Drosophila CTP synthase regulates collective cell migration by controlling the polarized endocytic cycle

Cell migration is an important process involved in developmental events and pathologies, such as inflammation and cancer cell metastasis. The migration process is categorized into two types: individual and collective cell migration. Both types of migration involve directed cell migration, implying cell interaction with the environment that includes many mechanisms for sensing guidance signaling and depositing molecules in a polarized manner (Maritzen et al., 2015; Montell et al., 2012; Rørth, 2011). Here, we investigated how cytidine triphosphate synthase (CTPsyn) affected collective cell migration.

Drosophila border cells are used to study collective cell migration in vivo. The border cell cluster migrates during Drosophila oogenesis and is composed of two polar cells at the center surrounded by six to eight border cells. In the stage 8 egg chamber, the polar cells direct nearby follicle cells to form border cells through activation of JAK/STAT signaling (Montell et al., 2012; Silver and Montell, 2001). In the stage 9 egg chamber, the border cells begin migration from the anterior of the egg chamber to the junction of nurse cells and the oocyte. Once they reach the junction, the border cells migrate to meet the dorsal-anterior follicle cells in the stage 10 egg chamber to participate in micropyle formation, which is the path for sperm entry in fertilization (Prasad et al., 2015) (Fig. 1A). During this process, the extracellular signals are ligands of receptor tyrosine kinase (RTK) which act as guidance cues, such as ligands of epidermal growth factor receptor (EGFR) and PVR (the Drosophila homolog of platelet-derived growth factor receptor and vascular endothelial growth factor receptor) (Duchek and Rørth, 2001; Duchek et al., 2001; McDonald et al., 2006). These activated RTKs are restricted to the leading edge of the migratory border cell cluster through endocytic recycling cycles (Assaker et al., 2010; Janssens et al., 2010). The endocytic recycling cycle comes from the PVR signal that acts on Actin polymerization through Rac, which controls Actin as the track for vesicle movement (Ramel et al., 2013; Wan et al., 2013). This leads to amplification of the initial small difference between RTK signaling levels at the front and at the back (Montell et al., 2012; Rørth, 2011; Saadin and Starz-Gaiano, 2016). However, the detailed mechanism of polarized establishment is unclear in collective cell migration.

Phosphoinositol phosphate (PIP) plays important roles in many biological functions from invertebrates to vertebrates, such as cell polarity, cell migration, cell signaling transduction and vesicle trafficking (Schink et al., 2016). PIP is composed of a myo-inositol head group and acyl chain tails. CTPsyn is involved in PIP biosynthesis to produce CTP for forming CDP-diacylglycerol (CDP-DAG) with phosphatidic acid (PA) in de novo synthesis (Chang and Carman, 2008). Subsequently, phosphatidylinositol synthase (Pis), an evolutionarily conserved enzyme in PIP biosynthesis, produces phosphatidylinositol (PI), which contains seven derivates that are generated by PIP kinases phosphorylating the 3-, 4-, and 5- groups of the inositol ring individually or in various combinations. Phosphatidylinositol 4-kinase alpha (PI4KIIIα) phosphorylates the 4-group of the inositol ring of PI to form PI4P (Tan and Brill, 2014). PI4P 5-kinase Skittles (Sktl), a phosphatidylinositol 4-phosphate 5-kinase (PIP5K) homolog in Drosophila, further phosphorylates the 5-group of inositol rings to produce PI 4,5-bisphosphate [PI(4,5)P₂, or PIP₂]. PIP₂ plays a major regulatory role on the cell surface, for example being a precursor of PI(3,4,5)P₃ and an adaptor for membrane and cytosolic protein binding (e.g. N-WASP, PLCδ, Profilin, and RhoGAPs/GEFs). PIP₂ regulates a variety of cellular processes, including endocytosis, exocytosis, cell motility, adhesion and signal transduction (Di Paolo and De Camilli, 2006; Murray et al., 2012). For example, PIP₂/PIP₃ is enriched at the leading edge and directs cell movement of several cell types, including Dictyostelium amoebae, mammalian neutrophils and Drosophila hemocytes (Funamoto et al., 2002; Kölsch et al., 2007; Lacalle et al., 2007; Rickert et al., 2000; Ridley et al., 2003). CTPsyn reportedly affects the activity of PI synthase in yeast and regulates PIP₂ production in Drosophila germline cells (McDonough et al., 1995; Strochlic et al., 2014). However, the role of PIP₂ in collective cell migration remains elusive, and the role of CTPsyn in cell migration is still unknown.

