Introduction

Correct cell polarity maintenance is essential for tissue homeostasis and cell physiology, whereas dysregulation of cell polarity has been characterized as an important step and a hallmark of cancer progression.1 Consequently, an overwhelming amount of studies have focused on the connections between polarity regulation and cancer development.2, 3, 4, 5 In particular, studies on Drosophila have defined numerous signaling modulators acting to establish and maintain epithelial polarity, including the basolaterally localized Scribble complex, which is composed of Scribble (Scrib), discs large (Dlg) and lethal giant larva (Lgl).6 In fly, loss of any of these proteins can cooperate with oncogenic Ras to induce invasive tumors;7 whereas in mammal, homologs of these proteins show consistently aberrant expression in multiple cancers, including lung, breast, colon and prostate.8, 9, 10 Despite the well-established connection between dysregulation of cell polarity and cancer, the factors that act downstream of the Scrib complex to mediate cancer progression remain poorly understood.

During the past decade, Drosophila melanogaster has become an excellent model to investigate the genetic mechanism of cancer biology,11, 12 and several invasion and metastasis models have been established in larvae and adult flies.12, 13 For instance, in the developing wing epithelia, loss-of-cell polarity genes or C-terminal SRC kinase (Csk) along the anterior/posterior compartment boundary using patched-Gal4 driver produces an invasive migration phenotype,14, 15, 16, 17 which has been used to study cell invasion in vivo.11

The structure and function of Scribble complex members have been highly conserved from fly to human,18, 19 which enables us to study the downstream signaling and underlying mechanism in simpler organisms like fly. To identify additional factors downstream of the Scribble complex involved in cell invasion and cancer progression, we performed a genetic screen using the previously described loss of scrib-induced cell invasion model in Drosophila wing discs.14, 15 We found that Rho1, a small GTPase, and Wallenda (Wnd), a Ser/Thr protein kinase of the mitogen-activated protein kinase kinase kinase (MAPKKK) family, are crucial regulators of loss-of-cell polarity or ectopic Src42A-induced cell invasion. Our genetic analysis indicates that Rho1 acts upstream of Wnd to regulate c-Jun N-terminal kinase (JNK)-dependent aspects of cell invasion, including loss of adherence junction, activation of metastasis markers and epithelial–mesenchymal transition (EMT). Furthermore, our biochemical studies confirm a physical interaction between Rho1 and Wnd. Finally, we show that Wnd cooperates with oncogenic Ras to stimulate wingless (Wg) expression, cell proliferation and tumor invasion.

Results

Rho1 and Wnd are required for cell invasion

As previously reported, RNA interference (RNAi)-mediated downregulation of the cell polarity gene scrib, driven by ptc-Gal4, triggered a strong cell invasion phenotype toward the posterior part of wing discs (Figures 1a and b), along with upregulation of matrix metalloprotease1 (MMP1; Supplementary Figure 1A),14, 15 a protein that is essential for basement membrane degradation and a hallmark of EMT.20, 21 This invasion phenotype and MMP1 upregulation depends on JNK signaling,14, 15, 22 which is activated by multiple upstream factors including mitochondrial respiratory complexes, small GTPases, MAPKKK and MAPKK family members, and so on,23, 24, 25, 26, 27, 28 yet it has remained unknown which factors mediate JNK activation induced by the loss-of-cell polarity. To investigate the molecular mechanism underlying loss-of-cell polarity-triggered cell invasion, we have conducted a candidate screen using the ptc>scrib-IR invasion phenotype. We have screened >500 UAS-RNAi lines from the Bloomington and National Institute of Genetics (NIG) stock centers targeting potential factors upstream of JNK, or factors that interact with JNK pathway components genetically or biochemically as reported in the literature. The screen is still ongoing, as more RNAi lines are being added to the stock centers. We called specific attention to two genes identified in the screen: Rho1 and wnd. Rho1 belongs to the GTPase family and is a major regulator of actin cytoskeleton dynamics and cell migration.29 Wnd encodes a MAPKKK homolog of vertebrate dual leucine zipper kinase (DLK), a protein implicated in neuronal cell radical migration and apoptosis.30, 31 We found that knocking down either Rho1 or wnd dramatically suppressed ptc>scrib-IR-induced cell invasion and MMP1 activation (Figures 1b–d and l, Supplementary Figures 1A–C) as well as lgl loss-induced cell invasion (Figures 1e–g and l) in wing discs, whereas loss of Rho1 or wnd alone gave no obvious phenotype (Supplementary Figures 1D and E). The knockdown efficiencies of RNAi lines were verified by quantitative reverse transcription polymerase chain reaction (Supplementary Figures 2E and F). To exclude potential off-target effects of RNAi, we examined at least two independent UAS-RNAi lines for each gene, including the previously described lines for Rho1 and wnd,32, 33 and obtained similar results (data not shown). Furthermore, we confirmed the suppression by wnd and Rho1 mutant alleles (Supplementary Figures 1G–K).

