complex with SKTL and PIP2, as is stated in subsection “SKTL and PAR3 act in synergy to form vesicular platforms”. The authors also suggest that Par3 could regulate SKTL activity, and that this would provide a positive feedback loop, but again without any supporting evidence.

6) In Figure 5 it looks as though the mutant SKTL is expressed at lower levels than the WT. Could this, rather than loss of kinase activity, explain the lower GFP-Par3 in cytoplasmic spots? The colocalization with Rab5 and 11 is not convincing. Moreover, it looks as though the aAPM has disintegrated – is this not an issue with interpretation of the data?

7) Finally, I think it is premature to conclude much from the IKK-eta data – clearly knockdown generates some kind of cytoplasmic inclusion in the oocytes, that accumulate actin and Rab11, but it’s not clear what they are; and since Par3-GFP is still strongly accumulated all over the plasma membrane, including the APM (Figure 6A – I), I don't think that it is clear that IKKe is necessary for unloading it from microtubules.

8) In general, the authors seem to interpret their data as if Par3 is an integral membrane protein that is being endocytosed and transported by the microtubule machinery, but of course it is not – it is a peripheral protein that associates with negatively-charged phospholipids in a highly dynamic way and is presumably able to diffuse through the cytoplasm between vesicles and plasma membrane.

Reviewer #3:

Jouette and colleagues have examined mechanisms that position the polarity protein Baz/Par-3 in a well-established model of cell polarization, the Drosophila oocyte. They examine a number of mechanisms involved in the cortical removal, sub-cellular trafficking and intracellular compartment release of Baz. The trafficking of Baz through the cytosol is poorly understood, and the results of this paper make significant advances in elucidating this process. However, major concerns should be addressed.

Major concerns

1) All of the Baz distributions are determined with an over-expressed Baz construct. Thus, it is unclear how relevant identified mechanisms are for endogenous levels of the protein. The newly identified mechanisms should be additionally evaluated with probes for endogenous Baz (using either Baz antibodies or an available Baz-GFP gene trap line).

2) Connections between the mechanisms that remove Baz from the cortex are unclear. Possible connections should be investigated with the tools already used to determine if the mechanisms are independent or interdependent.

2.1) Abnormal Baz accumulations arise along the later cortex with dynein RNAi. Are these accumulations dependent on cortical removal steps mediated by Par-1, PIP2 or Rab5? Perturbations of these proteins in pair-wise combinations could reveal dependent or independent cortical removal mechanisms.

2.2) Skittles overexpression is shown to increase the removal of Baz from the posterior domain. Does this removal require Par-1 or Rab-5? Again, combining the Skittles overexpression with the Par-1 RNAi or the Rab5DN could address how interdependent the removal mechanisms are.

3) Since Baz becomes isotropic around the cortex with IKK-epsilon knock down, does IKK-epsilon also play a role in the removal of Baz from the posterior cortex? If IKK-epsilon only promoted delivery to the anterior cortex, then removal from the posterior would still be expected. Some assessment of IKK-epsilon's direct or indirect role at the posterior would be expected within the overall context of the paper.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Dynein-mediated transport and membrane trafficking control PAR3 polarised distribution" for consideration by eLife. Your article has been reviewed by Anna Akhmanova as the Senior Editor, a Reviewing Editor, and two reviewers. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

This manuscript (an earlier version of which which has been previously reviewed by eLife but is now being treated as a new submission rather than a resubmission) provides evidence for a pathway that polarizes the distribution of Baz (Par-3) across the Drosophila oocyte, an important model of cell polarity. The model is an interesting and, if fully supported by the data, would be appropriate for publication in eLife. However, there are significant concerns that would need to be addressed before it could be accepted.

Essential revisions:

1) Generally, the quality of the biochemical evidence is not as strong as it should be. For example, in Figure 7E and F, the IPs of GFP-SKTL testing for pull-down of PAR3 and DLIC, the control was no addition of GFP antibody. Since addition of antibodies can increase non-specific binding, the better control is to IP an irrelevant GFP protein with the GFP antibody (as done in Figure 7D). Especially when the protein interaction they are trying to show is weak, good controls are required.

Even in 7D, which does have an appropriate control, the co-IP is pretty weak. This does not invalidate it, but it should be acknowledged in the text.

In the earlier version it was unclear that the 250kD band that was observed on western blots actually corresponded to Par3. The authors have used RNAi strains to address this point and the conclusion is a bit more convincing. It is still hard to fully appreciate, though. For instance, in Figure 7—figure supplement 2, a strain over-expressing Par3-GFP using the UAS-Gal4 driver is used. These embryos should be expressing a relatively high level of Par-GFP. In the total fraction from these embryos the most prominent band is around 100kD when probed with a Par3 antibody. However, this band is hard to detect with the GFP antibody. In fact, it is hard to see any bands in the GFP blot. Next, the sample was immunoprecipitated. It is unclear from the legend which antibody (Par3 or GFP) was used in the IP. In the immunoprecipitated fractions, the most prominent band is the 250kD band. The band is reduced in the strain expressing Par3 RNAi but it is not clear how it can be detected so prominently in the IP fraction when the same isoform is a minor species at best in the total fraction. Is the IP antibody somehow specifically precipitating an oligomeric form? The experiment with the endogenously expressed BAC-based Par3-GFP construct is more convincing.

