Main

Cancer is the leading cause of death worldwide, causing an estimated total of 577 190 deaths in United States alone in 2012.1 Cancer-related mortalities are mainly caused by metastasis, rather than the primary tumor that arises from a malignant lesion.2 Although significant progress has been made towards the understanding of cancer development, the molecular and genetic mechanisms of tumor metastasis have remained poorly understood. With reduced genome redundancy and amenable genetic tools, Drosophila melanogaster has become an excellent model system to investigate the genetic mechanism of cancer biology over the past decades.3, 4, 5, 6 Several invasion and metastasis models have been established in both adult flies and developing larvae.4, 7, 8 In the eye–antennal discs, expression of oncogenic mutants, such as RasV12, can cooperate with mutants that disrupt cell polarity, such as scrib or lethal giant larvae (lgl), to induce invasive tumors into the ventral nerve cord (VNC).9, 10, 11 In the epithelia of the developing wing disc, loss of the cell polarity gene (scrib) or C-terminal SRC kinase (Csk) along the anterior/posterior (A/P) boundary using patched-Gal4 driver produces an invasive migration phenotype,12, 13, 14 which has been used to model cell migration in vivo.3

The c-Jun N-terminal Kinase (JNK) pathway is evolutionarily conserved from fly to human and has an essential role in regulating a wide range of cellular activities including proliferation, differentiation, migration and apoptosis.15 Recently, JNK signaling has been shown to have an important role in modulating RasV12/scrib−/−-triggered tumor growth and metastasis, as well as RafGOF/RhoGEF2-induced tumorgenesis.14, 16, 17

Src was the first discovered oncogene encoding a non-receptor membrane-associated tyrosine kinase.18, 19, 20 Nine Src family members have been identified in mammals, whereas the Drosophila genome encodes only two Src homologs, Src42A and Src64B.21, 22, 23 Src42A and Src64B have a redundant role in regulating dorsal closure and are both required for tracheal cell morphogenesis.24, 25 Recent studies found that Src64B could induce JNK-dependent overgrowth and expression of Yorkie’s (Yki) target genes when cell death was blocked by expressing p35.26 When Src64B was overexpressed in a clone context, it induced non-autonomous tumor overgrowth through JNK-dependent propagation of Yki activity.27 Consistent with the well-documented role of Src family kinases in promoting mammalian tumor invasion,28 Cagan and colleagues found that inhibition of Csk, a negative regulator of Src family kinases, triggered JNK signaling-mediated cell invasion in Drosophila wing discs.13 However, a direct involvement of Src in tumor metastasis and cell invasion, and its underlying mechanisms, remain largely elusive.

We have performed a genetic screen and identified Src42A as a crucial regulator of RasV12/lgl−/−-induced tumor invasion, as well as the loss of scrib-triggered cell migration in development. Our genetic epistasis analysis demonstrated that Src42A modulates JNK signaling-mediated cell invasion, cell death and thorax closure upstream of Bendless (Ben) and dUev1a. Furthermore, we showed that ectopic expression of Ben/dUev1a not only induces cell migration and matrix metalloproteinase 1 (MMP1) upregulation but also cooperates with RasV12 to promote tumor growth and invasive behavior. Together, these data highlight the importance of Src42A-Ben/dUev1a-JNK signaling in regulating cell invasion and extend our knowledge towards the underlying mechanism of Src42A in tumor progression.

Results and discussion

Src42A is required for tumor invasion and cell migration

As previously described, co-expression of oncogenic Ras (RasV12) in lgl-mutant cells in eye–antenna discs using the ey-FLP/MARCM (mosaic analysis with a repressible cell marker) system induces strong tumor-like growths (Figure 1a), with invasive migration into the VNC 8 days after egg laying (AEL) (Figure 1a′).10, 16, 29 Such animals kept growing as oversized larvae carrying huge tumors in their head and died before pupation (Supplementary Figure S1A). Blocking JNK signaling by expressing the JNK phosphatase Puc dramatically suppressed the growth and invasion behaviors of RasV12/lgl−/− tumors (Figures 1b and b′).12, 16 To identify additional genes required for tumor growth and invasion, we performed a genetic screen for dominant suppressors of the RasV12/lgl−/− tumor progression phenotype and identified dUev1a as a crucial regulator.12 From this screen, we also found that the loss of src42A dramatically suppressed tumor cell invasion into the VNC (Figure 1c′) and enabled the animals to survive to the pupa stage (Supplementary Figure S1B), whereas the tumor size remained largely unaffected (compare Figures 1a and c). Loss of src64B produced a similar suppression effect (Supplementary Figure S2), consistent with previous finding that the two Drosophila Src proteins share redundant functions in development.24