Figure 1
figure 1

Rho1 and Wnd are required for impaired cell polarity or Src42A-induced cell invasion. Fluorescence micrographs of wing discs are shown. Compared with the control (a, a′), loss of scrib (b) or lgl (e) induced cell invasion is suppressed by knocking down Rho1 (c, f) or wnd (d, g). Compared with the control (h–h”), ectopic Src42A-induced cell invasion and MMP1 upregulation (i–i”) are dramatically impeded by knocking down Rho1 (j–j”) or wnd (k–k”). (l) and (m) Quantification data of cell migration in panels a–g and h–k. ***P<0.001. (mean+s.d., n=10). UAS-LacZ was included (b, e and i) as a negative control to demonstrate that the suppression by UAS-RNAi is specific, not a result of Gal4 titration by another UAS line.

Full size image

The oncoprotein Src is known to regulate various aspects of tumor progression ranging from proliferation and survival to migration and metastasis.17, 25, 34 Recent work has shown that expression of Drosophila Src42A recapitulates loss-of-cell polarity-induced JNK-mediated cell invasion.15, 17, 25 Consistently, we found Src42A-induced cell invasion and MMP1 activation (Figures 1i, i',i” and m) were significantly suppressed by knocking down Rho1 (Figures 1j, j',j” and m) or wnd (Figures 1k, k',K” and m). Together, these data indicate that both Rho1 and Wnd are crucial regulators for cell invasion in Drosophila wing discs.

Wnd induces JNK-mediated EMT

To further explore the role of Wnd in cell invasion, we drove Wnd expression by ptc-Gal4 in the wing disc with tub-Gal80ts, a temperature-sensitive form of the Gal4 inhibitor Gal80 (Gal80ts) that blocks Gal4 activity at low temperature (18 °C) but is inactivated at high temperature (29 °C).35 After Wnd induction, we observed that a large number of cells migrated significantly apart from the ptc domain toward the posterior part, coupled with intensive MMP1 activation (Figure 2e), whereas co-expression of a wnd RNAi completely suppressed the invasion behavior (Supplementary Figures 2A and B). Furthermore, Wnd-expressing cells display reduced cell adherent junction marker E-cadherin (Figure 2f) and increased accumulation of F-actin and β-integrin (Figures 2g and h). Taken together, these results suggest that Wnd-expressing epithelia cells exhibit molecular characteristics of EMT, including MMP1 induction, loss of E-cadherin and actin remodeling.36 In contrast, expression of a kinase dead form of Wnd (WndKD) shows no obvious EMT effects (Figures 2i–l), indicating that Wnd requires its kinase activity to induce cell invasion.

Figure 2
figure 2

Wnd promotes cell invasion and EMT. Fluorescence micrographs of wing discs are shown. Compared with the controls (a–d), cells expressing Wnd along the A–P boundary show invasive migration and MMP1 secretion (e), reduced E-cadherin (f), increased actin (g) and β-integrin (h) accumulation, whereas cells expressing WndKD fail to produce such phenotypes (i–l).