2) Another major issue is the over-reliance on overexpressed Par3. Key results should be repeated to look at endogenous Par3 expression. For example, are the Par3-positive cytoplasmic particles, and effects on these particles, observed when Par3 is expressed by its own regulatory regions? Without this additional data, there is a risk that the effects they report are artifacts of overexpression.

3) The results using a dominant negative rab5 should be confirmed with an shRNA construct (available from Bloomington). There is a risk that the DN might produce off target effects.

https://doi.org/10.7554/eLife.40212.056

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

You will see from the individual reviews below that there was significant appreciation of the potential interest of this work. Overall, however, the reviewers felt that there were too many significant revisions that would need to be done. There was, however, a feeling that this work might become appropriate for eLife were it to be further developed, and the reviewers did not preclude future submission of a more mature story. It would, however, be treated as a new submission.

Reviewer #1:

In this manuscript, Jouette and colleagues examine the mechanism by which opposing gradients of Par1 and Par3 are established in the Drosophila oocyte. These polarity proteins, and their mutual antagonism are critical to polarity establishment in a number of cell types and a variety of organisms. This is a highly significant topic and an understanding of this mechanism will be of interest to a wide audience.

The main focus of this paper is the mechanism by which Par3 is enriched along the anterior cortex of the oocyte and excluded from the posterior pole. Initial studies suggested that posterior exclusion was predominantly via Par1-mediated phosphorylation of Par3. The authors convincingly show that although this mechanism is important in stage 10 egg chambers, additional factors are also at play. The authors further suggest that Dynein, microtubules and endocytic trafficking are responsible for targeting Par3 to the anterior cortex. This latter point, however, is not convincingly supported by the data presented. In general, there are some very interesting findings in this paper. However, there are also major deficits in interpretation of results and in the controls that are used. Ideally, these issues would be fixed prior to publication.

Essential revisions:

1) The authors conclude that Par3 interacts with Dlic. The interaction is very hard to appreciate from the data that is presented. The enrichment of the Par3 bands in the "+" lane is barely above background. Secondly, why are there four bands corresponding to Par3? Are these degradation products or splice isoforms? According to FlyBase, the four isoforms of Par3 have sizes of 157, 163, 161 and 159 kD. That is not what we see here. The most prominent band appears to be around 55kD. The Par3-HA band that is shown in Figure 4G is around 250kD.

As suggested by the reviewer, we have reiterated several times the IP experience of DLICGFP to validate the interaction between DLIC and PAR3. We provide a new blot presented in the Figure 7. In the course of those new experiments, as our previous PAR3 antibody made in rat was ran out, we have used a different PAR3 antibody produced in rabbit and kindly provided by A. Wodarz. We still observed several bands between 250 and 115 kDA. We can see that one of these bands (250kDa) immunoprecipitates with DLIC and this immunoprecipitation is reproducible. We have verified that this band corresponds to PAR3 and not to a non-specific protein. For this extend, we used a Drosophila line that produces PAR3-GFP at an endogenous level, and we revealed, after western blot on ovarian extracts, PAR3 with PAR3 antibody or with a GFP antibody. To better visualize PAR3, we have also realised an IP with the same extracts. The >250 kDa band is detected by PAR3 and GFP antibodies, indicating that the 250kDa band is really PAR3 (Figure 7—figure supplement 2). Furthermore, we have also validated the specificity of PAR3 antibody by monitoring PAR3 knockdown after RNAi. (Figure 7—figure supplement 2). The abnormal weight of PAR3 could be explained by the presence of a dimeric form of PAR3, the oligomerisation of PAR3 being already described (Benton and St Johnston, 2003; Dickinson et al., 2017; Kullmann and Krahn, 2018). Moreover, high molecular weight bands are also shown in several papers without explanations (Krahn et al., 2010; Krahn et al., 2009).

2) Figure 3. Par3-GFP is found to accumulate in small vesicles that are positive for π (4,5)P2. This appears convincing. However, the authors go on to state that these same vesicles are also positive for Rab5 and Rab11. The localization of Rab5 and Rab11 to these vesicles is not nearly as convincing. The authors have done a good job of quantifying the localization of Par3 along the membrane. It would be good to also quantify the co-localization of Rab5 and Rab11 with these Par3-GFP vesicles (for instance using a Pearson's coefficient analysis).

As suggested by the reviewer, we have quantified the colocalisation of PAR3 with PI(4,5)P2, RAB5 and RAB11 vesicles. The Pearson’s coefficient was obtained by using the FiJi Macro Coloc2 on our images. The negative control was provided by quantifying colocalisation for the same images, but after rotation of one by 90 degrees, a condition in which only random colocalisation is observed. The results are now presented in the manuscript Figure 5A. As indicated in the text, the PAR3 particles do not colocalise with the endoplasmic reticulum (KDEL), the Golgi apparatus (SYX16) or late endosomes (HRS). However, they colocalise with PI(4,5)P2 and RAB5 enriched compartments and to a lesser extent with RAB11 enriched compartments.