In response to starvation, CTPsyn can assemble into filamentous structures which is well conserved across the species (Aughey et al., 2014; Calise et al., 2014; Carcamo et al., 2011; Ingerson-Mahar et al., 2010). We recently demonstrated that CTPsyn filaments are assembled along a cytokeratin network and histidine-mediated protein methylation promotes its formation in mammalian cancer cells (Chakraborty et al., 2020; Lin et al., 2018). Drosophila ovaries express three different isoforms of CTPsyn: isoform A is localized in the nucleus, isoform B is diffused in the cytosol and only isoform C can form the filament (Azzam and Liu, 2013). Based on our previous report, CTPsyn regulates S phase occurrence through its filamentous structure in Drosophila ovarian follicle cells, which is positively regulated by the E3 ubiquitin ligase Cbl (Wang et al., 2015). During initial border cell migration, the CTPsyn filamentous structure in anterior follicle cells is disassociated, while the epithelial cell type is transformed into migratory cells by JAK/STAT signaling activation. Therefore, we considered that CTPsyn may play another role in Drosophila border cells. Here, we found that CTPsyn is enriched at the outer cortical region of the border cell cluster together with Actin. Depletion of CTPsyn impaired the border cell cluster migration and affected the polarized distribution of PIP₂. Our results suggest that this could alter the PIP₂-involved endocytic recycling pathway and impair the activated RTK-polarized distribution in border cell clusters, which is important for guiding cell migration.

According to our previous study, CTPsyn filaments are present in every follicle cell of stage 2∼10A egg chambers and are involved in DNA endoreplication during stages 7∼10 (Wang et al., 2015). However, CTPsyn filamentous structures were absent in border cells after cell fate determination by JAK/STAT signaling. Here, we used the CTPsyn protein trap line in which GFP is trapped in-between the first and second exon of isoform C, to investigate the distribution of CTPsyn in border cell clusters (Azzam and Liu, 2013). The distribution of CTPsyn is well colocalized with Actin near the outer cortical region of migrating border cells, and especially at the leading edge (Fig. 1B-C). To investigate whether CTPsyn depletion can affect border cell migration, we expressed CTPsyn RNAi using Slbo-GAL4 in wild-type or CTPsyn^d06966 mutant background. CTPsyn^d06966 largely reduces the level of the filament-forming isoform C (Azzam and Liu, 2013). We found that with expression of CTPsyn RNAi in wild-type background, 65% of the border cell clusters migrated to the oocyte border by stage 10, according to the categories mentioned in Fig. 2A,B. However, the complete migratory phenotype further decreased by 32% when CTPsyn RNAi was expressed in the heterozygous CTPsyn^d06966 null mutant background (Fig. 2B). The enhanced migration defect in the CTPsyn mutant background indicated that it correlated with CTPsyn function (Fig. 2B-F).