Full size image

MMP1 serves as a transcriptional reporter for JNK signaling in tumor cell invasion.20, 21 Consistent with the MMP1 data, expression of Wnd (Figures 3a–d), but not WndKD (data not shown), promotes JNK phosphorylation and puc-LacZ expression, another reporter for JNK signaling. Conversely, Wnd-induced EMT behavior (Figures 3f, i, j and k) is fully suppressed by co-expression of BskDN (Figures 3g, i, l and m), a dominant negative form of Drosophila JNK homolog Bsk, suggesting that JNK signaling is indispensable for Wnd-triggered cell invasion.

Figure 3
figure 3

JNK signaling is required for Wnd-induced cell invasion and EMT. (a and b) Light micrographs of wing discs are shown. Compared with the control (a), Wnd expression upregulates puc transcription (b). (c–h and j–p) Fluorescence micrographs of wing discs are shown. (c–e) Compared with the control (c), ectopic Wnd-induced JNK phosphorylation (d) is dramatically suppressed by depletion of MKK4 and hep (e). (f–h) Wnd-induced cell invasion and MMP1 activation (f) are significantly suppressed by co-expression of BskDN (g) or knocking down MKK4 and hep (h). (i) Quantification data of cell invasion in panels f–h (mean+s.d., n=10). (j–m) Wnd-induced integrin accumulation (j) and E-cad reduction (k) are suppressed by co-expression of BskDN (l and m). (n) Co-expression of p35 cannot block Wnd-induced cell invasion and MMP1 activation. (o and p) Representative Flp-out clones expressing Wnd and p35 have ‘filopodia’ like structure and display intensive MMP1 activation (o) and F-actin accumulation (p).

Full size image

To determine how JNK signaling is regulated by Wnd, we analyzed the upstream components of the JNK pathway. Drosophila encodes two JNK kinases, the MKK7 homolog Hemipterous (Hep) and MKK4. Hep is involved in dorsal closure, cell death and tumor cell invasion,37, 38, 39 and MKK4 is a regulator of Eiger-induced cell death.40 Although knocking down either hep or mkk4 alone has no significant suppression on Wnd-induced cell invasion (data not shown), simultaneously knocking down both genes dramatically impedes Wnd-triggered JNK phosphorylation, MMP1 activation and cell invasion (Figures 3e, h and i), suggesting a functional redundancy between hep and mkk4 in mediating Wnd activity. Collectively, these data suggest that Wnd induces JNK-mediated EMT in wing epithelial through both Hep and MKK4.

Cell invasion is frequently accompanied with apoptosis.17 In accordance with this notion, we found that Wnd expression also induced strong cell death, especially in invading cells (Supplementary Figures 3A–D). Previous work showed that caspase activation below cell death threshold also induces JNK-dependent cell invasion,41 indicating that cell invasion is uncoupled from cell death. Consistent with this conclusion, we found that blocking cell death by expressing the caspase inhibitor p35 or deleting one copy of the proapoptotic RGH (reaper, hid and grim) genes did not impede Wnd-induced cell invasion and MMP1 expression (Figure 3n and Supplementary Figures 3E and F), confirming that the invasion behavior is not a secondary effect of cell death. To rule out the possibility that Wnd may have a non-autonomous effect on MMP1 activation, we generated Flp-out clones that co-express Wnd with p35 to block cell death, and observed a clear cell autonomous MMP1 induction (Supplementary Figure 3I). Furthermore, some cells within the clone displayed filopodia like structures and exhibited molecular characteristics of EMT,36 including intensive MMP1 activation and F-actin accumulation in a cell autonomous manner (Figures 3o and p). Some weak MMP1 staining was also observed outside the clones (Figure 3o”), likely representing their secretion to the extra cellular matrix.