3) The connection between SKTL activity and Par3 localization is convincing. The authors then go on to state that protein interaction partners of SKTL were identified using purification and mass spectrometry. However, the data for this mass spec experiment is not shown. How many peptides of Par3 were found in the SKTL pellet in comparison to the control? The authors then go on to validate this interaction. For this experiment, they used S2 cells. What was the rationale for using S2 cells? All of the rest of the experiments were performed in the ovary. Does this interaction occur in ovarian lysates? An additional point about experimental setup. The negative control for this experiment appears to be "no antibody". A better negative control would be to use cells or extracts expressing GFP alone.

We feel that the reviewer has misinterpreted our results. We have performed a proteomic screen to identify some SKTL partners but we do not found PAR3. We have found DLIC and IKKε amongst others. However, as suggested by the reviewer, we have added for DLIC and IKKε the mass spec data in the Materials and methods section.

“DLIC: SGSPGTGGPGGAGNPAGPGR score 80.

IKKε: VMQQQQQEVMAVMR and LLAIEEDQEGR score 24.

The mass spectrometry experiment has been performed by using S2 cells extract. We agree with the reviewer that this experiment would be more informative if we had been able to use ovarian extract. The expression of GFP-SKTL is particularly deleterious in the Drosophila oocyte and we cannot perform the immunoprecipitation (IP) with SKTL overexpressed extract. We tried to immunoprecipitate a GFP-SKTL form produced at endogenous level in a CRISPR transgenic line without success. Moreover, the SKTL antibody is not good enough to use it. Therefore, we used S2 cells to realize this mass spectrometry experiment. However, to validate the interaction between PAR3 and DLIC, we performed the IP with ovarian extracts.

These peptides are not present in control.”

Concerning the negative control of this experiment, we have realized two different types of control. As suggested by the reviewer, we have tested the IP with another GFP transgene (PDI-GFP) and have observed no co-IP of PAR3 or DLIC. But we have also performed the co-IP on the same extract with or without anti-GFP antibody. We had chosen to show this last control because if we overexpressed one of these proteins, we can modify the migration profile of others. For example, PAR3 expression modifies the migration profile of DLIC and conversely (variation of postraductional modifications or change in isoforms ratio). We feel that it can be hazardous to compare two wells in which the protein contents are very different. In the case of DLIC/PAR3 IP (Figure 7D), we show now the IP of DLIC-GFP and in control the IP of PDI-GFP.

4) The authors conclude that IKKε regulates transport of recycling endosomes in the oocyte. Although this is possible, the data that is presented do not provide proof of this. What is shown is that depletion of IKKε results in abnormal actin structures at the anterior of the oocyte that also contain Rab11. The images presented are from fixed samples that are stained with antibody. The authors suggest that these vesicles are formed due to blocking of an unloading step. While this is certainly possible, the data only addresses localization.

We agree with the reviewer that this experiment is indicative, but is not sufficient to support a strong conclusion. Our conclusions on recycling endosome transport are built upon confocal analysis on fixed issues and not through live imaging analysis. The published data concerning IKKε functions in Drosophila bristles report that IKKε regulates the localisation of RAB11 and Dynein along with the rapid shuttling of recycling endosomes (Otani et al., 2015; Otani et al., 2011). The activation of IKKε by a local phosphorylation and, the spNF interaction with IKKε and dynein are conserved between bristle and oocyte in Drosophila (Amsalem et al., 2013; Otani et al., 2015). With our results, we can conclude that microtubules are important to the accumulation of RAB11 positive-vesicles near the microtubules minus ends as it has been reported in the Drosophila bristles. But we cannot conclude to a direct transport of endosomes on microtubules; we can only indicate that our, like the others suggest this point. So as requested by the reviewer we have modified this point in the Results section of the new version of the manuscript.

5) The authors suggest that when the anterior (high)-posterior (low) distribution of Par3 is altered, oocyte polarity is affected and the oocyte nucleus is mis-localized. However, from Figure 2 it appears that Latrunculin treatment affects this Par3 AP distribution to a greater extent than Colchicin treatment. Yet, the oocyte nucleus is localized under Latrunculin conditions but delocalized under Colchicin treatment conditions. This does not seem consistent.

The reviewer raises an important point. Indeed, what seems important to maintain a correct polarity within the oocyte (with the nucleus positioning as readout), is the PAR3 asymmetry between the anterior and posterior rather than the posterior exclusion of PAR3 stricto senso. For example, in presence of latrunculin, the posterior exclusion of PAR3 is affected but its anterior accumulation a very little. Thus, the ratio between anterior and posterior remains high, and the nucleus is well located. However, in presence of colchicin, the posterior exclusion is always observable but there is no/weak anterior accumulation. So, the ant/post ratio is weak and the nucleus is mispositionned. In the Figure 2, the ant/post ratio (asymmetry ratio) is not indicated and we propose to add this to Figure 2 in the revised manuscript.

Reviewer #2:

In this study, Julie and coworkers quantify the plasma membrane distribution of Par3 in the Drosophila oocyte as it matures. During this process, Par3 becomes excluded from the posterior (PPM) and accumulates at the anterior plasma membrane (APM). In addition to the well-known effects of Par1, the authors report that exclusion from the PPM depends on Rab5 and PIP2-dependent endocytosis. They also argue that microtubules and dynein are required for transport to the APM, where it is unloaded through an IKK-related kinase. Interestingly, they show that Par3 can bind to Sktl, a PI4P5-kinase.