Efficiency of CTPsyn RNAi was validated in follicle cells and nurse cells by immunostaining and western blotting, respectively (Fig. S1A-D). In addition, CTPsyn RNAi was driven with c306-GAL4 and Slbo-GAL4 at 29°C instead of 25°C, to increase the RNAi effect. Following this, only 39% and 38% of the border cell clusters, respectively, migrated completely (Fig. S1E-I). c306-GAL4 expresses earlier in border cells and usually leads to stronger phenotypes than Slbo-GAL4. Furthermore, when the efficacy of CTPsyn gene silencing was increased using Dicer, complete border cell cluster migration reduced to 32% (Fig. S1E,J,K). To obtain a loss-of-function phenotype instead of just reduction, we used a mosaic assay to generate the CTPsyn^d06966 mutant allele clone in the border cell cluster. Approximately 70% of mosaic border cell clusters with more than two mutant cells reduced migration to less than 50% (Fig. 2G-I; Fig. S2A-C), and less than 5% of egg chambers with mosaic clones completed the migration. Although, no anterior versus posteriors positional preference was detected for CTPsyn mutant clones (Fig. S2D) at stage 9 during migration, live imaging of migrating border cell clusters containing CTPsyn mutant clones showed disrupted rotation of border cells and reduced migration of the cluster (see Movie 1 for control, Movie 2 for CTPsyn mutant). Taken together, these data suggest that CTPsyn is required for border cell cluster migration.

A previous study reported that knockdown of CTPsyn in germline cells reduced the PIP₂ reporter signal in the plasma membrane and decreased membrane integrity (Strochlic et al., 2014). However, the roles of CTPsyn-mediated PIP₂ biosynthesis in border cells has not been investigated. Here, we examined the distribution of PIP₂ by expressing the UAS-PH-PLCδ-GFP reporter, a fusion protein consisting of the PH domain of phospholipase C (PLC) and GFP (Gervais et al., 2008; Varnai et al., 2002). In wild type, the PH-PLCδ-GFP signals were localized between border cell–border cell junctions and border cell–nurse cell interface, and particularly enriched at the tip of the leading edge during entire migratory processes until reaching the oocyte border by stage 10 (Fig. 3A-E′,G,G′). This unique localization in the border cell cluster is consistent with previous reports indicating that PIP₂ is located at the leading edge of human (mammalian) migrating cells and is required for migration function (Czech, 2000). These polarized distributions of PH-PLCδ-GFP signals at the tip of the leading edge were impaired in the CTPsyn-depleted border cell cluster, but the signals of border cell–border cell junctions were not altered (Fig. 3F,F′,H,H′). The ratio of PH-PLCδ-GFP signals between the leading and trailing edge was 2.4-fold in wild type but was reduced to 1.5-fold in CTPsyn-depleted border cell clusters, while the total level of PH-PLCδ-GFP showed no difference (Fig. 3I; Fig. S3A). Depletion of CTPsyn resulted in the loss of enrichment of PIP₂ (GFP-PH^PLC) at the leading edge of the border cell cluster, indicating that CTPsyn regulates PIP₂ localization in border cells.

Furthermore, we found that knockdown of CTPsyn in border cell clusters reduced Actin enrichment at the leading edge, as monitored in early stage 9 egg chambers (Fig. 3J-L). Given that PIP₂ is known to promote Actin assembly, reduction of PIP₂ signal at the leading edge caused by CTPsyn depletion could diminish the Actin enrichment (Janmey et al., 2018; Lassing and Lindberg, 1988).

We then further examined whether CTPsyn regulates border cell migration through PIP₂ production by genetic interaction approaches. Here, we knocked down the genes encoding PIP biosynthesis enzymes (Pis, PI4KIIIα and Sktl, shown in Fig. 4A) in border cells under wild-type or CTPsyn^d06966 heterozygous mutant backgrounds. Knockdown of pis by itself did not show migration defects, however in the CTPsyn^d06966/+ background, 58% of stage 10 egg chambers showed border cell migration defects (Fig. 4B-F). Knockdown of PI4KIIIα in the wild-type background resulted in 31% of egg chambers with migration defects, which was enhanced in CTPsyn^d06966/+ background by 31%. Although knockdown of sktl displayed border cell migration defects, depletion of Sktl did not show genetic interaction with CTPsyn (Fig. 4B,G,H). In humans PI5P can be converted to PIP₂; however, in Drosophila this pathway is still unknown (Fig. 4A). These synergistic genetic interactions indicate that CTPsyn cooperates with Pis and PI4KIIIα in PIP biosynthesis to regulate border cell migration.