Wnd is required for Rho1-induced cell invasion and JNK activation

Rho1 has essential roles in regulating cell invasion and tumorigenesis.17, 42, 43 In accordance with this, we found that ptc-Gal4-driven Rho1 expression also induced JNK activation, MMP1 upregulation, cell death and invasion in an autonomous manner (Figures 4b–b”, Supplementary Figures 4F, I and L),27, 42 whereas depletion of Rho1 completely blocked the invasion phenotype (Supplementary Figures 2C and D). Co-expression of the caspase inhibitor p35 kept Rho1-expressing cells alive (the so called ‘undead cells’44) and generated a broader ptc domain with enhanced MMP1 production, whereas the ‘undead cells’ could still invade into the posterior part (Supplementary Figures 4G–G”), suggesting that cell death is uncoupled from invasion. Consistent with the notion that invasive behavior is associated with disruption of epithelial integrity,36 Rho1-expressing cells exhibited increased integrin and actin accumulation (Figure 4g and Supplementary Figure 4B). We found that Rho1-triggered EMT markers, including cell invasion, integrin and actin remodeling, and MMP1 activation, were significantly suppressed by knocking down wnd (Figures 4c, e and h and Supplementary Figures 4C and E) or inactivation of JNK (Figures 4d, e and i). It is worth noting that Rho1-induced actin accumulation remained unaffected by co-expression of BskDN (Supplementary Figures 4D and E, see Discussion). Consistently, loss of wnd also suppresses Rho1-induced JNK activation and cell death, as indicated by puc-LacZ and acridine orange staining respectively (Supplementary Figures 4H–J and K–M). Taken together, we conclude that Rho1 induces JNK-dependent cell death and invasion through Wnd. Intriguingly, Rudrapatna and colleagues recently reported a positive feedback loop between JNK and Rho1 in cell invasion.25 In accordance with this finding, we found that Wnd-induced cell invasion was also blocked by depletion of Rho1 (Supplementary Figures 3G and H).

Figure 4
figure 4

Wnd is required for Rho1-induced cell invasion. Fluorescence micrographs of wing discs are shown. (a–d) Compared with the control (a), ectopic Rho1-induced cell invasion and MMP1 activation (b) are dramatically suppressed by knocking down wnd (c) or expressing BskDN (d). (e) Quantification data of cell invasion in panels a–d (mean+s.d., n=10). (f–i) Compared with the control (f), Rho1-induced integrin accumulation (g) is strongly suppressed by reducing wnd (h) or Bsk (i) activity.

Full size image

Wnd physically interacts with Rho1

As MAPKKK proteins are typically activated by GTPases through direct binding,45 we reasoned that Rho1 may regulate Wnd activity via physical interaction. To test this hypothesis, we expressed enhanced green fluorescent protein (EGFP)-tagged Wnd (EGFP–Wnd) or HA-tagged Rho1 (HA–Rho1) in Drosophila S2 cells. Both Wnd and Rho1 were distributed uniformly in the cytoplasm (Figures 5a and b), and a significant co-localization was observed when the two proteins were co-expressed (Figure 5c). Consistently, co-immunoprecipitation experiments indicated that Wnd co-precipitated with Rho1 (Figure 5d), confirming a physical interaction between Rho1 and Wnd in vivo.

Figure 5
figure 5

Rho1 physically interacts with Wnd. (a–c) Fluorescence micrographs of S2 cells are shown. Wnd (a) or Rho1 (b) is evenly distributed in the cytoplasm, and show cytoplasmic co-localization when co-expressed (c). (d) Rho1 interacts with Wnd in S2 cells. HA-Rho1 is co-precipitated with EGFP-Wnd in immunoprecipitation assay.

Full size image

Wnd is dispensable for Rac1-triggered cell invasion

Apart from Rho1, Rac1 and Rac2 are known Rho-GTPase family members that regulate cell migration processes including dorsal closure and wound healing.29, 46, 47 Indeed, ectopic expression of Rac1, but not Rac2, driven by ptc-Gal4 induces JNK-dependent cell invasion behavior (Supplementary Figures 5A, F, K and data not shown). However, Rac1-triggered invasion cannot be suppressed by loss of wnd (Supplementary Figure 5B), indicating that Wnd is dispensable for Rac1-induced cell invasion. Consistent with this, we found that Wnd does not physically interact with Rac1 in immunoprecipitation experiments (Supplementary Figure 5L).