This is an interesting study and the careful quantification of Par3 localization is a useful approach. However, I found the work overall to be rather diffuse and the interpretations were highly over-stated. At a general level, it is not clear to me how the authors know that Par3 is in fact being transported from the PPM to the APM – each event (loss from the PPM and accumulation at the APM) could be independent of one another, and this possibility is not excluded in this study. Indeed, there does not seem to be any net gain of Par3 at the APM from stages 8 – 10 (Figure 1). Par3 association with membranes is highly dynamic, through a process likely controlled by phospho-inositide binding, oligomerization, and phosphorylation. It might diffuse through the cytoplasm and simply bind to any membrane with high levels of PIP2. In addition, new Par3 protein could be delivered from the nurse cells.

The reviewer raises two very important points concerning PAR3 localisation mechanism.

The first point concerns the posterior PAR3 accumulation and the origin of its accumulation. Currently we present the PAR3 quantification, normalised to the total amount of signal in the oocyte and related to the membrane length (density). This representation avoids some fluctuations associated to the membrane growth during oogenesis or to experimental procedures but also allows us to compare different genotypes. Thus, in our graphs, we can follow the evolution of asymmetry, but we cannot compare the quantity between different stages. To respond to the reviewer, we have added a new graph with the raw data of PAR3 quantity (grey levels) in oocyte (Figure 1D). We have also presented in sup data an intermediate step of the quantification in which PAR3 density are shown without normalisation (Figure 1—figure supplement 2).

So, we can observe a strong enrichment of PAR3 at the APM from stage 8 to 10. We can also observe the posterior exclusion of PAR3, at a lower level because of the small size of the posterior domain compared to the anterior domain. The PAR3 APM accumulation can arise in part from the oocyte but also from the nurse cells that transcript PAR3. There is no PAR3 transcription in oocyte, so the contribution of oocyte to the APM accumulation can be only associated with a relocalisation of PAR3 protein. In order to identify the oocyte contribution to the PAR3 anterior accumulation, we performed FRAP experiments on GFP-PAR3 expressed in the oocyte and the nurse cells at stage 9. We suppressed the GFP-PAR3 fluorescence at the APM and in all the nurse cells. Next, we followed the fluorescence recovery to the APM before the recurrence of GFP-PAR3 in the nurse cells due to neosynthesis. We can observe the recovery of the APM signal, which is correlated with a posterior fluorescence decrease. This new result is presented in Figure 8. It suggests that among the fraction of PAR3 accumulated at the APM at least a part of it corresponds to PAR3 already present in oocyte. This process depends on microtubules because if the same experiment is performed in presence of colcemid, the level of recovery is severely affected.

The second point raised by the reviewer concerns the surprising association of PAR3 with membrane. The localisation of PAR3 at the right place does not necessarily require association with membranes. PAR3 could simply diffuse in the cytoplasm. In this work, we do not demonstrate that a membrane association is important to the targeting of PAR3 to the APM. However, we show that i) PAR3 colocalises with PI(4.5)P2 enriched vesicles and that the PI5K SKTL is crucial to the polarised localisation of PAR3, ii) the vesicular trafficking is also important to this localisation and iii) the kinase IKKε which is previously shown implicated in the recycling endosome release of the dynein motor, strongly modifies the PAR3 polarity. Thus, it seems that PAR3 has close links with the vesicular trafficking. We cannot exclude the possibility that PAR3 associates with all PI(4,5)P2 enriched membranes.

Essential revisions:

1) The studies of the role of actin and microtubules are rather superficial – for instance the actin data are based only on latrunculin treatment, and the authors assume, based only on the similar phenotype induced by loss of Par1 or expression of a non-phosphorylatable Par3, that the actin network is mediating the effects of Par1 on Par3. This is not a valid conclusion – the same phenotype could be generated through 2 independent mechanisms.

Actin plays a role in the PAR3 distribution control. But we do not know by which mechanism. The actin cytoskeleton controls the polarised localisation of PAR1 (Doerflinger et al., 2006) but also regulate the endocytosis trafficking (Mooren et al., 2012). The two being important for the localisation of PAR3, we cannot conclude on a direct or indirect role of actin on PAR3. This point has been removed from the result section and is addressed only in the discussion.

2) Figure 2J (co-IP of Par3 with DLIC-GFP) is not convincing. The negative control lane input is lower than the positive. Also, a better negative control is needed (another GFP fusion protein); the DLIC-GFP input needs to be shown; and some evidence that the indicated bands actually represent Par3 needs to be provided. And in Figure 2—figure supplement 2B an input control is needed.

As suggested by the reviewer, we have reiterated several times the IP experience of DLICGFP to validate the interaction between DLIC and Par3. We provide a new blot presented in the Figure 8. In the course of those new experiments, as our previous PAR3 antibody made in rat was ran out, we have used a different PAR3 antibody produced in rabbit and kindly provided by A. Wodarz.

Concerning the negative control of this experiment, we have realized two different types of control. As suggested by the reviewer, we have tested the IP with another GFP transgene (PDI-GFP) and have observed no co-IP of PAR3 or DLIC. But we have also performed the co-IP on the same extract with or without anti-GFP antibody. We have chosen to show this last control because if we overexpressed one of these proteins, we can modify the migration profile of others. For example, PAR3 expression modifies the migration profile of DLIC and conversely (variation of postraductional modifications or change in isoforms ratio). In this case, it can be hazardous to compare two wells in which the protein contents are very different. In the case of DLIC/PAR3 IP (Figure 7D), we show now the IP of DLIC-GFP and in control the IP of PDI-GFP.