PIP₂ is involved in recruiting the correct coat and adaptor proteins during endocytosis, and PI phosphatases like synaptojanin and src-homology 2 containing 5-phosphatase (SHIP2) terminate this recruitment and subsequently PIP₂ is hydrolyzed to PI3P (Wallroth and Haucke, 2018). Therefore, we examined whether PI3P is also affected under CTPsyn depletion conditions. The distribution of PI3P in CTPsyn depletion of border cells was investigated by expressing a UAS-2-FYVE-GFP (2×FYVE) marker, as PI3P serves as a docking site on early endosomes for recruiting specific proteins containing the FYVE zinc-finger domain (Marat and Haucke, 2016). The 2×FYVE was asymmetrically localized at the front of the cluster in wild-type border cells as previously reported, but there were weak signals in cells at the back of the cluster during initial migration (Devergne et al., 2007) (Fig. 5A-A″). In contrast, in the CTPsyn-depleted border cell cluster, 2×FYVE signals lost their asymmetrical distribution and were not enriched in the leading cells (Fig. 5B-B″). The average ratio of 2×FYVE signals between the front and back was 4.3-fold in the wild type but was reduced to 1.2-fold in depletion of the CTPsyn border cell cluster (Fig. 5C). This result indicated that depletion of CTPsyn also resulted in an impaired polarized distribution of PI3P at the leading edge of the border cell cluster, suggesting that trafficking of the early endosome was affected in the border cell cluster.

Rab11 and exocyst components (Sec3, Sec5, Sec15) are reportedly required for border cell migration through their roles in endocytotic recycling trafficking of activated RTKs (Assaker et al., 2010; Wan et al., 2013). Although, under CTPsyn^d06966/+ mutant background, depletion of either Rab11 or Rab7 did not significantly suppress the border cell migration (Fig. S4A), we found that the asymmetric distribution of Rab11 at the leading edge of the border cell cluster was significantly affected, but the total level of Rab11 showed no difference (Fig. S3B) (Assaker et al., 2010; Janssens et al., 2010; Wan et al., 2013) (Fig. 5D-F). In the wild type, Rab11-GFP accumulated at the front of the border cell cluster, and an average of 1.8-fold difference in intensity was measured by comparing the signal at the front and the back of the border cell cluster (Fig. 5F). The intensity difference was reduced to 1.2-fold under depletion of CTPsyn (Fig. 5F). As PI4P plays a role in recruiting Rab11-mediated exocysts, the above data suggests that, following CTPsyn depletion, PI4P distribution in border cell clusters might also be affected (Ketel et al., 2016). In addition, knockdown of sec3 and sec15 but not sec5 in the CTPsyn^d06966/+ mutant background affected border cell migration (Fig. 5G; Fig. S4B). Given that RNAi knockdown of sec3 and sec15 did not cause a migration defect, this synergetic genetic interaction between CTPsyn and sec3/sec15 indicate that CTPsyn depletion in border cells might affect the polarized distribution of PIPs, thereby affecting the recycling/endocytic pathways during border cell cluster migration.

