EGFR signaling controls directionality of epithelial multilayer formation upon loss of cell polarity Authors: Aiguo Tian, Xian‐Feng Wang, Yuting Xu, Virginia Morejon, Yi‐Chun Huang, Chidi Nwapuda, Wu‐Min Deng Published: 2023 ABSTRACT AbstractApical‐basal polarity is maintained by distinct protein complexes that reside in membrane junctions, and polarity loss in monolayered epithelial cells can lead to formation of multilayers, cell extrusion, and/or malignant overgrowth. Yet, how polarity loss cooperates with intrinsic signals to control directional invasion toward neighboring epithelial cells remains elusive. Using the Drosophila ovarian follicular epithelium as a model, we found that posterior follicle cells with loss of lethal giant larvae (lgl) or Discs large (Dlg) accumulate apically toward germline cells, whereas cells with loss of Bazooka (Baz) or atypical protein kinase C (aPKC) expand toward the basal side of wildtype neighbors. Further studies revealed that these distinct multilayering patterns in the follicular epithelium were determined by epidermal growth factor receptor (EGFR) signaling and its downstream target Pointed, a zinc‐finger transcription factor. Additionally, we identified Rho kinase as a Pointed target that regulates formation of distinct multilayering patterns. These findings provide insight into how cell polarity genes and receptor tyrosine kinase signaling interact to govern epithelial cell organization and directional growth that contribute to epithelial tumor formation.Keywords: cell polarity, Drosophila, EGFR signaling, follicular epithelium, growth directionsSubject Categories: Cancer; Cell Adhesion, Polarity & Cytoskeleton; Development Introduction Epithelial cells, which form continuous sheets to provide protective, absorptive, and secretory functions, are polarized and characterized by cell membranes with apical and basolateral domains separated by adherens and tight junctions. These different domains are defined by specific protein complexes, including the Scribble (Scrib) complex located at the septate junctions and regulates basolateral membrane identity, the Par and the Crumbs complexes that are localized at the tight junctions to regulate apical membrane identity (Bilder et al, 2000, 2003; Hong et al, 2001; Zulueta‐Coarasa & Rosenblatt, 2021). The genes coding for these protein complexes (namely, polarity genes) are conserved across metazoans and they are the gatekeeper to prevent malignant growth, and disruption of their function has major implications in development and tumorigenesis (Coradini et al, 2011; Royer & Lu, 2011). The behavior of polarity‐deficient epithelial cells can be affected by locally available signals. In the hinge area of Drosophila wing imaginal discs, loss of lethal (2) giant larvae (lgl) or Scrib leads to apical delamination and tumor formation, and the process requires local Jak–STAT signaling (Slattum & Rosenblatt, 2014; Tamori et al, 2016; Richardson & Portela, 2018; Nanavati et al, 2020). Similarly, in the follicular epithelium (FE) during oogenesis, the polarity‐deficient follicle cells form multilayers primarily at the terminal regions of the egg chambers (Abdelilah‐Seyfried et al, 2003; Bilder, 2004; Chatterjee et al, 2022). The polarity‐disrupted follicle cells invade into the germline region (Goode & Perrimon, 1997; Abdelilah‐Seyfried et al, 2003; Bilder, 2004; Chatterjee et al, 2022), and the intensity of invasion is enhanced by Keap1‐Nrf2 signaling (Chatterjee et al, 2022). These findings indicate a significant contribution of tissue microenvironment in malignant transformation of polarity‐loss epithelial cells. The FE in Drosophila is maintained by a population of follicle stem cells located in the germarium. The descendant follicle cells experience a series of signaling events that pattern them along the axis of the egg chamber and control their differentiation and maturation. The signaling events include the Jak–STAT signaling regulated development of the polar cells, Notch‐induced transition from mitotic to endocycle, and EGFR signaling‐induced posterior follicle cell fate (Gonzalez‐Reyes et al, 1995; Roth et al, 1995; Deng et al, 2001; Lopez‐Schier & St Johnston, 2001; Bastock & St Johnston, 2008; Antel & Inaba, 2020). EGFR signaling is activated by interaction between the TGF‐α‐like ligand Gurken (Grk) in the oocyte and the epidermal growth factor receptor (EGFR, also called Torpedo) in the abutting follicle cells. When the cell polarity genes are mutated in the FE, the terminal regions of egg chambers produce the multilayering phenotype (Abdelilah‐Seyfried et al, 2003; Li et al, 2011; Shahab et al, 2015; Jevitt et al, 2021), which appears to be more prominent in the posterior follicle cells (PFCs) (Shahab et al, 2015). In this study, we found that PFCs with loss of the polarity genes lgl or Dlg accumulate toward the apical side of their wildtype neighbors, whereas loss of Baz or aPKC caused basal expansion of mutant cells. The direction of accumulation or expansion is dependent on the levels of EGFR signaling. Further studies with single‐cell RNA sequencing (scRNA‐seq) and genetic epistasis analyses revealed that EGFR signaling determines formation of distinct multilayering patterns through the Pnt‐Rok‐Sqh regulatory axis. In summary, we show that cell polarity acts as a gatekeeper to prevent epithelial cells from invading their neighbors and that the tissue microenvironment regulates the growth direction of the polarity‐deficient cells. Results Distinct multilayering patterns caused by loss of apical‐basal polarity in follicle cells The FE is a monolayered epithelium that shows a characteristic apical‐basal polarity (referred to as cell polarity hereafter) with its apical surface facing the germline cells (Fig 1A–A″″ and E–E″′). When follicle cell (FC) clones mutant for lgl (lgl 4 ) were generated using the FLP/FRT technique (Xu & Rubin, 1993), multilayers were formed at the terminal areas of egg chambers (Appendix Fig S1; 100%, n = 90), as reported previously (Abdelilah‐Seyfried et al, 2003; Li et al, 2011; Shahab et al, 2015; Jevitt et al, 2021). Focusing on egg chambers during mid‐oogenesis (stage 7–10), we noticed that the multilayers contained both mutant lgl and wildtype cells (Fig 1B–C″″″). In the majority of posterior follicle cell (PFC) clones carrying multilayers (92.1%, n = 63, Fig 1D), lgl mutant cells were located at the apical side (Fig 1B–B″″, GFP−, red arrows) relative to their neighboring wildtype FCs (Fig 1B–B″″, GFP+, green arrows), and these adjacent wildtype FCs were insulated from the oocyte by apically located mutant cells. In only a small fraction of clones with the multilayers (7.9%, n = 63, Fig 1D), lgl mutant cells (Fig 1C–C″″, GFP−, red arrows) were located at the basal side of their wildtype neighbors (Fig 1C–C″″, GFP+, green arrows). To further observe the connection between apically located mutant cells and wildtype neighbors, we performed the Z‐stack analysis and marked the cytoskeleton with fluorescent Phalloidin. The mutant cells of lgl were apically located (Fig EV1A and A′, GFP−, red arrows, Movie EV1) and showed clear connection with wildtype neighbors (Fig EV1B–C′, green arrows). These findings suggest that lgl mutant cells tend to be apically accumulated within the multilayers containing both mutant and wildtype cells. Loss of lgl in PFCs produces the apical location pattern A–A″′Wildtype PFC clones without GFP showed the normal monolayer.B–B″′Mutant PFCs of lgl 4 without GFP (red arrows) accumulated at the apical side and insulated the wildtype neighbors from the oocyte (green arrows), which generates the apical location pattern.C–C″′Mutant PFCs of lgl 4 without GFP (red arrows) are localized at the basal side of the neighboring wildtype cells (green arrows) to form the basal location pattern.DThe frequency of each location pattern was quantified (n (total clones with apical or basal location) = 63).E–E″′Wildtype PFC clones without GFP showed the normal monolayer.F–F″′At the edge of multilayers with small clones, mutant PFCs (red arrows) of lgl 4 locate toward the apical side of their neighboring wildtype cells (green arrows).G–G″′At the edge of multilayers with small clones, mutant PFCs (red arrows) of lgl 4 locate toward the basal side of their neighboring wildtype cells (green arrows).HThe frequency of clones with apical and basal location was quantified (n (total clones with apical or basal location) = 77). A–A″′Wildtype PFC clones without GFP showed the normal monolayer. Wildtype PFC clones without GFP showed the normal monolayer. B–B″′Mutant PFCs of lgl 4 without GFP (red arrows) accumulated at the apical side and insulated the wildtype neighbors from the oocyte (green arrows), which generates the apical location pattern. Mutant PFCs of lgl 4 without GFP (red arrows) accumulated at the apical side and insulated the wildtype neighbors from the oocyte (green arrows), which generates the apical location pattern. C–C″′Mutant PFCs of lgl 4 without GFP (red arrows) are localized at the basal side of the neighboring wildtype cells (green arrows) to form the basal location pattern. Mutant PFCs of lgl 4 without GFP (red arrows) are localized at the basal side of the neighboring wildtype cells (green arrows) to form the basal location pattern. DThe frequency of each location pattern was quantified (n (total clones with apical or basal location) = 63). The frequency of each location pattern was quantified (n (total clones with apical or basal location) = 