Wnd belongs to the MAPKKK family that also includes Slipper (Slpr) and dTAK1, both of which have been implicated in Rho1-induced cell death.27 Whereas knocking down slpr or dTAK1 alone had no visible effect on Rac1-induced cell invasion (Supplementary Figures 5C, D and K), depletion of both slpr and dTAK1 significantly suppressed Rac1-induced invasion (Supplementary Figures 5E and K). Similarly, knocking down both slpr and dTAK1 dramatically impeded Rho1-induced cell invasion (Supplementary Figures 5G–K), suggesting that dTAK1 and Slpr act redundantly downstream of Rac1 and Rho1 in regulating cell invasion.

Wnd cooperates with p35 or RasV12 to promotes cell proliferation and tissue growth

JNK signaling cooperates with oncogenic Ras to promote growth and invasion.14, 20, 39, 48, 49 Consistent with its ability to activate JNK signaling, Wnd could cooperate with p35 or oncogenic Ras to stimulate cell proliferation. Co-expression of Wnd and either p35 or oncogenic Ras resulted in increased phospho-histone 3 (PH3) staining (Figure 6), increased CycE expression (Figures 7i–l), and increased tissue growth (Figure 6) as compared with expression of either p35 or oncogenic Ras alone. Moreover, co-expression of Wnd with RasV12 in eye disc clones resulted in massive tumor growth in larva heads (Supplementary Figures 6A–C), with about 20% of the larvae showing a phenotype of tumor invasion into the ventral nerve cord, which were never seen in larvae expressing RasV12 or Wnd alone (Supplementary Figures 6A'–C').

Figure 6
figure 6

Wnd cooperates with p35 and RasV12 to induce proliferation and growth. Fluorescence micrographs of wing discs are shown. Compared with the control (a), or expression of Wnd (b), p35 (c) or RasV12 (e) alone, co-expression of Wnd with p35 (d) or RasV12 (f) results in a significant increase of PH3+ cells and GFP+ region. (g) Ratio of GFP+ region versus the whole disc in a–f were shown. **P<0.01, *P<0.05 (mean+s.d., n⩾5).

Full size image
Figure 7
figure 7

Wnd induces JNK-dependent Wg expression. Fluorescence micrographs of wing discs are shown. Figures a–d are higher magnification of the wing pouch region. (a–d) Compared with the control (a), Wnd-induced Wg expression (b) is suppressed by BskDN (c), which does not affect the expression of endogenous Wg (d). (e–h) Ectopic expression of p35 (e) does not affect Wg expression, expression of or RasV12 slightly induces non-autonomously Wg serection (g), whereas co-expression of Wnd with p35 (f) or RasV12 (h) induces intensive Wg expression. (i–l) Compared with control (i), expression of Wnd or RasV12 fails to induce CycE expression (j and k), whereas co-expression of both synergistically upregulates CycE expression (l).

Full size image

Wg has been reported to act downstream of JNK signaling to stimulate cell proliferation, and Wg expression is upregulated by JNK signaling.44 Consistently, ectopic Wnd is sufficient to induce Wg expression in a JNK-dependent manner (Figures 7a–d). In addition, oncogenic cooperation between Wnd and p35 or RasV12 is coupled with dramatic increase of Wg induction (Figures 7e–h). Induction of Wg expression has an important role during compensatory proliferation,50 and consistent with this view, we found that co-expression of Wg and RasV12 is sufficient to induce an overgrowth phenotype (Supplementary Figures 4N–P). Taken together, these data suggest that Wnd is able to promote cell proliferation and tissue growth by upregulating the expression of CycE and Wg, and collaborates with oncogenic Ras to induce tumor growth and invasion.