The asymmetrical enrichment of activated RTK at the leading edge in the border cell cluster is generated by a polarized endocytic recycling cycle. Cbl, an E3 ligase, is involved in internalizing activated RTKs during border cell migration (Jékely et al., 2005). Knockdown of Cbl in CTPsyn^d06966/+ background showed a synergetic interaction in border cell migration defect phenotypes in 40% of stage 10 egg chambers (Fig. 5G). Thus, we further investigated the distribution of activated RTKs in border cell clusters by immunostaining with an anti-pTyr antibody, which has been validated as a reliable local readout of endogenous RTK activity (Jékely et al., 2005). In wild-type cells, pTyr staining showed restriction in front of the leading edge during migration initiation, as previously reported (Fig. 6A). However, depletion of CTPsyn impaired the asymmetrical localization of pTyr signals (Fig. 6B). The pTyr signal ratio between the front and back was 1.5:1 in wild type but was reduced to be 1:1 with depletion of the CTPsyn border cell cluster (Fig. 6I). Furthermore, the enrichment of pTyr signal is also reduced in CTPsyn^d06966 mosaics (Fig. 6J-K″″). Our results suggest that depletion of CTPsyn abolished the polarized distribution of RTK signaling, indicating that CTPsyn is necessary for regulating the restriction of activated RTKs at the leading edge.

To determine whether CTPsyn is involved in the general sorting of membrane proteins, we visualized adherens junction protein Ecad (also known as Shg) and its partner, Arm, by immunostaining. The patterns of Ecad and Arm were not altered in the border cell cluster with depletion of CTPsyn (Fig. 6C-F; Fig. S2). As the function of PIPs is also required for the polarized architecture of cells, we considered the distribution of cell polarity proteins aPKC and Dlg. The location of aPKC and Dlg was not altered in CTPsyn-depleted border cells or CTPsyn mutant clones (Fig. 6G,H; Fig. S5), suggesting that the failure of restricting activated RTKs did not result from cell polarity disruption.

In summary, CTPsyn regulated the polarized distribution of activated RTK at the leading edge through regulating the polarized endocytic cycle via PIP₂ metabolism in collective cell migration.

It is well known that CTPsyn is essential for cell growth by providing building blocks such as DNA, RNA and phospholipids, and the expression of CTPsyn is frequently elevated in cancers (Kizaki et al., 1980; van den Berg et al., 1993; Williams et al., 1978). CTPsyn expression level can regulate the synthesis of CDP-DAG, which is required for the production of phosphoinositides (McDonough et al., 1995). Here, we reveal a role for CTPsyn in collective cell migration through regulation of the asymmetrical distribution of PIP₂ at the leading edge of migrating border cell cluster, suggesting that CTPsyn might have additional roles in tumorigenesis (Fig. 6L). PIP₂ enrichment at the leading edge is essential for directing migration in a single cell (Czech, 2000). PIP₂ is spatially controlled by the activities of metabolized kinases such as phosphoinositide 4-kinase (PI4K) and PIP5K (Clayton et al., 2013; van den Bout and Divecha, 2009). Our results also showed that during collective cell migration, PIP₂ is enriched at the leading edge (Fig. 3). Knockdown of PIP₂ biosynthesis enzymes in the CTPsyn mutant background further suppressed the border cell migration, supporting the idea of PIP₂ regulation by CTPsyn (Fig. 4). This finding raised an interesting question regarding how a biosynthesis enzyme regulates the spatial distribution of phospholipids. Purine biosynthetic enzymes are compartmentalized at the leading edge in motile renal cell carcinoma (Wolfe et al., 2019). Intriguingly, we found that CTPsyn is also enriched at the cortex of border cell clusters, suggesting a spatial proximity of PI biosynthesis and asymmetrical enrichment of PIP₂ (Fig. 1). One possible explanation for our results is that local high levels of PIs generated by enrichment of CTPsyn is essential for subsequent functioning of phosphoinositide-metabolizing kinases that are also spatiotemporally distributed; however, this hypothesis needs further experiments to verify. Interestingly, live images showed an impairment in mutant CTPsyn border cell rotation during migration at stage 9, indicating the coordination between border cells was disrupted. However, the molecular mechanism of CTPsyn in coordinating border cell rotation needs further investigation.