Although this PAR3 antibody is published, we have further addressed its specificity on PAR3. We have used a Drosophila strain that express PAR3-GFP on its own promoter and in absence of endogenous PAR3: BAC PAR3GFP = P[w+, FRT9-2]18E, f, PAR3[815.8], P{CaryP, PB[BAC PAR3- sfGFP2]attP18} (Besson et al., 2015). We have compared the PAR3 protein profile revealed with anti-PAR3 antibody or with anti-GFP antibody. Furthermore, we have also validated the specificity of this antibody by monitoring PAR3 knockdown after RNAi.

This new result is presented in Figure 7—figure supplement 2. As requested by the reviewer we added an input control (using α-tubulin) in Figure 4—figure supplement 1.

3) Figure 3. The authors claim that Par3 particles are associated with PIP2-containing membranes and that Par3 is removed from the PPM by endocytosis on Rab5/PIP2 endosomes, but in Figure 3, most of the Par3 in cytoplasmic spots is NOT associated with Rab5 OR PIP2; and the single Rab5+ vesicle is much larger than any of the PIP2 positive spots. Overall, these data do not agree with the stated hypothesis. The staining for Rab5 and Rab11 is also very diffuse – it would be better to use FP-tagged Rabs for this type of experiment.

We have quantified the amount of PAR3 that colocalises with PIP2 and RAB5 positive particles (by using Manders coefficient / Coloc2 plugin in Fiji). We observed that around 50% of PAR3 co localise with PI(4,5)P2 and 50% co localise with RAB5. We do not know if it is the same 50%. We show in this study that RAB5 is as crucial to the posterior exclusion of PAR3 than the PI5Kinase SKTL. The endocytosis seems important to the exclusion however, we cannot conclude on a direct association of PAR3 with the budding vesicle or on a later association with RAB5 positive vesicles. We have added these news results in the Results section.

The RAB proteins have been tagged with YFP at their endogenous chromosomal loci (Dunst et al., 2015). Like described by Dunst et al., the fluorescence of RAB proteins is very weak in the oocyte and in order to observe it, we must use an anti-GFP antibody to enhance the detection. Unfortunately, this would prevent us from performing a colocalisation with PARGFP.

4) In Figure 3—figure supplement 1, how do we know the antibodies are specific? – the HRS antibody does not seem to detect vesicles.; and the syntaxin16 antibody decorates the plasma membrane but no detectable trans-Golgi.

The antibody anti-HRS and anti-syntaxin16 were now well referenced in the Materials and methods section. The anti-HRS antibody is a gift of H. Bellen and was published in “Hrs Regulates Endosome Membrane Invagination and Tyrosine Kinase Receptor Signaling in Drosophila” by (Lloyd et al., 2002). This antibody detects very well the vesicles in follicular cells but there are clearly less vesicles in the oocyte.

The anti-syntaxin16 stained Golgi and lysosome compartments (Akbar et al., 2009). From the stage 9, Golgi begins to accumulate in the subcortical region (Lee and Cooley, 2007). Nevertheless, now we show a magnification on cytoplasm where we can observe some vesicles.

5) The interaction between SKTL and Par3 is based only on over-expressed proteins. The blot needs the GFP-SKTl input and a better (GFP fusion) negative control. Also, I could not find data to support the contention that Par3 forms a tripartite complex with SKTL and PIP2, as is stated in subsection “SKTL and PAR3 act in synergy to form vesicular platforms”. The authors also suggest that Par3 could regulate SKTL activity, and that this would provide a positive feedback loop, but again without any supporting evidence.

We have added the input of the samples, and a control with a GFP alone, rather than a control without anti-GFP.

Concerning the second point, indeed we had no evidence of a tripartite complex. We quantified the colocalisation of PAR3 with PI(4,5)P2 or SKTL (Menders coefficient). Around 50% of PAR3 colocalise with PI(4,5)P2 or SKTL. Since we performed a colocalisation of PAR3, SKTL and PI(4,5)P2 and the result is presented in Figure 3G.

Finally, we have suppressed the positive feedback loop hypothesis of the text.

6) In Figure 5 it looks as though the mutant SKTL is expressed at lower levels than the WT. Could this, rather than loss of kinase activity, explain the lower GFP-Par3 in cytoplasmic spots? The colocalization with Rab5 and 11 is not convincing. Moreover, it looks as though the aAPM has disintegrated – is this not an issue with interpretation of the data?

The Myc-SKTL and Myc-SKTLDNRQ transgenes are inserted at the same site in the genome thanks to phiC31 integrase system (chromosome 3L, position 68A4). The expression of these two transgenes are thus of the same level. We have performed a western blot on ovary extracts and quantified with α-tubulin as loading control.

The results are presented in Author response image 1.

Author response image 1
Our results show that the mutant form is more stable than wild type.