63). E–E″′Wildtype PFC clones without GFP showed the normal monolayer. Wildtype PFC clones without GFP showed the normal monolayer. F–F″′At the edge of multilayers with small clones, mutant PFCs (red arrows) of lgl 4 locate toward the apical side of their neighboring wildtype cells (green arrows). At the edge of multilayers with small clones, mutant PFCs (red arrows) of lgl 4 locate toward the apical side of their neighboring wildtype cells (green arrows). G–G″′At the edge of multilayers with small clones, mutant PFCs (red arrows) of lgl 4 locate toward the basal side of their neighboring wildtype cells (green arrows). At the edge of multilayers with small clones, mutant PFCs (red arrows) of lgl 4 locate toward the basal side of their neighboring wildtype cells (green arrows). HThe frequency of clones with apical and basal location was quantified (n (total clones with apical or basal location) = 77). The frequency of clones with apical and basal location was quantified (n (total clones with apical or basal location) = 77). Data information: The relative location of GFP+ cells (green) and GFP− cells (red) are illustrated in A″″, B″″, C″′, E″′, F″′, and G″′. a: apical side; b: basal side; O: oocyte. Phall: Phalloidin. Scale bars represent 20 μm. Three independent experiments were performed, and error bars are ±SEM (D and H). Source data are available online for this figure. The lgl mutant cells are apically located compared with their wildtype neighbors A–C′The representative confocal Z‐stack with nine planes of the lgl 4 mutant clones (A–A′), and the first (B‐B′) and fourth planes of the clone (C–C′). Red arrows show the mutant cells and green arrows show the wildtype neighboring cells.D–G″At the edge of multilayers with small clones or large clones, mutant PFCs (red arrows) of Dlg m52 locate toward the apical side of their neighboring wildtype cells (D‐D″, F‐F″) or toward the basal side of their neighboring wildtype cells (E–E″, G–G″). A–C′The representative confocal Z‐stack with nine planes of the lgl 4 mutant clones (A–A′), and the first (B‐B′) and fourth planes of the clone (C–C′). Red arrows show the mutant cells and green arrows show the wildtype neighboring cells. The representative confocal Z‐stack with nine planes of the lgl 4 mutant clones (A–A′), and the first (B‐B′) and fourth planes of the clone (C–C′). Red arrows show the mutant cells and green arrows show the wildtype neighboring cells. D–G″At the edge of multilayers with small clones or large clones, mutant PFCs (red arrows) of Dlg m52 locate toward the apical side of their neighboring wildtype cells (D‐D″, F‐F″) or toward the basal side of their neighboring wildtype cells (E–E″, G–G″). At the edge of multilayers with small clones or large clones, mutant PFCs (red arrows) of Dlg m52 locate toward the apical side of their neighboring wildtype cells (D‐D″, F‐F″) or toward the basal side of their neighboring wildtype cells (E–E″, G–G″). Data information: GFP− marked lgl mutant cells and fluorescent Phalloidin marks the actin cytoskeleton. Scale bars represent 20 μm. The apical accumulation pattern suggests that lgl mutant cells could expand apically in the area between the wildtype neighbors and the oocyte. To corroborate this, we closely examined egg chambers bearing small lgl 4 PFC clones for the relative position between mutant cells and neighboring wildtype cells at the clone boundary. In the majority of these mosaic FE (93.1%, n = 77, Fig 1H and F–F″′ compared with Fig 1G–G″′), lgl 4 mutant cells were found to be apically located at the interface (Fig 1F–F″′, GFP−, red arrows) and separated their wildtype neighbors from the oocyte (Fig 1F–F″′, GFP+, green arrows). Their relative positions at the interface indicate that lgl mutant cells show a preferential apical accumulation to form multilayers at the apical side of adjacent wildtype cells in the FE. The analysis of a distinct mutant allele lgl 27S3 validated apical location of lgl mutant cells in mosaic FE containing both large (Fig EV2A, 91.4%, n = 70) and small clones (Fig EV2B, 92.5%, n = 80). To determine whether the preferential apical accumulation is specific to lgl mutant cells, we examined mosaic egg chambers containing PFC clones of Dlg, which belongs to the same Scrib complex as Lgl, and found that mutant cells of two Dlg alleles (Dlg m52 and Dlg m30 ) showed similar apical accumulation in large PFC clones (Fig 2A, Dlg m52 , 93.3%, n = 75; Figs EV1F–F″ vs. G‐G″, and EV2C, Dlg m30 , 92.3%%, n = 65). In small clones, we noticed that Dlg mutant cells predominantly locate toward the apical interface of the neighboring wildtype cells (Fig 2B, Dlg m52 , 95%, n = 81, Fig EV1D–D″ compared with Figs EV1E–E″ and EV2D, Dlg m30 , 92.5%, n = 65). The quantification of each location pattern in clones with loss of lgl or Dlg A–DThe frequency of each location pattern in large clones was quantified for lgl 27S3 (A, n = 70) and Dlg m30 (C, n = 65). The frequency of small clones with apical or basal location was quantified for lgl 27S3 (B, n = 80) and Dlg m30 (D, n = 65). A–DThe frequency of each location pattern in large clones was quantified for lgl 27S3 (A, n = 70) and Dlg m30 (C, n = 65). The frequency of small clones with apical or basal location was quantified for lgl 27S3 (B, n = 80) and Dlg m30 (D, n = 65). The frequency of each location pattern in large clones was quantified for lgl 27S3 (A, n = 70) and Dlg m30 (C, n = 65). The frequency of small clones with apical or basal location was quantified for lgl 27S3 (B, n = 80) and Dlg m30 (D, n = 65). Data information: Three independent experiments were performed. Error bars are ±SEM. The quantification of each location pattern in clones with loss of cell polarity genes AThe frequency of each location pattern in large clones was quantified for Dlg m52 (n = 75).BThe frequency of each location pattern in small clones was quantified for Dlg m52 (n = 81).CThe frequency of each location pattern in large clones was quantified for aPKC (n = 81).DThe frequency of each location pattern in small clones was quantified for aPKC (n = 105).EThe frequency of each location pattern in large clones was quantified for Baz (n = 78).FThe frequency of each location pattern in small clones was quantified for Baz (n = 123).G–G″″′The mitotic cell cycle marker phosphor‐Histone H3 (PH3) was detected in FCs before stage 7 (red arrowheads), but not afterward in the wildtype egg chambers.H–H″′PH3 was detected in lgl 4 mutant clones after stage 7 (arrows).I–I″′PH3 was detected in Baz mutant clones after stage 7 (arrows).JThe quantification of PH3‐positive cells in control, lgl, and Baz mutant clones at stage 8. AThe frequency of each location pattern in large clones was quantified for Dlg m52 (n = 75). The frequency of each location pattern in large clones was quantified for Dlg m52 (n = 75). BThe frequency of each location pattern in small clones was quantified for Dlg m52 (n = 81). The frequency of each location pattern in small clones was quantified for Dlg m52 (n = 81). CThe frequency of each location pattern in large clones was quantified for aPKC (n = 81). The frequency of each location pattern in large clones was quantified for aPKC (n = 81). DThe frequency of each location pattern in small clones was quantified for aPKC (n = 105). The frequency of each location pattern in small clones was quantified for aPKC (n = 105). EThe frequency of each location pattern in large clones was quantified for Baz (n = 78). The frequency of each location pattern in large clones was quantified for Baz (n = 78). FThe frequency of each location pattern in small clones was quantified for Baz (n = 123). The frequency of each location pattern in small clones was quantified for Baz (n = 123). G–G″″′The mitotic cell cycle marker phosphor‐Histone H3 (PH3) was detected in FCs before stage 7 (red arrowheads), but not afterward in the wildtype egg chambers. The mitotic cell cycle marker phosphor‐Histone H3 (PH3) was detected in FCs before stage 7 (red arrowheads), but not afterward in the wildtype egg chambers. H–H″′PH3 was detected in lgl 4 mutant clones after stage 7 (arrows). PH3 was detected in lgl 4 mutant clones after stage 7 (arrows). I–I″′PH3 was detected in Baz mutant clones after stage 7 (arrows). PH3 was detected in Baz mutant clones after stage 7 (arrows). JThe quantification of PH3‐positive cells in control, lgl, and Baz mutant clones at stage 8. The quantification of PH3‐positive cells in control, lgl, and Baz mutant clones at stage 8. Data information: Scale bars represent 20 μm. Three independent experiments were performed (A–F, and J). Error bars are ±SEM. ***P < 0.001 (Student's t‐test). Source data are available online for this figure. During the development of egg chambers, the wildtype FCs enter the endoreplication cycle after stage 7, leading them to cease division and no longer increase in number (Deng et al, 2001) (Fig 2G–G″″′). In contrast, the lgl/Dlg mutant FCs exhibited continuous division, as indicated by expression of the mitotic cell marker PH3 (Phospho‐Histone H3) (Fig 2H–H″′). This continued proliferation of lgl/Dlg mutant cells results in an increase in their numbers at the apical side, leading to the insulation of wildtype neighbors from the oocyte. We further examined the mosaic egg chambers bearing aPKC (aPKC k06403 ) or Baz (Baz Xi106 ) PFC clones. Baz and aPKC are components of the Par complex that regulate apical membrane identity. Contrary to lgl and Dlg, mosaic egg chambers carrying large aPKC or Baz PFC clones showed a preponderance of