In epithelial cells, PIP₂ is enriched in the microdomain of the plasma membrane (PM) named the lipid raft, which is characterized by cholesterol enrichment and resistance to Triton X-100 (Johnson et al., 2008; Munro, 2003; Wang and Richards, 2012). This enrichment of PIP₂ on the PM promotes the localization of EGFR in the lipid raft microdomain, which may upregulate PI3K signaling to activate Rac/Cdc42 for Actin polymerization (Abd Halim et al., 2015). PIP₂ is also known to regulate Actin dynamics through interactions with Actin associated proteins (Janmey et al., 2018). Our results showed that depletion of CTPsyn resulted in loss of PIP₂ and Actin enrichment at the leading edge (Fig. 3), indicating that CTPsyn may control border cell migration through regulation of Actin dynamics by the levels of PIP₂. Furthermore, loss of function of CTPsyn in the border cell cluster abolished the pTyr signaling enrichment in the leading edge (Fig. 6B,I-K″″), suggesting that CTPsyn may be required for the formation of raft microdomains of the PM for enriching activated RTKs through the promotion of PIP₂ production. Recently, a rapid dynamic for PIP₂ turnover by PLC and PI4 kinase pathways was discovered in the raft microdomain (Myeong et al., 2021), suggesting that the CTPsyn-mediated biosynthesis of PIP₂ might be crucial for raft domain formation and signaling.

During collective border cell migration, spatial restriction of RTK at the leading edge is maintained by a positive-feedback loop through RTK downstream Rac signaling, which induces polarized distribution of recycling endosome and exocyst, involving Cbl, Rab5, Rab11 and exocyst components (Assaker et al., 2010; Jékely et al., 2005; Wan et al., 2013). The migratory defect of the CTPsyn-depleted border cell cluster was enhanced by interfering with the activity of Cbl and exocyst components (Fig. 5G), suggesting that CTPsyn function is involved in the endocytic recycling cycle to promote directional migration. Furthermore, the polarized distribution of Rab11-GFP and FYVE signals was also abolished in the CTPsyn-depleted border cell cluster (Fig. 5A-F). As asymmetric distribution of RTK was significantly reduced with CTPsyn depletion (Fig. 6B,I-K″″), it is possible that RTK-regulated high levels of recycling vesicles and asymmetrically distributed exocyst were also disturbed. Alternatively, PIP₂ at the PM is required for nucleation of endocytic clathrin-coated pits, and PI(3,4)P₂ is the intermediate for the conversion to the PI3P-containing endosome (Posor et al., 2013). Thus, asymmetric PIP₂ enrichment may regulate the asymmetric distribution of endosomal PI3P through endocytosis. Furthermore, the conversion of PI3P to PI4P by MTM1 and PI4KIIα is required for exocyst-dependent endosomal exocytosis due to the binding of PI4P to Rab11 (Ketel et al., 2016; Wallroth and Haucke, 2018). These studies in mammals support our finding that CTPsyn depletion disrupted the asymmetric distribution of Rab11 and PI3P, which may be through the loss of PIP₂ asymmetrical distribution. However, the possibility that CTPsyn might regulate polarized trafficking is not excluded from our study. In summary, we provide here the link between CTPsyn function and PIP₂ production in asymmetrical endocytic and recycling activity during collective cell migration. This may lead us to investigate further the roles of CTPsyn in cancer metastasis and find alternative therapeutic treatments.