Thus, the absences of effect of DNRQ form cannot be explain by protein instability.

https://doi.org/10.7554/eLife.40212.052

In order to clarify the results concerning the colocalisations, we have quantified the colocalisation of PAR3 with RAB5 and RAB11 vesicles. The Pearson’s coefficient was obtained by using the FiJi Macro Coloc2 on our images. The negative control was provided by quantifying colocalisation for the same images, but after rotation of one by 90 degrees, a condition in which only random colocalisation is observed. The results are now presented in the manuscript Figure 6A. We have also added, in the result part, the quantification of PAR3 that colocalises with RAB5 or RAB11 (Manders ‘s coefficient).

We agree with the reviewer on the difficulties to visualise the APM in Figure 5G. It is just because the plasma membrane is refolded and appears in XY plan. For clarity, this image has been removed with this new version.

7) Finally, I think it is premature to conclude much from the IKK-eta data – clearly knockdown generates some kind of cytoplasmic inclusion in the oocytes, that accumulate actin and Rab11, but it’s not clear what they are; and since Par3-GFP is still strongly accumulated all over the plasma membrane, including the APM (Figure 6A – I), I don't think that it is clear that IKKe is necessary for unloading it from microtubules.

We have reorganised the text concerning IKKe in the result section, we have simplified the conclusion and remove the interpretations. From our results, we can conclude that ikke knockdown leads to the accumulation of (PI,4,5)P2 et RAB11 enriched vesicles at the MT minus end and that the PI(4,5)P2 and microtubules are required to this accumulation.

Moreover, PAR distribution is strongly affected by this knockdown.

8) In general, the authors seem to interpret their data as if Par3 is an integral membrane protein that is being endocytosed and transported by the microtubule machinery, but of course it is not – it is a peripheral protein that associates with negatively-charged phospholipids in a highly dynamic way and is presumably able to diffuse through the cytoplasm between vesicles and plasma membrane.

We agree with reviewer and we have modified the Discussion section:

“Although PAR3 does not need a priori an association with membrane to be transported, we found it very frequently close to endocytic membranes. PAR3 on the endosome could be a passenger of the vesicles, which are then transported to the anterior. In parallel, PAR3 could also control its own association with vesicles. Indeed, we hypothesize PAR3 could recruit SKTL that in return produces PI(4,5)P2 that stabilises PAR3 to the membrane and recruits the vesicle budding machinery.” We do not exclude that PAR3 associates with vesicles by simple electrostatic interactions without going by endocytosis. However, the formation of the endocytosis vesicles seems to be important to the final distribution of PAR3.

Reviewer #3:

Jouette and colleagues have examined mechanisms that position the polarity protein Baz/Par-3 in a well-established model of cell polarization, the Drosophila oocyte. They examine a number of mechanisms involved in the cortical removal, sub-cellular trafficking and intracellular compartment release of Baz. The trafficking of Baz through the cytosol is poorly understood, and the results of this paper make significant advances in elucidating this process. However, major concerns should be addressed.

Major concerns

1) All of the Baz distributions are determined with an over-expressed Baz construct. Thus, it is unclear how relevant identified mechanisms are for endogenous levels of the protein. The newly identified mechanisms should be additionally evaluated with probes for endogenous Baz (using either Baz antibodies or an available Baz-GFP gene trap line).

To identify and to characterize the PAR3 regulation mechanism, firstly we have used a Drosophila strain that expresses PAR3-GFP on its own promoter and in absence of endogenous PAR3: BAC PAR3GFP = P[w+, FRT9-2]18E, f, baz[815.8], P{CaryP, PB[BAC Baz- sfGFP2]attP18} (Besson et al., 2015). The level of expression is clearly weaker than with UAS PAR3-GFP expression. We have not used thereafter this strain during this study because of the ubiquitous expression of PAR3. Indeed, the accumulation of PAR3 at the apex of somatic follicular cells around the oocyte masks the potential localisation of PAR3 at the plasma membrane of oocyte, the two membranes being very close. For the same raison, we did not employ PAR3-GFP trap line. Then, the use of antibody is not so easy in oocyte. The follicular cells layer is a natural barrier that prevents the antibody penetration. To stain PAR3, we must permeabilize the membrane with 0,3% Triton X-100 detergent, and the final staining is not enough reproducible to precise quantification.

To clarify this interesting point, we have added images of these experiments in the Figure 1—figure supplement 1A, and precise in the text why we do not continue with these transgenic lines.

In this study, we have used a UAS PAR3-GFP line, expressed with a line in which GAL4 is expressed under the control of maternal α tubulin promoter. Thus, in this condition, PAR3 is expressed only in germline cells and not in follicular cells. We have verified that PAR3GFP expression did not induce phenotype like nucleus mispositioning and we have controlled that the mechanisms described here was always saturable by a PAR3 expression increase. We agree with the reviewer that our conditions are not the best, but it is the one that we can use to answer to this precise question.

2) Connections between the mechanisms that remove Baz from the cortex are unclear. Possible connections should be investigated with the tools already used to determine if the mechanisms are independent or interdependent.

2.1) Abnormal Baz accumulations arise along the later cortex with dynein RNAi. Are these accumulations dependent on cortical removal steps mediated by Par-1, PIP2 or Rab5? Perturbations of these proteins in pair-wise combinations could reveal dependent or independent cortical removal mechanisms.

As suggested by the reviewer, we tried to combine the perturbations induced by these proteins on PAR3 localisation despite the genetic association difficulties.