basal expansion of mutant cells (aPKC k06403 : 80%, n = 81, Figs 2C and EV3I–I″ vs. K–K″; Baz Xi106 , 85%, n = 78, Figs 2E and EV3H–H″ compared with Fig EV3J–J″). Small PFC clones revealed more clearly that the aPKC and Baz mutant cells showed a strong preference to accumulate at the basal side of the nearby wildtype cells (aPKC k06403 : 83.2%, n = 105, Figs 2D and EV3E–E″ vs. G–G″, Baz Xi106 , 86.7%, n = 123, Figs 2F and EV3D–D″ compared with Fig EV3F–F″). Additionally, the Z‐stack analysis showed that the mutant cells of Baz were basally located (Fig EV3A and A′, GFP−, red arrows) and showed clear connection with wildtype neighbors (Fig EV3B–C′, green arrows). Further analysis of expression of PH3 in Baz mutant clones showed that mutant cells can continue to divide after stage 7 (Fig 2I–I″′ and J). Therefore, aPKC or Baz mutant cells show basal growth and form multilayers at the basal side of wildtype neighbors. The Baz mutant cells are basally located compared with their wildtype neighbors A–C′The representative confocal Z‐stack with six planes of the Baz xi106 mutant clones (A–A′), and the first (B–B′) and fourth planes of the clone (C–C′). Red arrows show the mutant cells and green arrows show the wildtype neighboring cells.D–K″At the edge of multilayers with small clones or large clones, mutant PFCs (yellow arrows) of Baz xi106 and aPKC K06043 locate toward the basal side of their neighboring wildtype cells (green arrows) (D–E″, H–I″) or toward the apical side of their neighboring wildtype cells (green arrows) (F–G″, J–K″). A–C′The representative confocal Z‐stack with six planes of the Baz xi106 mutant clones (A–A′), and the first (B–B′) and fourth planes of the clone (C–C′). Red arrows show the mutant cells and green arrows show the wildtype neighboring cells. The representative confocal Z‐stack with six planes of the Baz xi106 mutant clones (A–A′), and the first (B–B′) and fourth planes of the clone (C–C′). Red arrows show the mutant cells and green arrows show the wildtype neighboring cells. D–K″At the edge of multilayers with small clones or large clones, mutant PFCs (yellow arrows) of Baz xi106 and aPKC K06043 locate toward the basal side of their neighboring wildtype cells (green arrows) (D–E″, H–I″) or toward the apical side of their neighboring wildtype cells (green arrows) (F–G″, J–K″). At the edge of multilayers with small clones or large clones, mutant PFCs (yellow arrows) of Baz xi106 and aPKC K06043 locate toward the basal side of their neighboring wildtype cells (green arrows) (D–E″, H–I″) or toward the apical side of their neighboring wildtype cells (green arrows) (F–G″, J–K″). Data information: GFP− marked Baz mutant cells and fluorescent Phalloidin marks the actin cytoskeleton. Scale bars represent 20 μm. Taken together, these observations indicate that FCs with loss of polarity genes from different complexes can form distinct multilayering patterns in the FE. Downregulation of EGFR signaling is necessary for lgl/Dlg‐loss‐induced apical growth As EGFR signaling is activated in the PFCs to induce a PFC fate (Gonzalez‐Reyes et al, 1995; Roth et al, 1995), we asked whether its activation is involved in the regulation of multilayering patterns of polarity‐deficient cells. To this end, we examined the expression of pointed (pnt)‐lacZ, an EGFR signaling readout (Morimoto et al, 1996), and found that it was significantly reduced in lgl 4 PFC clones (Fig 3B–B″ vs. A–A″, red arrows, n = 67), indicating downregulation of EGFR signaling in lgl PFC clones. Additionally, downregulation of pnt‐lacZ was rescued by overexpression of an active form of EGFR (EGFR A887T ) (Fig EV4B–B″′ vs. A–A″′, n = 25). Loss of EGFR signaling in PFCs alone was not, however, sufficient to induce multilayering as loss‐of‐function (LOF) of EGFR PFC clones generated with EGFR f24 or EGFR‐RNAi did not show an obvious defect in monolayered epithelial organization (Gonzalez‐Reyes et al, 1995; Roth et al, 1995) (Fig EV5B–B′: EGFR f2 , GFP−, 100%, n = 94; C–C′: EGFR‐RNAi, GFP+, 100%, n = 112, compared with Control (A–A:, 100%, n = 65)). In addition, the EGFR signaling activity in EGFR f2 was confirmed by pnt‐lacZ expression (Fig EV5E vs. D, n = 19). Downregulation of EGFR signaling is required for apical accumulation A–B″The EGFR signaling reporter pnt–lacZ, which was highly expressed in PFCs (A–A″), was significantly reduced in lgl 4 mutant PFCs (GFP−; B–B″, red arrows), whereas neighboring wildtype cells showed normal expression (white arrows).C–C″′The MARCM PFC clones (GFP+) with UAS–EGFR A887T expression showed the monolayer.D–E″′Apical accumulation that was induced in lgl 4 MARCM PFC clones (GFP+) (D–D″′) was suppressed by activated EGFR signaling with UAS–EGFR A887T to form basal expansion (E–E″, white arrows, E″′, black arrow).FThe frequency of apical accumulation or basal expansion was quantified for indicated genotypes (n = 64). A–B″The EGFR signaling reporter pnt–lacZ, which was highly expressed in PFCs (A–A″), was significantly reduced in lgl 4 mutant PFCs (GFP−; B–B″, red arrows), whereas neighboring wildtype cells showed normal expression (white arrows). The EGFR signaling reporter pnt–lacZ, which was highly expressed in PFCs (A–A″), was significantly reduced in lgl 4 mutant PFCs (GFP−; B–B″, red arrows), whereas neighboring wildtype cells showed normal expression (white arrows). C–C″′The MARCM PFC clones (GFP+) with UAS–EGFR A887T expression showed the monolayer. The MARCM PFC clones (GFP+) with UAS–EGFR A887T expression showed the monolayer. D–E″′Apical accumulation that was induced in lgl 4 MARCM PFC clones (GFP+) (D–D″′) was suppressed by activated EGFR signaling with UAS–EGFR A887T to form basal expansion (E–E″, white arrows, E″′, black arrow). Apical accumulation that was induced in lgl 4 MARCM PFC clones (GFP+) (D–D″′) was suppressed by activated EGFR signaling with UAS–EGFR A887T to form basal expansion (E–E″, white arrows, E″′, black arrow). FThe frequency of apical accumulation or basal expansion was quantified for indicated genotypes (n = 64). The frequency of apical accumulation or basal expansion was quantified for indicated genotypes (n = 64). Data information: The relative location between GFP+ cells (green) and GFP− cells (red) are illustrated in C″′, D″′ and E″′. a: apical side; b: basal side; O: oocyte. Phall: Phalloidin. Scale bars represent 20 μm. Three independent experiments were performed (F). Error bars are ± SEM. ***P < 0.001 (Student's t‐test). Source data are available online for this figure. Expression of EGFR A887T can rescue downregulation of pnt in lgl mutant clones A–B″′The EGFR signaling reporter pnt–lacZ was significantly reduced in lgl 4 mutant cells (red arrows, A–A″′), but was rescued by expression of EGFR A887T (red arrows, B–B″′). Scale bars represent 20 μm. Red arrows show the mutant cells. A–B″′The EGFR signaling reporter pnt–lacZ was significantly reduced in lgl 4 mutant cells (red arrows, A–A″′), but was rescued by expression of EGFR A887T (red arrows, B–B″′). Scale bars represent 20 μm. Red arrows show the mutant cells. The EGFR signaling reporter pnt–lacZ was significantly reduced in lgl 4 mutant cells (red arrows, A–A″′), but was rescued by expression of EGFR A887T (red arrows, B–B″′). Scale bars represent 20 μm. Red arrows show the mutant cells. Loss of EGFR signaling in the FE did not induce multilayering A–C′Loss of EGFR signaling by mutation (B–B′) or RNAi knockdown (C–C′) produced monolayered follicle cells, similar to that in the wildtype FE (A–A′). The GFP− clones are outlined with dashed white lines.D–E″The EGFR signaling reporter pnt–lacZ was downregulated in EGFR f24 mutant cells (white arrows, E–E″) compared with that in the control (D–D″). A–C′Loss of EGFR signaling by mutation (B–B′) or RNAi knockdown (C–C′) produced monolayered follicle cells, similar to that in the wildtype FE (A–A′). The GFP− clones are outlined with dashed white lines. Loss of EGFR signaling by mutation (B–B′) or RNAi knockdown (C–C′) produced monolayered follicle cells, similar to that in the wildtype FE (A–A′). The GFP− clones are outlined with dashed white lines. D–E″The EGFR signaling reporter pnt–lacZ was downregulated in EGFR f24 mutant cells (white arrows, E–E″) compared with that in the control (D–D″). The EGFR signaling reporter pnt–lacZ was downregulated in EGFR f24 mutant cells (white arrows, E–E″) compared with that in the control (D–D″). Data information: Scale bars represent 20 μm. To determine whether downregulation of EGFR signaling is necessary for apical growth of lgl mutant FCs, we re‐activated EGFR signaling by mis‐expressing a constitutively active form of EGFR (EGFR A887T ) in lgl 4 PFC clones (referred as lgl 4 ‐EGFR A887T clones) using the mosaic analysis with a repressible cell marker (MARCM) technique (Lee & Luo, 1999). As a control, follicle cells with clonal overexpression of EGFR A887T alone maintained a monolayered epithelia and showed no defect in the epithelial organization (GFP+, Fig 3C–C″′, 100%, n = 65). In contrast, activating EGFR signaling in lgl 4 PFC clones (namely, lgl 4 ‐EGFR A887T ) suppressed the apical accumulation phenotype of lgl 4 mutant cells (Fig 3D–D″′). Instead, the lgl 4 ‐EGFR A887T cells were found over the basal side of their wildtype neighbors (Fig 3E–E″′ and F). These findings suggest that activated EGFR signaling promotes basal growth of polarity‐deficient lgl mutant FCs. The EGFR activity promotes basal growth induced by loss of aPKC/Baz in the FE To determine whether EGFR signaling is also involved