All flies were maintained at 25°C. The following fly lines from Bloomington Drosophila Stock Center (BDSC) were used: wild type (Oregon-R), UAS-PHPLCδ-GFP, UAS-GFP-myc-2×FYVE and UAS-Sktl^RNAi (TRIP line JF02796). The following fly lines from the Vienna Drosophila Resource Center (VDRC) were used: CTRL^GD60000 (CTRL^RNAi), CTPsyn^GD12759, CTPsyn^GD12762, PI4KIIIα^GD15993, Pis^GD11852, Cbl^GD22335, Sec3^GD35806, Sec5^GD28873 and Sec15^GD35161. To drive expression in border cells, Slbo-GAL4, c306-GAL4 and e22c-GAL4 was used. To drive expression in germ cells nanos-GAL4 was used. CTPsyn mutant stock (UAS-CD8::GFP, CTPsyn^d06966/TM3Ser) was obtained from BDSC. Wild type (e22cFLP/+;Ubi-GFP FRT80B/FRT80B) and mutant mosaic clones (e22cFLP/+; Ubi-GFP FRT80B/CTPsyn^d06966, FRT80B) were generated using Flp-FRT method. To induce clones, flies were incubated at 29°C. GFP-CTP synthase (CA06746 and CA07332) protein traps were from Ji-Long Liu (Shanghai Tech University, Shanghai, China). For genetic interaction experiments, all crosses were maintained at 29°C unless otherwise mentioned.

Ovary dissection was performed in Schneider's medium within 5 min and then fixed in 4% paraformaldehyde for 20 min. After washes in PBT (0.3% Triton X-100 in PBS), ovaries were stained with primary antibodies overnight at 4°C. The primary antibodies used for immunostaining included: rabbit anti-CTPsyn (1:300; y-88, sc-134457; Santa Cruz Biotechnology), mouse anti-Ecad [1:50; Developmental Studies Hybridoma Bank (DSHB)], mouse anti-arm (1:50; 7A1; DSHB), mouse anti-phosphotyrosine (1:200; Millipore), rabbit anti-PKCζ (1:200; C-20; Santa Cruz Biotechnology) and mouse anti-Dlg (1:50; 4F3; DSHB). Secondary antibodies included Alexa Fluor 488 or 546 anti-mouse/rabbit/rat antibodies [1:1000; A11001 (anti-mouse), A11008 (anti-rabbit), A11030 (anti-mouse), A11035 (anti-rabbit) and A21050 (anti-mouse); Thermo Fisher Scientific]. To visualize DNA, we added 0.5 mg/ml DAPI (Sigma-Aldrich) for 5 min at room temperature. To visualize Actin, we used Phalloidin (1:100; Molecular Probes).

Stage 10 egg chambers were selected, and the position of border cell clusters was analyzed. We defined stage 10 as when the oocyte spanned the posterior half of the egg chamber. As an index for migration, these stage 10 egg chambers were categorized based on the location of the border cell cluster as depicted in Fig. 2A.

Images from fixed tissues were acquired using a ZEISS LSM 780 inverted confocal microscope. The quantification methods are similar to those previously described (Zhang et al., 2011). For measurement of the front/back (leading/lagging) ratios, an area around the leading edge of the cluster but excluding polar cells (labeled by anti-FasIII antibodies; 1:20, 7G10, DSHB) was chosen as the front region, and an area including the lagging end but excluding polar cells was chosen as the back region. Fluorescence intensity and area were measured using Fiji software for each region. The graphs were plotted with Prism 6. For experimental data, a Student's two-tailed unpaired t-test was used for the analysis.

The time-lapse image was performed as previously described (Chang et al., 2018). In brief, female flies were dissected in S2 live media which comprised S2 Cell medium, fetal bovine serum (Gibco), insulin (Sigma-Aldrich) and female fly extract. Egg chambers were cultured in ibidi glass-bottomed dishes after dissection. The images were taken every 5 min using a Zeiss Observer D1 microscope. All live images were processed by Axio Vision SE64Rel.4.8.2 software (Media Cybernetics) and the NIS-Elements AR Analysis 4.50.00 64-bit software (Nikon).

We are grateful to the Microscopy Center of Chang Gung University, for assistance with image capture. We thank Mark A. Peifer and Robert J. Duronio for critical comments on the manuscript. We also thank the Fly Core in Taiwan, the DSHB, the BDSC and VDRC for providing reagents.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200190.reviewer-comments.pdf.