First of all, to test whether PAR3 accumulation along the lateral cortex observed in dhc64c knockdown was dependent on PAR1, we expressed PAR3-GFP with RNAi PAR1 and RNAi DHC64c. The combined expression of these transgenes strongly affects the morphology and the development of the egg chamber. As it can be observed (Author response image 2), the oocyte in the egg chamber present an abnormal small size compared to the nurse cells. Moreover, this phenotype is also observed and seems stronger, when PAR3AA (a form non phosphorylable by PAR1) is overexpressed in dhc64c knock-down context. These kinds of phenotypes are usually related to a defect in the growth of the oocyte due to a transport failure from the nurse cells to the oocyte. Unfortunately, these interactions make it difficult to quantify PAR3 polarity during oogenesis and it is therefore difficult to conclude about the interdependence between these mechanisms.

We also tried to test the interdependence between DHC64c and SKTL, or RAB5. Sadly, the expression of RNAi DHC64c with either RAB5 DN (Dominant Negative) or in context sktl mutant, disorganise the structure of the egg chamber making the quantification impossible. To test the involvement of endocytosis in these processes, we also used the drug dynasore, which is an inhibitor of dynamin and thus of the endocytosis vesicle scission. However, this drug is associated with pleiotropic phenotypes in the oocyte and we were not able to answer this point to the reviewer.

2.2) Skittles overexpression is shown to increase the removal of Baz from the posterior domain. Does this removal require Par-1 or Rab-5? Again, combining the Skittles overexpression with the Par-1 RNAi or the Rab5DN could address how interdependent the removal mechanisms are.

We also tried to address the interdependence between SKTL and PAR1 or RAB5 for PAR3 posterior exclusion.

Firstly, we combined the expression of SKTL (UAS-myc-SKTL) with PAR1 knockdown (UAS RNAi PAR1) and we observed a genetic interaction between them concerning PAR3 posterior exclusion. We showed that SKTL overexpression is always able to exclude PAR3 of the posterior in PAR1 RNAi knockdown. The same observation was realised by using PAR3 AA mutant form (non phosphorylable by PAR1). This new data is presented Figure 4.

To observe the interdependence between SKTL and RAB5, we expressed PAR3-GFP with Myc-SKTL and RAB5DN (Dominant Negative). However, as illustrated in the image below, the expression of all these proteins induces precocious oogenesis defects. The PAR3 quantification was therefore not possible.

3) Since Baz becomes isotropic around the cortex with IKK-epsilon knock down, does IKK-epsilon also play a role in the removal of Baz from the posterior cortex? If IKK-epsilon only promoted delivery to the anterior cortex, then removal from the posterior would still be expected. Some assessment of IKK-epsilon's direct or indirect role at the posterior would be expected within the overall context of the paper.

The kinase IKKε has a role probably more complex that the one described in this work. It seems for example, to have a slight effect on PAR1 posterior localisation. Therefore, we have chosen to focus on the already described phenotype of IKKε, the ectopic accumulation of RAB11 cargo at the microtubule minus ends (Otani et al., 2011).

[Editors' note: the author responses to the re-review follow.]

This manuscript (an earlier version of which which has been previously reviewed by eLife but is now being treated as a new submission rather than a resubmission) provides evidence for a pathway that polarizes the distribution of Baz (Par-3) across the Drosophila oocyte, an important model of cell polarity. The model is an interesting and, if fully supported by the data, would be appropriate for publication in eLife. However, there are significant concerns that would need to be addressed before it could be accepted.

Essential revisions:

1) Generally, the quality of the biochemical evidence is not as strong as it should be. For example, in Figure 7E and F, the IPs of GFP-SKTL testing for pull-down of PAR3 and DLIC, the control was no addition of GFP antibody. Since addition of antibodies can increase non-specific binding, the better control is to IP an irrelevant GFP protein with the GFP antibody (as done in Figure 7D). Especially when the protein interaction they are trying to show is weak, good controls are required.

The immunoprecipitation experiment was reiterated with another GFP-protein in control (GFP-SNAP). The GFP-SNAP protein does not precipitate PAR3 or DLIC. The new IP blot is presented in Figure 7E and the charge control has been added in Figure 7—figure supplement 2.

Even in 7D, which does have an appropriate control, the co-IP is pretty weak. This does not invalidate it, but it should be acknowledged in the text.

This precision has been added in subsection “DLIC, PAR3 and SKTL are interacting partners”.

In the earlier version it was unclear that the 250kD band that was observed on western blots actually corresponded to Par3. The authors have used RNAi strains to address this point and the conclusion is a bit more convincing. It is still hard to fully appreciate, though. For instance, in Figure 7—figure supplement 2, a strain over-expressing Par3-GFP using the UAS-Gal4 driver is used. These embryos should be expressing a relatively high level of Par-GFP. In the total fraction from these embryos the most prominent band is around 100kD when probed with a Par3 antibody. However, this band is hard to detect with the GFP antibody. In fact, it is hard to see any bands in the GFP blot. Next, the sample was immunoprecipitated. It is unclear from the legend which antibody (Par3 or GFP) was used in the IP. In the immunoprecipitated fractions, the most prominent band is the 250kD band. The band is reduced in the strain expressing Par3 RNAi but it is not clear how it can be detected so prominently in the IP fraction when the same isoform is a minor species at best in the total fraction. Is the IP antibody somehow specifically precipitating an oligomeric form? The experiment with the endogenously expressed BAC-based Par3-GFP construct is more convincing.