in regulating basal expansion of the aPKC or Baz mutant cells, we examined expression of pnt‐lacZ in aPKC mutant clones and found that its expression remained high in aPKC k06403 mutant PFCs adjacent to the oocyte, suggesting that EGFR signaling is likely intact in these cells (Fig 4A–A″′, white arrows). In aPKC mutant cells that were located more basally and did not maintain contact with the oocyte, pnt‐lacZ appeared to be either diminished or lost (Fig 4A–A″′, yellow arrowheads), probably because the oocyte‐expressed ligand Grk (Gonzalez‐Reyes et al, 1995; Roth et al, 1995) could no longer reach them. Basal expansion of aPKC mutant FCs is regulated by EGFR signaling A–A″′The EGFR signaling reporter pnt‐lacZ remained highly expressed in aPKC k06403 mutant cells (white arrow) and in neighboring wildtype cells (green arrows) but was downregulated once these cells (yellow arrowhead) lost contact with the oocyte.B–B″′ aPKC MARCM PFC clones (GFP+) showed basal expansion (white arrows).C–C″′The MARCM PFC clones (GFP+) with UAS–EGFR–RNAi expression showed the monolayer.D–D″′The expression of UAS–EGFR–RNAi in aPKC k06403 cells (GFP+) switched basal expansion in aPKC k06403 PFC clones to apical expansion (red arrows). The blue arrows show the neighboring wildtype cells.E–E″′The MARCM PFC clones (GFP+) with UAS–EGFR A887T expression showed the monolayer.F–F″′ UAS‐EGFR A887T expression (GFP+) in aPKC k06403 PFC clones enhanced basal expansion (white arrows, black arrow in F″′).G, HThe frequency of apical accumulation or basal expansion was quantified for indicated genotypes, n = 60 for each genotype. Error bars are ±SEM. ***P < 0.001.IActivation of EGFR signaling with UAS‐EGFR A887T at the lateral region (white arrow) did not induce multilayers.J aPKC k06403 mutant cells did not induce basal expansion and multilayers at the lateral region (white arrow).KActivation of EGFR signaling with UAS‐EGFR A887T in aPKC k06403 FC clones induced the multilayered phenotype and produced basal expansion at the lateral regions (yellow arrows). The red arrows show the wildtype neighbors.LThe frequency of clones with multilayers at the lateral region of egg chambers, n = 95 for each genotype. A–A″′The EGFR signaling reporter pnt‐lacZ remained highly expressed in aPKC k06403 mutant cells (white arrow) and in neighboring wildtype cells (green arrows) but was downregulated once these cells (yellow arrowhead) lost contact with the oocyte. The EGFR signaling reporter pnt‐lacZ remained highly expressed in aPKC k06403 mutant cells (white arrow) and in neighboring wildtype cells (green arrows) but was downregulated once these cells (yellow arrowhead) lost contact with the oocyte. B–B″′ aPKC MARCM PFC clones (GFP+) showed basal expansion (white arrows). aPKC MARCM PFC clones (GFP+) showed basal expansion (white arrows). C–C″′The MARCM PFC clones (GFP+) with UAS–EGFR–RNAi expression showed the monolayer. The MARCM PFC clones (GFP+) with UAS–EGFR–RNAi expression showed the monolayer. D–D″′The expression of UAS–EGFR–RNAi in aPKC k06403 cells (GFP+) switched basal expansion in aPKC k06403 PFC clones to apical expansion (red arrows). The blue arrows show the neighboring wildtype cells. The expression of UAS–EGFR–RNAi in aPKC k06403 cells (GFP+) switched basal expansion in aPKC k06403 PFC clones to apical expansion (red arrows). The blue arrows show the neighboring wildtype cells. E–E″′The MARCM PFC clones (GFP+) with UAS–EGFR A887T expression showed the monolayer. The MARCM PFC clones (GFP+) with UAS–EGFR A887T expression showed the monolayer. F–F″′ UAS‐EGFR A887T expression (GFP+) in aPKC k06403 PFC clones enhanced basal expansion (white arrows, black arrow in F″′). UAS‐EGFR A887T expression (GFP+) in aPKC k06403 PFC clones enhanced basal expansion (white arrows, black arrow in F″′). G, HThe frequency of apical accumulation or basal expansion was quantified for indicated genotypes, n = 60 for each genotype. Error bars are ±SEM. ***P < 0.001. The frequency of apical accumulation or basal expansion was quantified for indicated genotypes, n = 60 for each genotype. Error bars are ±SEM. ***P < 0.001. IActivation of EGFR signaling with UAS‐EGFR A887T at the lateral region (white arrow) did not induce multilayers. Activation of EGFR signaling with UAS‐EGFR A887T at the lateral region (white arrow) did not induce multilayers. J aPKC k06403 mutant cells did not induce basal expansion and multilayers at the lateral region (white arrow). aPKC k06403 mutant cells did not induce basal expansion and multilayers at the lateral region (white arrow). KActivation of EGFR signaling with UAS‐EGFR A887T in aPKC k06403 FC clones induced the multilayered phenotype and produced basal expansion at the lateral regions (yellow arrows). The red arrows show the wildtype neighbors. Activation of EGFR signaling with UAS‐EGFR A887T in aPKC k06403 FC clones induced the multilayered phenotype and produced basal expansion at the lateral regions (yellow arrows). The red arrows show the wildtype neighbors. LThe frequency of clones with multilayers at the lateral region of egg chambers, n = 95 for each genotype. The frequency of clones with multilayers at the lateral region of egg chambers, n = 95 for each genotype. Data information: a: apical side; b: basal side; O: oocyte. Phall: Phalloidin. Scale bars represent 20 μm. Three independent experiments were performed (G, H, L). Error bars are ±SEM. ***P < 0.001 (Student's t‐test). Source data are available online for this figure. When EGFR was knocked down with UAS‐EGFR‐RNAi in aPKC PFC clones (namely, aPKC‐EGFR RNAi ) by means of the MARCM system, the frequency of basal expansion of polarity‐deficient cells was markedly reduced (50% in aPKC‐EGFR RNAi , n = 60 vs. 80.1% in aPKC, n = 60), whereas the frequency of apical accumulation was dramatically increased (50% in aPKC‐EGFR RNAi , n = 60 vs. 19.9% in aPKC; n = 60) (Figs 4D–D″′ vs. B–B″′ and G). In contrast, the control FE with loss of EGFR signaling alone did not show apical or basal multilayering (Fig 4C–C″′). These results are consistent with the observations in lgl mosaic FE that EGFR signaling promotes basal growth of polarity‐deficient FCs. Furthermore, upregulating EGFR signaling by expressing EGFR A887T , which did not cause multilayering by itself (Fig 4E–E″′), in aPKC (named as aPKC‐EGFR A887T ) PFC clones resulted in an enhanced multilayering phenotype with greater number of mutant cells accumulating at the basal side of adjacent wildtype cells (Fig 4F–F″, white arrows, Fig 4F″′, black arrows), and increased percentage of PFC clones with basal buildup (94.5% in aPKC‐EGFR A887T vs. 80.1% in aPKC, n = 60 in each genotype; Fig 4F–F″′ vs. B–B″′ and H). Moreover, we found that activating EGFR signaling in aPKC k06403 FC clones located at the middle section of the egg chamber, where EGFR signaling is not activated and multilayers do not normally form, resulted in multilayers in 88% clones (n = 95) with basal expansion of mutant cells (Fig 4K, yellow arrows, compared with Fig 4I with EGFR A887T expression and Fig 4J with aPKC clones at the middle section (white arrows), Fig 4L), further suggesting that EGFR signaling promotes basal expansion of polarity‐deficient cells in the FE. The effect of EGFR signaling on lgl and aPKC FCs prompted us to examine the relationship between lgl/Dlg and aPKC/Baz complexes in regulating formation of the multilayering patterns. To this end, we performed genetic epistatic analysis with the MARCM system and quantified the frequency of apical or basal growth of lgl 4 PFCs with or without aPKC knockdown, and of aPKC k06403 PFCs with or without lgl knockdown. The results revealed that lgl knockdown in aPKC k06403 PFC clones substantially reduced the frequency of basal growth (Appendix Fig S2, 55% in aPKC k06403 with lgl‐RNAi vs. 80.1% in aPKC k06403 , n = 60 in each genotype). In contrast, aPKC knockdown in lgl 4 PFC clones did not change the frequency of apical growth (91.2% in lgl 4 with aPKC‐RNAi, n = 85 vs. 92% in lgl 4 alone, n = 75), thereby indicating that lgl is epistatic to aPKC in regulating the multilayering patterns. This epistatic relationship is consistent with the different levels of EGFR signaling found in lgl and aPKC mutant PFCs, i.e., EGFR activity is absent in lgl but present in some aPKC mutant PFCs, and EGFR signaling promotes basal expansion of polarity‐deficient cells. scRNA‐seq analysis identified cell clusters enriched with Rok expression To determine how EGFR signaling regulates the growth direction, we performed single‐cell RNA sequencing (scRNA‐seq) analysis on ovaries bearing MARCM clones of aPKC k06403 with EGFR A887T overexpression (aPKC‐EGFR A887T ) and the wildtype control. To compare cell populations that were present in the wildtype and aPKC‐EGFR A887T datasets, the two datasets were merged into an integrated dataset, which was ultimately divided into 42 clusters (Fig 5A). Based on the expression of GFP, Act5C, and EGFR (Fig 5B–G), 11 clusters from aPKC‐EGFR A887T mutant cells were identified (clusters 14, 15, 33, 34, 35, 37, 38, 39, 40, 41, and 42; Fig 5A, red arrows and green arrows). Among these clusters, six (clusters 37, 38, 39, 40, 41, and 42) were identified as specific clusters containing only aPKC‐EGFR A887T cells (red arrows in Fig 5A). In addition, all 36 clusters except for the six specific ones were annotated on the basis of expression of marker genes (Fig 5A and Appendix Fig S3) identified from the previous scRNA‐seq analysis of the wildtype egg chamber (Jevitt et al, 2020). Upregulation