Figure 1
figure 1

Src42A is required for Ras-lgl oncogenic cooperation and loss of scrib-induced invasion. (a–c) GFP-labeled clones of cells with indicated genotypes are created in the developing eye–antennal discs. RasV12/lgl−/−-induced tumor growth (a) and invasion of the VNC (a′) were dramatically suppressed by expressing Puc (b–b′). The invasion behavior was also strongly suppressed by expressing an src42A RNAi (c′), whereas the tumor overgrowth was not significantly changed (c). Scale bars, 200 μm in a–c and 100 μm in a′–c′. (d–f) Fluorescence micrographs of wing discs are shown. Compared with the control (d–d′′), loss of scrib-induced cell invasion and MMP1 expression (e–e′′) was suppressed by the expression of an src42A RNAi (f–f′′). Scale bars, 20 μm

Full size image

To examine whether Src42A modulates cell invasion in other contexts, we turned to another well-characterized model in the larval wing imaginal disc.3, 12, 13, 30, 31, 32 Compared with the controls (Figures 1d–d′′), RNAi downregulation of the cell polarity gene scrib driven by ptc-Gal4 triggered strong cell migration phenotype towards the posterior part of the discs (Figure 1e′), along with JNK-dependent upregulation of matrix metalloprotease1 (MMP1, Figure 1e′′),12, 14 a protein that is essential for basement membrane degradation.29, 33 We found that the loss of src42A dramatically suppressed the epithelial migration phenotype as well as the upregulation of MMP1 (Figures 1f–f′′). Taken together, these data indicate that Src42A is required for RasV12/lgl−/−-triggered tumor invasion in the eye discs as well as the loss of cell-polarity-induced cell migration and MMP1 expression in the developing wing.

Src42A induce JNK-mediated cell migration through Ben/dUev1a

Ectopic expression of Src42A along the A/P boundary of wing imaginal discs driven by ptc-Gal4 (ptc>Src42A) resulted in strong cell migration phenotype along with the upregulation of MMP1 (Figures 2b–b′′′). To examine whether cell invasion is a primary consequence of Src42A activation, or a secondary effect of cell death triggered by elevated Src activity,34, 35, 36 we expressed the caspase inhibitor p35 to block cell death. Co-expression of p35 with Src42A resulted in a widening of the ptc domain and enhanced the production of MMP1 (Supplementary Figure S3). Importantly, these ‘undead cells’37, 38 can still migrate to the posterior part (Supplementary Figure S3), indicating that cell migration is a primary event induced by Src42A expression.

Figure 2
figure 2

Src42A induces JNK-mediated cell invasion in developing wings. Fluorescence micrographs of wing discs are shown. Compared with the controls (a–a′′′), ectopic Src42A expression-induced MMP1 secretion (b′) and cell invasion phenotype (b′′′) were completely suppressed by co-expressing of BskDN (c–c′′′). Scale bars, 100 μm in a–c and 20 μm in a′′′–c′′′

Full size image

JNK signaling is misregulated in many human cancers and contributes to invasion in different Drosophila models.7, 13, 29, 31 We found that ptc>Src42A-induced cell migration and MMP1 induction (Figure 2b) were significantly suppressed by the expression of a dominant-negative form of Basket (Bsk) (Figure 2c), indicating that JNK signaling is required for Src42A-induced cell migration. We performed genetic epistasis analysis between Src42A and components of JNK signaling and found that ptc>Src42A-induced migration phenotype was significantly blocked by the loss of JNK kinase Hep, JNK Kinase Kinase dTAK1 and the E3 ubiquitin ligase dTRAF2 (Figure 3c), suggesting that Src42A induces cell migration upstream of dTRAF2.

Figure 3
figure 3

Src42A induces cell invasion through Ben/dUev1a. (a–b) Fluorescence micrographs of wing discs are shown. Ectopic Src42A expression-induced cell invasion (a′) and MMP1 secretion (a′′) were strongly suppressed by co-expressing a ben and dUev1a RNAi (b–b′′). Scale bars, 20 μm. (c) Quantification of invasion phenotype induced by Src42A expression. Results are shown as mean±S.D., n≥30. Student’s t-test was used to compute P-values, significance is indicated with asterisks (***P<0.001, **P<0.01)

Full size image

Our recent work identified the E2 ubiquitin-conjugating enzyme complex consisting of Bendless (Ben) and dUev1a as the upstream regulators of dTRAF2 in JNK signaling.12, 39 We explored the genetic interactions between the Src42A and Ben/dUev1a complex and found that ptc>Src42A-triggered migration phenotype and MMP1 expression (Figures 3a–a′′) were suppressed strongly by RNAi downregulation of either Ben or dUev1a alone (Figure 3c), and completely by both (Figures 3 b–b′′), suggesting that the Ben/dUev1a complex is necessary for Src42A-induced cell migration. We previously showed that co-expression of Ben and dUev1a resulted in JNK activation;12 consistently, Ben and dUev1a expression along the A/P boundary induced a mild cell migration phenotype and upregulation of MMP1 (Figures 4a), which was significantly enhanced by removing one copy of endogenous puc (Figure 4c), indicating that Ben/dUev1a expression could induce JNK-dependent cell migration.