2) Another major issue is the over-reliance on overexpressed Par3. Key results should be repeated to look at endogenous Par3 expression. For example, are the Par3-positive cytoplasmic particles, and effects on these particles, observed when Par3 is expressed by its own regulatory regions? Without this additional data, there is a risk that the effects they report are artifacts of overexpression.

To address this issue, we have used a Drosophila strain, PAR3-Trap, in which a GFP is inserted in frame in PAR3 (Januschke and Gonzales, 2010). We have added in a new figure (Figure 2—figure supplement 2) the illustration of PAR3 trap line that shows the presence of dotted structure in the oocyte cytoplasm. This observation was realized on previtellogenic stage because the accumulation of autofluorescent vitellus masks the presence of these structures thereafter. Furthermore, we have assayed the microtubule requirement for the cytoplasmic transport of these structures. We performed a colchicin treatment on PAR3 trap egg chambers. We then observed that PAR3 accumulates in numerous dotted structures in the cytoplasm, as observed previously with the PAR3-GFP overexpression in presence of colchicin. The expression of PAR3GFP in the follicular cells with this strain precludes the analysis of PAR3-GFP at the oocyte plasma membrane. These new results are presented in the Figure 2—figure supplement 2.

3) The results using a dominant negative rab5 should be confirmed with an shRNA construct (available from Bloomington). There is a risk that the DN might produce off target effects.

We have quantified PAR3 distribution in a RAB5 RNAi context. Among the twodrosophila transgenic lines available in Bloomington stock center, the BL34832 stock affects most strongly endocytosis during oogenesis according defect in yolk uptake (Compagnon et al., 2009). RAB5 RNAi alters PAR3 distribution in the same manner that RAB5 dominant negative but a little less strongly. This new result is present in Figure 5.

https://doi.org/10.7554/eLife.40212.057

Article and author information

Author details

  1. Julie Jouette

    Institut Jacques Monod, CNRS, UMR 7592, Paris Diderot University, Sorbonne Paris Cité, Paris, France
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Performed experiments and data analysis, Conducted data interpretation and writing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0115-0065
  2. Antoine Guichet

    Institut Jacques Monod, CNRS, UMR 7592, Paris Diderot University, Sorbonne Paris Cité, Paris, France
    Contribution
    Funding acquisition, Project administration, Writing—review and editing, Conducted data interpretation and writing
    For correspondence
    antoine.guichet@ijm.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7216-1944
  3. Sandra B Claret

    Institut Jacques Monod, CNRS, UMR 7592, Paris Diderot University, Sorbonne Paris Cité, Paris, France
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing, Performed experiments and data analysis, Conducted data interpretation and writing
    For correspondence
    sandra.claret@ijm.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7167-510X

Funding

Ministère de l’Enseignement Supérieur, de la Recherche Scientifique

  • Julie Jouette

Ligue Contre le Cancer

  • Julie Jouette

Fondation ARC pour la Recherche sur le Cancer (SLR20130607102)

  • Antoine Guichet

Ligue Contre le Cancer (RS14/75-58)

  • Antoine Guichet

Fondation ARC pour la Recherche sur le Cancer (PJA 20141201756)

  • Antoine Guichet

Fondation ARC pour la Recherche sur le Cancer (PJA 20161204931)

  • Antoine Guichet

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We acknowledge the ImagoSeine core facility of the Institut Jacques Monod, member of IBiSA and the France-BioImaging (ANR-10-INBS-04) infrastructure. We thank the proteomic platform of Institut Jacques Monod. We thank A Wodarz, J McDonald, D St Johnston, R Karess, F Schweisguth, M Sanial, and M Erdélyi for reagents. We thank F Bernard, C Jackson, JA Lepesant, and L Pintard for critical comments on the manuscript; C Seiler for technical assistance; L Gervais for directed mutagenesis of SKTLDNRQ; the Bloomington Stock Center for fly stocks; the Developmental Studies Hybridoma Bank for monoclonal antibodies; BestGene Inc Service for Drosophila embryo injections. This work was supported by the CNRS, by the ARC (grant SLR20130607102, grant PJA 20141201756 and PJA 20161204931), and by the ‘Ligue Contre le Cancer’ (grant RS14/75-58).

JJ was supported by a fellowship from the Ministry of Research and Technology (MRT) and by a fellowship from Fondation ‘Ligue Contre le Cancer’. The authors declare that they have no financial interest that might influence the results or interpretation of their manuscript.

Senior Editor

  1. Anna Akhmanova, Utrecht University, Netherlands

Reviewing Editor

  1. Matthew Freeman, University of Oxford, United Kingdom

Publication history

  1. Received: July 19, 2018
  2. Accepted: January 3, 2019
  3. Accepted Manuscript published: January 23, 2019 (version 1)
  4. Version of Record published: February 1, 2019 (version 2)
  5. Version of Record updated: February 1, 2019 (version 3)

Copyright

© 2019, Jouette et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 934
    Page views
  • 225
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    2. Developmental Biology
    Piyal Taru Das Gupta, Maithreyi Narasimha
    Research Article
    1. Cell Biology
    2. Computational and Systems Biology
    Simeon Leupold et al.
    Research Advance Updated