of Rok expression in aPKC‐EGFR A887T clones was identified by scRNA‐seq analysis AHigh‐quality cells from wildtype (Con) and aPKC‐EGFR A887T were grouped into 42 clusters, which were labeled according to the expression of marker genes.B, CAccording to expression of GFP, aPKC‐EGFR A887T mutant clones were identified in A (red arrows and green arrowheads). The red arrows indicate the specific clusters for aPKC‐EGFR A887T cells.D, EThe expression of Act5c was used as the control.F, GEGFR expression in each cluster was shown.H, ICells in clusters 38 and 39 (blue arrows) show enriched expression of Rok (I, blue arrows, compared with H in the wildtype).J Rok expression was measured by qRT–PCR in wildtype, lgl knockdown, and aPKC knockdown FCs (n = 3). AHigh‐quality cells from wildtype (Con) and aPKC‐EGFR A887T were grouped into 42 clusters, which were labeled according to the expression of marker genes. High‐quality cells from wildtype (Con) and aPKC‐EGFR A887T were grouped into 42 clusters, which were labeled according to the expression of marker genes. B, CAccording to expression of GFP, aPKC‐EGFR A887T mutant clones were identified in A (red arrows and green arrowheads). The red arrows indicate the specific clusters for aPKC‐EGFR A887T cells. According to expression of GFP, aPKC‐EGFR A887T mutant clones were identified in A (red arrows and green arrowheads). The red arrows indicate the specific clusters for aPKC‐EGFR A887T cells. D, EThe expression of Act5c was used as the control. The expression of Act5c was used as the control. F, GEGFR expression in each cluster was shown. EGFR expression in each cluster was shown. H, ICells in clusters 38 and 39 (blue arrows) show enriched expression of Rok (I, blue arrows, compared with H in the wildtype). Cells in clusters 38 and 39 (blue arrows) show enriched expression of Rok (I, blue arrows, compared with H in the wildtype). J Rok expression was measured by qRT–PCR in wildtype, lgl knockdown, and aPKC knockdown FCs (n = 3). Rok expression was measured by qRT–PCR in wildtype, lgl knockdown, and aPKC knockdown FCs (n = 3). Data information: Three independent experiments were performed (J). Error bars are ±SEM. **P < 0.01, and *P < 0.05 (Student's t‐test). Source data are available online for this figure. Among the aPKC‐EGFR A887T ‐specific clusters, clusters 38 and 39 showed enrichment of genes such as Rho kinase (Rok) (Fig 5I, blue arrows, compared with Fig 5H), pointed (pnt, a downstream gene of EGFR signaling), cappuccino (capu), Rho1, myospheroid (mys), and inflated (if). Rok, which phosphorylates Spaghetti squash (Sqh), the regulatory myosin light chain (RMLC) of nonmuscle myosin II in Drosophila (Edwards & Kiehart, 1996; Wang & Riechmann, 2007), is involved in the remodeling of actin/actomyosin cytoskeleton and regulation of cell extrusion/delamination (Fernandez et al, 2007; Cram, 2014). Our further qRT–PCR analysis with aPKC‐RNAi knockdown in FCs confirmed that Rok expression was upregulated (Fig 5J). In addition, we examined Rok expression in lgl mutant cells with lgl‐RNAi and found that Rok expression was downregulated (Fig 5J). The Rok kinase activity can be detected by an antibody against phospho‐specific RMLC (pRMLC), and the level of pRMLC was used as the readout for the Rok activity (Wang & Riechmann, 2007). In wildtype follicle cells, pRMLC is enriched at the apical cortex (Fig 6A–A″, n = 35), which was confirmed by sqh mutant clones (Appendix Fig S4). To determine whether EGFR signaling regulates Rok activity in follicle cells, we examined the level of pRMLC in aPKC mutant cells. We found that the intensity of the pRMLC in aPKC mutant cells is comparable to that in wildtype cells (Fig 6B–B″ and I). However, pRMLC in aPKC mutant cells was distributed through the entire cortex (Fig 6B–B″, n = 40). Moreover, when EGFR A887T was expressed in aPKC mutant cells, we observed a significant upregulation of the pRMLC level in the entire cortex (Fig 6C–C″, n = 35 clones, Fig 6I). These results suggest that activation of EGFR signaling upregulates the pRMLC level. In support of this conclusion, we found that the pRMLC level was reduced in lgl mutant cells where EGFR activity was decreased (Fig 6D–D″, n = 40, Fig 6J). This decrease of pRMLC was rescued by EGFR A887T overexpression in lgl mutant cells (Fig 6E–E″, n = 38, Fig 6J). In addition, we examined the level of pRMLC when EGFR A887T or EGFR‐RNAi was expressed in follicle cells and found that pRMLC was upregulated by EGFR A887T (Fig 6F–F″, n = 34, Fig 6K), but downregulated by EGFR‐RNAi expression (Fig 6G–G″, n = 31, Fig 6K). As a control, expression of the dominant‐negative Rok in FCs can reduce pRMLC level (Fig 6H–H″, n = 17, Fig 6K). Together, these results suggest that Rok and Sqh/RLMC are critical targets of EGFR signaling in FCs. The phosphorylation of Sqh by Rok was regulated by EGFR signaling A–A″The phosphorylation of Sqh by Rok was detected by an antibody against pRMLC (red) at the apical cortex of FCs (white arrows).B–B″pRMLC staining was detected in both the apical and basal cortex (white arrows) in aPKC mutant PFCs.C–C″ aPKC mutant PFCs with EGFR A887T overexpression showed increase pRMLC signal in the cortex.D–D″The pRMLC signal was reduced in the cortex of lgl mutant PFCs (white arrows).E–E″The pRMLC level in lgl mutant PFCs was upregulated by EGFR A887T overexpression (white arrows).F–H″The level of pRMLC (red) in EGFR A887T (F–F″, white arrows), EGFR‐RNAi (G–G″, white arrows) and Rok DN expressing clones (H–H″).IThe quantification of pRMLC intensity in the control, aPKC mutant cells, and aPKC mutant cells with EGFR A887T (n = 3).JThe quantification of pRMLC intensity in the control, lgl 4 mutant cells, and lgl 4 mutant cells with EGFR A887T (n = 3).KThe quantification of pRMLC intensity in the control, EGFR A887T , EGFR‐RNAi, and Rok DN mutant cells (n = 3). A–A″The phosphorylation of Sqh by Rok was detected by an antibody against pRMLC (red) at the apical cortex of FCs (white arrows). The phosphorylation of Sqh by Rok was detected by an antibody against pRMLC (red) at the apical cortex of FCs (white arrows). B–B″pRMLC staining was detected in both the apical and basal cortex (white arrows) in aPKC mutant PFCs. pRMLC staining was detected in both the apical and basal cortex (white arrows) in aPKC mutant PFCs. C–C″ aPKC mutant PFCs with EGFR A887T overexpression showed increase pRMLC signal in the cortex. aPKC mutant PFCs with EGFR A887T overexpression showed increase pRMLC signal in the cortex. D–D″The pRMLC signal was reduced in the cortex of lgl mutant PFCs (white arrows). The pRMLC signal was reduced in the cortex of lgl mutant PFCs (white arrows). E–E″The pRMLC level in lgl mutant PFCs was upregulated by EGFR A887T overexpression (white arrows). The pRMLC level in lgl mutant PFCs was upregulated by EGFR A887T overexpression (white arrows). F–H″The level of pRMLC (red) in EGFR A887T (F–F″, white arrows), EGFR‐RNAi (G–G″, white arrows) and Rok DN expressing clones (H–H″). The level of pRMLC (red) in EGFR A887T (F–F″, white arrows), EGFR‐RNAi (G–G″, white arrows) and Rok DN expressing clones (H–H″). IThe quantification of pRMLC intensity in the control, aPKC mutant cells, and aPKC mutant cells with EGFR A887T (n = 3). The quantification of pRMLC intensity in the control, aPKC mutant cells, and aPKC mutant cells with EGFR A887T (n = 3). JThe quantification of pRMLC intensity in the control, lgl 4 mutant cells, and lgl 4 mutant cells with EGFR A887T (n = 3). The quantification of pRMLC intensity in the control, lgl 4 mutant cells, and lgl 4 mutant cells with EGFR A887T (n = 3). KThe quantification of pRMLC intensity in the control, EGFR A887T , EGFR‐RNAi, and Rok DN mutant cells (n = 3). The quantification of pRMLC intensity in the control, EGFR A887T , EGFR‐RNAi, and Rok DN mutant cells (n = 3). Data information: a: apical side; b: basal side; O: oocyte. Phall: Phalloidin. Scale bars represent 20 μm. Three independent experiments were performed (I, J, K). Error bars are ±SEM. *P < 0.05 (Student's t‐test). Source data are available online for this figure. Rok and RMLC (sqh) regulate the direction of multilayer formation in FE To determine whether different levels of pRMLC in lgl and aPKC mutant cells control apical or basal growth of FCs with loss of polarity genes, we lowered the pRMLC level in aPKC mutant cells by overexpressing a dominant‐negative and kinase‐dead form of Rok (Rok DN ) or a nonphosphorylatable form of sqh (sqh DN ). As expected, overexpression of either UAS‐Rok DN or UAS‐sqh DN in aPKC k06403 PFC clones reduced the frequency of basal growth (86.7% in aPKC k06403 vs. 63.9% in aPKC k06403 ‐Rok DN , and 67.2% in aPKC k06403 ‐sqh DN ; n = 60 in each genotype) and increased the frequency of apical growth (13.3% in aPKC k06403 vs. 36.1% in aPKC k06403 ‐Rok DN , and 32.8% in aPKC k06403 ‐sqh DN ; n = 60 in each genotype) (Fig 7A–C″ and D). Additionally, overexpression of UAS‐Rok DN suppressed the basal growth of aPKC mutant cells with EGFR A887T expression (Fig 7L and K–K″ vs. I–J″, n = 35). As the control, overexpression of UAS‐Rok DN or UAS‐sqh DN in follicle cells with MARCM did not induce the multilayering phenotypes (Appendix Fig S5B and D). Therefore, downregulation of Rok‐Sqh activity promotes a switch from basal to apical growth of polarity‐deficient FCs. Directional growth is regulated by the Rok‐Sqh axis A–A″Loss of aPKC‐induced basal growth (red arrows).B–C″Overexpression of Rok DN (B–B″) or sqh DN (C–C″) in aPKC mutant PFCs induced apical