Discussion

Maintenance of cell polarity is critical for tissue homeostasis and normal morphogenesis, whereas loss-of-cell polarity is associated with tumor invasion in both fly and mammal.4, 18 The JNK pathway is evolutionarily conserved and has a crucial role in regulating many aspects of tumor progression.39, 51 Mounting evidence suggests that JNK signaling is required for loss-of-cell polarity-triggered tumor invasion,14, 17, 20, 22, 39 yet the molecular links between cell polarity disruption and JNK activation remain largely elusive. By performing a genetic screen using loss of scrib-induced cell invasion phenotype in Drosophila wing epithelia, we have identified the missing link between loss-of-cell polarity and cell invasion, and have shown that Rho1–Wnd signaling connects cell polarity genes and the MKK-JNK cascade (Supplementary Figure 7). We provide in vivo evidence that both Rho1 and Wnd are crucial for cell invasion in Drosophila wing epithelia, and that by forming a complex with Wnd, Rho1 promotes JNK-mediated cell invasion through both Hep and MKK4. This study, though carried out exclusively in Drosophila, also advances our limited understanding toward the upstream regulators of mammalian DLK.52

Interestingly, a recent study suggested a Rho1–ARP (actin remodeling protein)–JNK-positive feedback loop in regulating wing epithelial cell migration.25 In contrast, our data demonstrate that although both cell invasion and MMP1 induction caused by loss-of-cell polarity or Rho1 activation can be dramatically suppressed by blocking Wnd–Bsk signaling, the actin remodeling process could only be impeded by depletion of wnd, but remains unaffected by inactivation of Bsk (Supplementary Figures 4A–E), suggesting that this Wnd function is mediated by downstream factors other than the JNK pathway. Consistent with this notion, Khoo et al.28 have shown that Rho1-Myosin II signaling regulates actin accumulation independent of JNK. As recent studies have linked cytoskeleton regulation with the Hippo pathway,53, 54 and our unpublished data show that Wnd is able to inactivate Hippo signaling, we cannot exclude the possibility that Rho1–Wnd-induced cell invasion also depends on Hippo pathway activity. However, it is worth noting that unlike JNK activation, expression of Hippo pathway transcription co-activator Yorkie (Yki) along the anterior/posterior boundary of wing disc cannot induce a cell invasion phenotype, despite massive MMP1 induction.55 Given that the regulatory mechanism between JNK and Hippo in diverse physiological functions are context dependent,56 it would be interesting to investigate the potential roles of Hippo–Yki signaling in cell invasion and tumor metastasis.

Another promising downstream candidate that modulates the actin remodeling activity of Wnd is p38, which is known to have essential roles in regulating actin polymerization and microtubule dynamics.57, 58 In support of this hypothesis, studies in C aenorhabditis elegans suggest that DLK-1 (homolog of Wnd) requires p38 activity to regulate axon regeneration.59 Further investigation will be of great interest to clarify the role of p38 in Rho1–Wnd signaling-induced actin remodeling and cell invasion.

In this study, we provide the first in vivo evidence that Wnd is both necessary and sufficient for controlling cell invasion, epithelial organization and tumorigenesis. Our genetic data suggest that Wnd is a key regulator of Rho1 and Src42A-induced cell invasion, and that Wnd functions to induce cell invasion and EMT in a JNK-dependent manner. These data raise the possibility that DLK, the mammalian homolog of Wnd, may exhibit a conserved role in regulating cell migration and invasion via a mechanism similar to what we reported in Drosophila. Indeed, although most studies on DLK are focused on neurodegeneration and regeneration,52 some work indicates that DLK may regulate radial migration of neuronal cells during mouse corticogenesis.30 Further investigation is required to clarify the potential role of DLK in tumor cell invasion.

In conclusion, our study in Drosophila has uncovered a novel physiological function for Wnd in modulating cell invasion and EMT, and has identified a molecular link between impaired cell polarity and JNK-mediated cell invasion, which further emphasizes the value of the Drosophila model system to gain insight into human cancer biology. Given that the pathway and components identified in Drosophila are highly conserved, similar molecules and mechanisms could be involved in human cancer progression.