growth (red arrows). White arrows indicated the wildtype neighbors.E–E″ lgl mutant cells showed apical growth (red arrows). White arrows indicated the wildtype neighbors.F–G″Overexpression of Rok CA (F–F″) or sqh CA (G–G″) in lgl 4 PFC clones switched apical growth to basal growth (red arrows). White arrows indicated the wildtype neighbors.I–I″Loss of aPKC‐induced basal growth (red arrows).J–J″Expression of EGFR A887T enhances basal growth (red arrows).K–K″Expression of Rok DN repressed basal growth in aPKC mutant clones with EGFR A887T (red arrows).D, H, LThe frequency of apical growth or basal growth was quantified for indicated genotypes (n = 60). Scale bars represent 20 μm. A–A″Loss of aPKC‐induced basal growth (red arrows). Loss of aPKC‐induced basal growth (red arrows). B–C″Overexpression of Rok DN (B–B″) or sqh DN (C–C″) in aPKC mutant PFCs induced apical growth (red arrows). White arrows indicated the wildtype neighbors. Overexpression of Rok DN (B–B″) or sqh DN (C–C″) in aPKC mutant PFCs induced apical growth (red arrows). White arrows indicated the wildtype neighbors. E–E″ lgl mutant cells showed apical growth (red arrows). White arrows indicated the wildtype neighbors. lgl mutant cells showed apical growth (red arrows). White arrows indicated the wildtype neighbors. F–G″Overexpression of Rok CA (F–F″) or sqh CA (G–G″) in lgl 4 PFC clones switched apical growth to basal growth (red arrows). White arrows indicated the wildtype neighbors. Overexpression of Rok CA (F–F″) or sqh CA (G–G″) in lgl 4 PFC clones switched apical growth to basal growth (red arrows). White arrows indicated the wildtype neighbors. I–I″Loss of aPKC‐induced basal growth (red arrows). Loss of aPKC‐induced basal growth (red arrows). J–J″Expression of EGFR A887T enhances basal growth (red arrows). Expression of EGFR A887T enhances basal growth (red arrows). K–K″Expression of Rok DN repressed basal growth in aPKC mutant clones with EGFR A887T (red arrows). Expression of Rok DN repressed basal growth in aPKC mutant clones with EGFR A887T (red arrows). D, H, LThe frequency of apical growth or basal growth was quantified for indicated genotypes (n = 60). Scale bars represent 20 μm. The frequency of apical growth or basal growth was quantified for indicated genotypes (n = 60). Scale bars represent 20 μm. Data information: Three independent experiments were performed (D, H, L). Error bars are ±SEM. ***P < 0.001 (Student's t‐test). Source data are available online for this figure. Furthermore, we enhanced Rok‐Sqh activity by expressing a constitutively active form of Rok (Rok CA ) or a phosphomimetic form of sqh (sqh CA ) in lgl 4 PFC clones, and found that expression of either Rok CA or sqh CA in lgl 4 PFC clones alleviated the apical growth phenotype with less frequency (93.3% in lgl 4 vs. 42% in lgl 4 ‐Rok CA , and 47.2% in lgl 4 ‐sqh CA ; n = 60 in each genotype) and increased the frequency of basal growth (6.7% in lgl 4 vs. 58% in lgl 4 ‐Rok CA , and 52.8% in lgl 4 ‐sqh CA ; n = 60 in each genotype) (Fig 7E–G″ and H). In control egg chambers, overexpression of UAS‐Rok CA and UAS‐sqh CA in FCs with MARCM did not induce multilayering (Appendix Fig S5A and C). These findings suggest that upregulation of Rok‐Sqh activity promotes basal growth. Together, these genetic epistasis analyses indicate that Rok and Sqh phosphorylation is critical in the FE to regulate the direction of accumulation of polarity‐deficient cells. Rok acts as a downstream target of the EGFR‐Pnt pathway EGFR signaling in the FCs is mediated by a downstream transcription factor Pnt (Morimoto et al, 1996). The study with chromatin immunoprecipitation sequencing (ChIP‐seq) in the Drosophila intestine showed that the enhancer of Rok was a binding target of Pnt (Zhang et al, 2022). Thus, we wondered if Rok is transcriptionally regulated by Pnt in the FE. To test this possibility, we analyzed the promoter region of Rok using JASPAR (Sandelin et al, 2004) and identified three potential Pnt‐binding sites (Fig 8A). Using an anti‐GFP antibody to perform a ChIP analysis with Pnt‐GFP expressing ovary, the three potential Pnt target segments (Amplicons 1, 2, and 3) were selected for the qPCR analysis and the result showed that Pnt strongly bound to two sites (sites 1 and 3) and weakly bound to one site (site 2) (Fig 8B and C). Furthermore, a qRT–PCR analysis with ovaries containing Pnt overexpressing FCs showed that Rok expression was upregulated (Fig 8D). Together, these studies indicate that Rok is a transcriptional target of Pnt in the FE. Rok acts as a transcriptional target of Pnt AThe consensus Pnt‐binding sites and three consensus Pnt‐binding sequences were found in 1 kb promoter/enhancer region of Rok.B, CThe ChIP experiment detects direct binding of Pnt‐GFP to the promoter/enhancer region of Rok. Transgenic flies with pnt‐GFP were subject to ChIP experiment using the anti‐GFP antibody. The enhancer regions detected by different primers for sites 1, 2, and 3 were shown in B.DThe qRT–PCR shows that overexpression of Pnt in follicle cells upregulated Rok expression.EThe frequency of apical growth or basal growth was quantified for indicated genotypes, n = 60 for each genotype. AThe consensus Pnt‐binding sites and three consensus Pnt‐binding sequences were found in 1 kb promoter/enhancer region of Rok. The consensus Pnt‐binding sites and three consensus Pnt‐binding sequences were found in 1 kb promoter/enhancer region of Rok. B, CThe ChIP experiment detects direct binding of Pnt‐GFP to the promoter/enhancer region of Rok. Transgenic flies with pnt‐GFP were subject to ChIP experiment using the anti‐GFP antibody. The enhancer regions detected by different primers for sites 1, 2, and 3 were shown in B. The ChIP experiment detects direct binding of Pnt‐GFP to the promoter/enhancer region of Rok. Transgenic flies with pnt‐GFP were subject to ChIP experiment using the anti‐GFP antibody. The enhancer regions detected by different primers for sites 1, 2, and 3 were shown in B. DThe qRT–PCR shows that overexpression of Pnt in follicle cells upregulated Rok expression. The qRT–PCR shows that overexpression of Pnt in follicle cells upregulated Rok expression. EThe frequency of apical growth or basal growth was quantified for indicated genotypes, n = 60 for each genotype. The frequency of apical growth or basal growth was quantified for indicated genotypes, n = 60 for each genotype. Data information: Three independent experiments were performed (D, E). Error bars are ±SEM. ***P < 0.001 (Student's t‐test). Source data are available online for this figure. To corroborate that Pnt mediates the role of EGFR signaling in regulating apical or basal growth of polarity‐deficient PFCs, pnt was overexpressed in lgl mutant FCs using the MARCM technique. These mosaic FE showed that the growth direction of lgl PFCs was switched from a predominantly apical to a basal preference (Fig 8E, 93.6% in lgl 4 vs. 31% in lgl 4 ‐Pnt; n = 67), This phenotype is similar to lgl PFC clones with re‐activated EGFR signaling (Fig 3E–E″′), suggesting that Pnt acts downstream of EGFR signaling to regulate the growth direction of polarity‐deficient FCs. Taken together, these findings connect EGFR signaling and the actomyosin cytoskeletal regulators Rok‐Sqh through transcription factor Pnt in the control of polarity loss‐induced multilayering. Discussion Intact epithelial cell polarity maintains the integrity and protects cells from unnecessary growth, extrusion or invasion. The direction through which they escape the epithelial layer has a profound impact on their fates, which include cell death and overgrowth (Tamori et al, 2016). In this study, using the Drosophila FE model, we found that EGFR signaling was differentially regulated in the PFCs with loss of lgl/Dlg or Baz/aPKC, and demonstrated that different levels of EGFR signaling activity in PFCs controlled the direction of formation of multilayers when apical‐basal polarity was disrupted (Fig 9B and C, compared with Fig 9A). Using scRNA‐seq analysis, we identified enrichment of Rok expression in mutant cells with simultaneous aPKC loss and EGFR activation. Further studies with qRT–PCR and ChIP analyses suggest that Rok is a transcriptional target of Pnt, which is downstream of EGFR signaling (Gabay et al, 1996). Rok, through the control of phosphorylation of RMLC, modulates the actomyosin network and epithelial organization. pRMLC is enhanced in mutant cells with basal accumulation where EGFR signaling is intact. Upregulation of EGFR signaling in these cells resulted in further accumulation of pRMLC around the cell cortex. These findings thereby suggest an EGFR‐Pnt‐Rok‐RMLC regulatory axis that governs the pattern of epithelial overgrowth upon polarity loss. A model for the regulation of apical and basal growth in polarity‐deficient cells by EGFR signaling and the Rok‐Sqh regulatory axis AThe wildtype follicle cells show a monolayer.B, CIn follicle cells with polarity loss, reduction of EGFR signaling downregulates the Rok‐Sqh signal to promote apical growth (B), but EGFR signaling upregulates Rok‐Sqh signal to induce basal growth (C). AThe wildtype follicle cells show a monolayer. The wildtype follicle cells show a monolayer. B, CIn follicle cells with polarity loss, reduction of EGFR signaling downregulates the Rok‐Sqh signal to promote apical growth (B), but EGFR signaling upregulates Rok‐Sqh signal to induce basal growth (C). In follicle cells with polarity loss, reduction of EGFR signaling downregulates the Rok‐Sqh signal to promote apical growth (B), but EGFR signaling upregulates Rok‐Sqh signal to induce basal growth (C). Epithelial cells experiencing a loss of polarity can exit the epithelial layer through extrusion or delamination processes (Slattum & Rosenblatt, 2014). The mechanisms governing directional extrusion/delamination involve the disruption of intercellular junctions and the contraction of actomyosin (Rosenblatt et al, 2001; Slattum et al, 2009; Gu et al, 2011; Tamori et al, 2016; Ohsawa et al, 2018). Typically, cortical actomyosin first contracts apically within the extruding cell, while actomyosin cables in surrounding cells collaborate to squeeze out the extruding cell (Rosenblatt et al, 2001; Zulueta‐Coarasa & Rosenblatt, 2021). In the FE model, polarity‐deficient cells were not observed to undergo apoptosis, and they maintained connections with wildtype cells, forming multilayers alongside their neighboring wildtype cells. The specific regulatory factors governing the directional localization, whether it involves contraction or cell shape changes, remain to be investigated. Our studies have found that EGFR signaling is differentially regulated in FCs when various cell polarity complexes are lost, but the underlying mechanism remains elusive. The fate of EGFR protein, either being sorted to lysosomes for degradation or the plasma membrane for recycling (Madshus & Stang, 2009; Tomas et al, 2014) (Ceresa, 2006; Rappoport & Simon, 2009) (Nishimura & Itoh, 2019), plays a fundamental role in regulating EGFR signaling. The distinctions in EGFR signaling regulation by the Lgl/Dlg/Scrb and Par complexes might be due to their specific molecular functions. The Scrib complex, for example, is involved in retromer‐dependent sorting events that facilitate the return of internalized cargo to the cell surface (de Vreede et al, 2014) (Stephens et al, 2018), which could affect the recycling of the EGFR protein. On the other hand, the Par complex has been found to act downstream of Cdc42 to regulate the apical endocytotic pathway (Harris & Tepass, 2008), which may not significantly impact EGFR membrane trafficking. Hence, the regulation of EGFR signaling by these cell polarity complexes could involve their influence on EGFR localization at the plasma membrane or their impact on EGFR degradation processes. As a result, further investigation into how different cell polarity complexes affect the trafficking of the EGFR protein and its subsequent impact on EGFR signaling is intriguing and should be pursued in future studies. This may shed light on the intricate molecular mechanisms underlying EGFR regulation and its connections to cell polarity pathways. During oogenesis, multiple signaling pathways are activated to specify the fates of anterior follicle cells (AFCs) and PFCs. Without EGFR signaling, the PFC may adopt an AFC‐like behavior, which includes the invasion by border cells that migrate through the germline nurse cells. The unmasking of EGFR signaling by lgl/Dlg mutations may allow the mutant PFC behave like invasive border cells, which still depend on Jak–STAT signaling for their migration. This may explain why lateral follicle cells and early follicle cells do not normally invade inwards upon polarity loss. The complexity of the AFCs also includes the activation of PDGF signaling, another RTK signaling that may be needed for apical growth prevention. Therefore, it would be interesting to determine whether the AFCs undergo directional growth upon polarity loss and what the underlying mechanisms are. In studies, the Drosophila wing imaginal disc epithelia (Appendix Fig S6A–A″′) have been used as a model system to investigate cell delamination and invasion. Specifically, we examined the localization of lgl and aPKC mutant cells in comparison with neighboring wildtype cells in the notum regions. Interestingly, we observed that lgl mutant cells tended to be localized at the apical side of wildtype neighboring cells (Appendix Fig S6B–B″′, 89%, n = 23). On the other hand, aPKC mutant cells were found to accumulate at the basal side of neighboring cells (Appendix Fig S6C–C″′, 91%, n = 32). These observations indicate that the location patterns of these mutant cells closely resemble those observed in FCs, highlighting the conserved phenotype in the different tissues. However, whether the EGFR signaling pathway is involved in regulation of this process in the wing discs will be explored in our future studies. Clinically, tumors from pancreatic ducts, retina, kidney, and tongue of mammals show exophytic and endophytic growth, which is defined by tumor cells growing out of or crawling under the epithelium (Harasymczuk et al, 2013). The apical and basal location of mutant cells revealed in the Drosophila FE model resembles the endophytic and exophytic growth of tumor in humans, respectively. Clinical studies revealed that exophytic tumor growth is associated with a high level of EGFR expression (Phuc et al, 2021), and the increased expression of EGFR has also been identified in a variety of cancers to promote tumor growth (Sharma et al, 2007; Sasaki et al, 2013). Our studies indicate that activation of EGFR signaling is associated with basal growth direction and upregulation of EGFR signaling can further enhance their basal expansion, suggesting that the regulatory mechanism learned in the Drosophila FE model could be applied to cancer studies. In addition, the studies in the mouse found that the tension imbalance and tissue curvature in tubular epithelia regulate different tumor growth patterns (Messal et al, 2019). As endophytic or exophytic tumors have different prognosis and postoperative local recurrence (Kirita et al, 1994; Tsivian et al, 2010; Eslami et al, 2015), these models from Drosophila and mammals provide platforms for identifying intrinsic signaling pathways, biophysical parameters, and microenvironment involved in the regulation of these different growth pattern tumors. Materials and Methods Fly strains and clone generation. The strains we used were raised at 25°C on standard media. The following flies were used: lgl 4 FRT40A/Cyo (strong LOF allele), lgl 27S3 FRT40A/Cyo (LOF allele), FRT42D apkc k06403 /Cyo (strong LOF allele), Baz Xi106 FRT9‐2, Dlg m30 FRT19A, and Dlg m52 FRT19A. Transgenic lines: UAS‐EGFR‐RNAi (VDRC43267), UAS‐EGFR A887T (BL#9536), UAS‐aPKC‐RNAi (BL#25946), UAS‐lgl‐RNAi (VDRC51249), pointed‐LacZ (pnt‐LacZ), UAS‐Rok DN (kinase dead, BL#6671), UAS‐sqh DN (nonphosphorylatable sqh, BL#64114), UAS‐Rok CA (active form of Rok, BL#6669), UAS‐sqh CA (phospho‐mimic, BL#64411), sqh AX3 FRT19A/FM7c (BL#25712), hsFlp, act > CD2 > Gal4, UAS‐RFP, Pnt‐GFP (BL#42680). Follicular mosaic clones and MARCM clones were generated by FLP‐FRT‐induced mitotic recombination (Xu & Rubin, 1993), and the following strains were used: ywhsFLP; Ubi‐GFPFRT40A/CyO, ywhsFLP; FRT42D Ubi‐GFPP/CyO, Ubi‐hGFP FRT9‐2/FM7c; hsFlp/TM6, hsFlp RFP FRT19A, hsFLP; FRT42D tubGAL80/cyo; actin‐GAL4 UAS‐mCD8:GFP, hsFLP; tubGAL80 FRT40A; actin‐GAL4 UAS‐mCD8:GFP. For generation of mosaic clones and MARCM clones, adult flies with the appropriate genotypes were subjected to heat shock at 37°C for 1 h, and flies were raised at 25°C for different periods before dissection. To ensure consistency and avoid spatio‐temporal discrepancies in phenotypes, we examined posterior follicle cells (PFCs) in stage 7–10 egg chambers with 5–7 days of incubation time after clone induction (ACI). Immunohistochemistry Ovaries were dissected out and fixed in 4% paraformaldehyde solution, and the following primary antibodies were applied according to standard antibody staining procedures (Tian & Deng, 2008): rabbit anti‐galactosidase, 1:2000 (Sigma); goat anti‐GFP (Abcam, Cambridge, MA); Rabbit anti‐pRMLC, 1:100 (Cell Signaling); and Alexa Fluor 633 Phalloidin and 488 Phalloidin, 1:50 (Invitrogen). Secondary antibodies conjugated to Alexa Fluor 546 donkey anti‐mouse and anti‐rabbit (Molecular Probes) and Alexa Fluor 633 donkey anti‐mouse and anti‐rabbit and 488 Donkey anti‐goat (Jackson immunoresearch) were used at 1:400. Fluorescently labeled samples were counterstained with DAPI for visualization of DNA. Images were captured with a Zeiss LSM 800 confocal microscope and assembled in Adobe Photoshop. scRNA‐seq sample preparation, sequencing, and data analysis Drosophila ovaries with the desired genotypes (hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T and hsFLP; FRT40A tubGAL80/FRT40A; actin‐GAL4 UAS‐mCD8:GFP) at 6‐d ACI were then dissected in complete medium (Grace's Insect Basal Medium supplemented with 15% fetal bovine serum) as previously described (Jevitt et al, 2020). To prevent cell clumping, ovaries were transferred to a tube containing 300 μl Earle's Balanced Salt Solution (EBSS) (no calcium, magnesium, and phenol red) and gently washed for 2 min. The EBSS was then removed, and the tissue was dissociated in 100 μl Papain (50 U/ml in EBSS and previously heat activated in 37°C for 15 min) for 30 min. The suspension was mechanically dissociated every 3 min by gentle pipetting up and down. To quench the digestion, 500 μl complete medium was added to dissociated cells. The suspension was then passed through a 40 μl sterile cell strainer and centrifuged for 10 min at 700 RCF to remove debris and large eggs with intact eggshells that could not be dissociated. This process also filtered out larger germline cells, which increase dramatically in size around stage 9 (Kolahi et al, 2009). The supernatant was removed, and single cells were resuspended