Materials and methods

Drosophila stocks and genetics

All stocks were raised on standard Drosophila media and crosses were performed at 25 °C unless otherwise indicated. For experiments involving tub-Gal80ts, flies were raised at 18 °C to restrict Gal4 activity for 5–6 days, then shifted to 29 °C for 2–3 days to inactivate Gal80ts. The following stocks were used: GMR-Gal4, ptc-Gal4, UAS-p35, UAS-GFP, UAS-Rac1 (#6680), UAS-Rho1 (#7334), UAS-LacZ (#3956), UAS-Rho1-IR (#27727), UAS-wnd-IR (#27525), UAS-lgl-IR (#31089), tub-Gal80ts (Bloomington Stock Center, Bloomington, IN, USA), UAS-Wnd, UAS-WndKD (gifts from Aaron DiAntonio60), pucE69, UAS-dTAK1-IR, UAS-BskDN, UAS-hep-IR, UAS-Puc,61 UAS-wnd-IR,33 UAS-Rho1-IR,32 UAS-Src42A,15 UAS-MKK4-IR,62 UAS-slpr-IR27 and UAS-scrib-IR22 were previously described.

Immunostaining

Third instar larvae wing discs, eye discs and S2 cells were fixed in 4% paraformaldehyde and stained as described previously,63 using mouse anti-MMP1 (1:200), rat anti-DE-cadherin (1:100), mouse anti-β-Gal (1:1000), mouse anti-β-integrin (1:100), mouse anti-Wg (1:300, DSHB; Developmental Studies Hybridoma Bank, Iowa City, IA, USA), rabbit anti-phospho-JNK (1:200, Calbiochem, San Diego, CA, USA), rabbit anti-PH3 (1:100), rabbit anti-active Caspase 3 (1: 400) and Alexa Fluor555 (1:100; Cell Signaling Technology, Danvers, MA, USA), Rat anti-CycE (gift from HE Richardson, 1:200). Secondary antibodies were anti-rabbit-Alexa (1:1000, CST), anti-Rat-Cy3 (1:1000, Jackson Immuno Research, West Grove, PA, USA) and anti-mouse-Cy3 (1:1000, Jackson Immunochemicals, West Grove, PA, USA).

X-gal staining

Eye and wing discs were dissected from third instar larvae in PBST (1x phosphate-buffered saline pH 7.0, 0.1% Triton-X 100) and stained for β-galactosidase activity.

AO staining

Eye and wing discs were dissected from third instar larvae in PBST and incubated in 1 × 10−5 M acridine orange for 5 min at room temperature before imaging.

Plasmids, cell culture and transfection

wnd, Rho1 and Rac1 cDNA were subcloned into pUAST to generate UAS-EGFP-Wnd, UAS-HA-Rho1 and UAS-Flag-Rac1 constructs. S2 cells were cultured in Insectagro DS2 (Corning, Manassas, VA, USA) supplemented with 10% fetal bovine serum (Thermo Scientific, Waltham, MA, USA), 50 U/ml of penicillin and 50 μg/ml of streptomycin. Transfection was performed using Effectene Transfection Reagent (QIAGEN, Valencia, CA, USA) according to manufacturer’s instructions. An actin-Gal4 construct was co-transfected with the pUAST expression vector for all the transfection experiments.

Immunoprecipitation and western blot analysis

S2 cells were transfected with indicated plasmids for 36 h, and immunoprecipitation and Western analyses were performed as previously described.63 In brief, pre-cleared cell lysates were incubated with the indicated antibodies followed by precipitation with protein G sepharose beads (Sigma, St Louis, MO, USA). Immune complexes were washed with lysis buffer, eluted in 2 × SDS sample buffer, and then subjected to Western blot using corresponding antibodies. Antibodies used in this study were as follows: rabbit anti-Wnd antibody (a kind gift from Aaron DiAntonio, St Louis, MO, USA), mouse anti-EGFP, rabbit anti-HA and rabbit anti-Flag (CMCTAG, San Diego, CA, USA), goat anti rabbit/mouse IgG-HRP (CST).

Statistical analysis

Quantification of the data was presented in bar graphs created with Graphpad Prism 5 (San Diego, CA, USA). Data represents mean values+s.d. We used one-way analysis of variance with corrected Bonferroni multiple comparison test to calculate statistical significance in Figures 1 and 6.