in 100 μl. Cell viability was assayed with AOPI solution, and cell number was determined with a Nexcelom Cellometer. Cells were then further diluted to an approximate, final density of 2,000 cells/μl according to 10× Genomics. Libraries were generated by Chromium Next GEM Single Cell v3.1 and were sequenced in a NovaSeq 6000 (Illumina) at Florida State University. Cell Ranger (version 3.0.0) processing resulted in a total of 307,977,702 reads with mean 39,945 reads per cell for wildtype (con), and 234,288,861 reads with mean 21,245 reads per cell for aPKC‐EGFR A887T . Seurat (Butler et al, 2018; Stuart et al, 2019) was used for log‐normalization and scaling of the data by means of default parameters and was used to integrate the two datasets from the wildtype and aPKC‐EGFR A887T into a single dataset, and RunUMAP with a resolution of 2 was used to group the integrated dataset. Chromatin immunoprecipitation coupled with qPCR Approximately 50 pairs of ovaries were used in each experiment, dissected from females raised at 25°C on standard cornmeal/agar medium. Samples are prepared as previously described (Evolutionarily Conserved Roles for Apontic in Induction and subsequent Decline of Cyclin E Expression). Briefly, dissected ovaries were fixed with 1% formaldehyde for 20 min at room temperature. After fixation, samples were terminated with 2.5 M glycine, homogenized with lysis buffer, and stored at −80°C until needed. Samples were sonicated with the Covaris M220 130‐μl microtube using the setting of Peak Incident Power 50, Duty Factor 20%, Cycle per Burst 200, and Treatment Time 150 s. An average size of 150–200 bp was produced. The resulting chromatin was processed in three ways. First, 5% of the chromatin was heat treated to reverse cross‐links and phenol–chloroform extracted to generate the input DNA fraction. Second, the remaining chromatin was divided into two fractions. One fraction was incubated with anti‐GFP antibody (2 μl) (abcam‐ab290) and the other with control Rabbit IgG (2 μl). Enriched DNA was obtained by incubation with protein A beads (Sigma 16‐661) and phenol–chloroform extraction. RT–qPCR Total RNAs were prepared from approximately 10 pairs of ovaries using Rneasy Plus Mini Kit (Zymo Research, Cat. number R2050). cDNA was synthesized using the iScript cDNA synthesis kit (Bio‐Rad). RT–qPCR was performed using iQ SYBR Green System (Bio‐Rad). RpL11 was used as a normalization control. Primers for Rok: 5′‐TACGAATGCAAGAGATGC‐3′ and 5′‐CGGGTCGTGTTTGTCCACAT‐3′. Relative quantification of mRNA levels was calculated using the comparative CT method. Statistical analyses were performed by Student's t‐test. Statistical analyses Statistical analyses were performed with a two‐tailed unpaired t‐test. P‐value is indicated by asterisks in the Figures: **P < 0.01; ***P < 0.001. Differences at P < 0.01 were considered significant. Genotypes for flies in each figure Figure 1. (A–A″′, E–E″) ywhsFLP; Ubi‐GFPFRT40A/FRT40A. (B–B″′, C–C″′, F–F″, G‐G″) ywhsFLP; Ubi‐GFPFRT40A/lgl 4 FRT40A. Figure 2. (A, B) hsFLP Ubi‐RFP FRT19A/Dlg m52 FRT19A. (C, D) ywhsFLP; FRT42DUbi‐GFPP/FRT42DaPKC k06403 . (E, F) Baz Xi106 FRT9‐2/Ubi‐His2A‐GFPP FRT9‐2; hsFlp/+. (G–G″″) ywhsFLP; Ubi‐GFPFRT40A/FRT40A. (H–H″′) ywhsFLP; Ubi‐GFPFRT40A/lgl 4 FRT40A. (I–I″′) Baz Xi106 FRT9‐2/Ubi‐His2A‐GFPP FRT9‐2; hsFlp/+. Figure 3. (A–A″) ywhsFLP; Ubi‐GFPFRT40A/FRT40A; pnt‐LacZ/+. (B–B″) ywhsFLP; Ubi‐GFPFRT40A/lgl 4 FRT40A; pnt‐LacZ/+. (C–C″) hsFLP; tubGAL80 FRT40A/FRT40A; actin‐GAL4 UAS‐GFP/UAS‐EGFR A887T . (D–D″) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP. (E–E″) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T . Figure 4 (A–A″′) ywhsFLP; FRT42DUbi‐GFP/FRT42D aPKC k06403 ; pnt‐LacZ. (B–B″, J) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP. (C–C″) hsFLP; FRT42D tubGAL80/FRT42D; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR‐RNAi. (D–D″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR‐RNAi. (E–E″, I) hsFLP; FRT42D tubGAL80/FRT42D; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T . (F–F″, K) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T . Figure 6. (A–A″) hsFLP; FRT42D tubGAL80/FRT42D; actin‐GAL4 UAS‐mCD8:GFP. (B–B″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP. (C–C″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T . (D–D″) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP. (E–E″) hsFLP; tubGAL80 FRT40A/ lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T . (F–F″) hsFLP; tubGAL80 FRT40A/FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR A887T . (G–G″) hsFLP; tubGAL80 FRT40A/FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐EGFR‐RNAi. (H–H″) hsFLP; tubGAL80 FRT40A/FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐Rok DN . Figure 7. (A–A″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/+. (B–B″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/UAS‐Rok DN . (C–C″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/UAS‐sqh DN . (E–E″) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP. (F–F″) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐Rok CA . (G–G″) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP/UAS‐sqh CA . (I–I″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/+. (J–J″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/EGFR A887T . (K–K″) hsFLP; FRT42D tubGAL80/FRT42DaPKC k06403 ; actin‐GAL4 UAS‐mCD8:GFP/EGFR A887T Rok DN . Genotypes for flies in each supplementary figure Figure EV1. (A–C′) ywhsFLP; Ubi‐GFPFRT40A/lgl 4 FRT40A. (D–G″) hsFLP Ubi‐RFP FRT19A/Dlg m52 FRT19A. Figure EV2. (A–B) ywhsFLP; Ubi‐GFPFRT40A/lgl 27S3 FRT40A. (C–D) hsFLP Ubi‐RFP FRT19A/Dlg m30 FRT19A. Figure EV3. (A–C′, D–D″, F–F″, H–H″, J–J″) Baz Xi106 FRT9‐2/Ubi‐His2A‐GFPP FRT9‐2; hsFlp/+. (E–E″, G–G″, I–I″, K–K″) ywhsFLP; FRT42DUbi‐GFPP/FRT42DaPKC k06403 . Figure EV4. (A–A″′) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP/pnt‐lacZ. (B–B″′) hsFLP; tubGAL80 FRT40A/lgl 4 FRT40A; actin‐GAL4 UAS‐mCD8:GFP/EGFR A887T pnt‐lacZ. Figure EV5. (A–A′) ywhsFLP; Ubi‐GFPFRT40A/FRT40A. (B–B′) ywhsFLP; Ubi‐GFPFRT40A/EGFR f24 FRT40A. (C–C′) hsFLP; tubGAL80 FRT40A/FRT40A; actin‐GAL4 UAS‐mCD8:GFP/EGFR‐RNAi. (D–D″) ywhsFLP; Ubi‐GFPFRT40A/FRT40A; pnt‐lacZ/+. (E–E″) ywhsFLP; Ubi‐GFPFRT40A/EGFR f24 FRT40A; pnt‐lacZ/+. Author contributions Aiguo Tian: Conceptualization; formal analysis; funding acquisition; investigation; methodology; writing – original draft; project administration; writing – review and editing. Xian‐Feng Wang: Formal analysis; investigation. Yuting Xu: Data curation; investigation. Virginia Morejon: Data curation; investigation. Yi‐Chun Huang: Data curation; investigation. Chidi Nwapuda: Formal analysis. Wu‐Min Deng: Conceptualization; funding acquisition; writing – review and editing. Disclosure and competing interests statement The authors declare that they have no conflict of interest. Supporting information Appendix Click here for additional data file. Expanded View Figures PDF Click here for additional data file. Movie EV1 Click here for additional data file. PDF+ Click here for additional data file. Source Data for Figure 1 Click here for additional data file. Source Data for Figure 2 Click here for additional data file. Source Data for Figure 3 Click here for additional data file. Source Data for Figure 4 Click here for additional data file. Source Data for Figure 5 Click here for additional data file. Source Data for Figure 6 Click here for additional data file. Source Data for Figure 7 Click here for additional data file. Source Data for Figure 8 Click here for additional data file. Acknowledgements We thank Dr. David Bilder, Bloomington Drosophila Stock Center for stocks (NIH P400D018537), and Developmental Studies Hybridoma Bank for providing fly lines and antibodies. We also thank Drs. Hong Liu and Jun‐yuan Ji for careful reading of the manuscript and Kejing Song from Tulane University, Yanming Yang, and Cynthia Vied from Florida State University for single‐cell library preparation and sequencing. This work was supported by the National Institutes of Health P20GM103629, A Carol Lavin Bernick Faculty Grant, Bridge Research Fund at Tulane University and Bridge Research Fund at SOM to AT, and the National Institutes of Health GM072562, CA224381, and CA227789 to WMD. The EMBO Journal (2023) 42: e113856 Contributor Information Aiguo Tian, Email: atian@tulane.edu. Wu‐Min Deng, Email: wdeng7@tulane.edu. Data availability All raw sequence files are available from the SRA repository: SRR26353200 and SRR26353201 (http://www.ebi.ac.uk/ena/data/view/SRR26353200 and http://www.ebi.ac.uk/ena/data/view/SRR26353201). References Supplementary Materials Appendix Click here for additional data file. Expanded View Figures PDF Click here for additional data file. Movie EV1 Click here for additional data file. PDF+ Click here for additional data file. Source Data for Figure 1 Click here for additional data file. Source Data for Figure 2 Click here for additional data file. Source Data for Figure 3 Click here for additional data file. Source Data for Figure 4 Click here for additional data file. Source Data for Figure 5 Click here for additional data file. Source Data for Figure 6 Click here for additional data file. Source Data for Figure 7 Click here for additional data file. Source Data for Figure 8 Click here for additional data file. Data Availability Statement All raw sequence files are available from the SRA repository: SRR26353200 and SRR26353201 (http://www.ebi.ac.uk/ena/data/view/SRR26353200 and http://www.ebi.ac.uk/ena/data/view/SRR26353201).