Figure 4
figure 4

Ben/dUev1a expression promotes tumor growth and metastasis of RasV12 cells. (a–d) Fluorescence micrographs of wing discs are shown. Compared with the controls (a), Ben/dUev1a expression induced mild cell invasion phenotype (b) was intensely enhanced by deleting one copy of endogenous puc (c). Ben/dUev1a expression under the ptc promoter upregulates MMP1 secretion. Crosses were performed at 29 °C. Scale bars, 20 μm. (e–g) Dorsal views of the cephalic complexes (e–g) and NVCs are shown (e′–g′). RasV12-expressing clones show a moderate growth advantage (e), but never invade into NVC (e′). Clones with elevated Ben/dUev1a activity gave no obvious overgrowth or invasion phenotypes (f–f′). Massive tumor growth (g) and invasion to the VNC (g) were observed by co-expression of RasV12 and Ben/dUev1a. Scale bars, 200 μm in (e–g) and 100 μm in (e′–g′)

Full size image

Together, the above data suggested that the Ben/dUev1a complex is necessary and sufficient for Src42A-induced JNK-dependent cell migration.

Ben/dUev1a cooperates with RasV12 to induce tumor growth and invasion

Gain of function of cell morphology regulators or loss of cell polarity could promote JNK-dependent cell migration and cooperate with RasV12 to drive tumor growth and invasion.36, 40 As Ben/dUev1a expression could also induce JNK-dependent cell migration, we wonder whether Ben/dUev1a could cooperate with RasV12 to promote tumor progression. As reported previously, clonal expression of RasV12 in eye disc induced benign overgrowth (Figure 4e), with no cells invading into the VNC (Figure 4e′).16 On the other hand, expression of Ben/dUev1a resulted in no obvious growth or invasion phenotype (Figures 4f and f′). When Ben/dUev1a was simultaneously expressed with RasV12, tumor overgrowth (Figure 4g) and invasion into the VNC region (Figure 4g′) were observed in the eye discs 8 days AEL, suggesting that Ben/dUev1a could cooperate with RasV12 to promote tumor growth and invasion behavior.

Src42A activates JNK signaling in vivo

To investigate whether Src42A activates JNK signaling in vivo, we examined JNK phosphorylation and puc expression, a read-out of JNK activity,41 in response to Src42 expression. We found that ectopic expression of Src42A under the ptc promoter in the wing disc resulted in strong JNK phosphorylation (compare Figures 5a and b) and expression of a puc-LacZ reporter (Figures 5c and d). Furthermore, ectopic Src42A expression driven by GMR-Gal4 is sufficient to activate puc transcription in developing eye discs (Figures 5e and f). Together, these results indicate that Src42A can activate JNK signaling in vivo.

Figure 5
figure 5

Src42A expression activates JNK in vivo. (a–b) Fluorescence micrographs of wing discs are shown. Compared with the controls (a), Src42A expression induced strong JNK phosphorylation (b). Scale bars, 100 μm. (c–f) X-Gal staining of the puc-LacZ reporter gene in the developing wing (c–d) and eye (e–f) are shown. Ectopic Src42A expression under the ptc promoter strongly upregulated puc expression along the A/P boundary (d). The weak endogenous puc expression posterior to the morphogenetic furrow of eye disc (e) was significantly enhanced by expressing a wild-type Src42A (f)

Full size image

Src42A triggers JNK-dependent cell death in the developing eye

Apart from the role of Src42A in regulating tumor invasion and cell migration showed above, we and others35, 36 also found that ectopic Src42A expression in developing eyes could induce extensive cell death, as shown with acridine orange (AO) staining (Supplementary Figure S4B) that detects dying cells42 and produce a small eye phenotype (Figure 6b). Consistent with the genetic epistasis analysis in cell invasion, the GMR>Src42A-triggered cell death and small eye phenotype were dramatically suppressed by inactivation of Ben, dUev1a or downstream components of the JNK signaling (Supplementary Figure S4 and Figures 6c–g). Besides, the ectopic Src42A-caused small eye phenotype was also reverted by co-expressing two independent RNAi lines against src42A (Figure 6h and Supplementary Figure S5), which further confirmed the efficiency of the two RNAi lines.

Figure 6
figure 6

Src42A induces Ben/dUev1a-mediated cell death in developing eyes. Light micrographs of Drosophila adult eyes are shown. Compared with the wild-type eye (a), the GMR>Src42A small eye phenotype (b) was dramatically suppressed by the expression of BskDN (c), a hep RNAi (d), dTAK1DN (e), a dTRAF2 RNAi (f), reducing ben and dUev1a activity (g), or a src42A RNAi (h)

Full size image

Ben/dUev1a is physiologically required for Src42A

Apart from its role in modulating cell migration and cell death, src42A also regulates JNK-dependent dorsal closure.24 Consistently, JNK signaling, as revealed by puc expression, is activated in the dorsal patch of the third instar larval wing disc (Figure 7a).43 Downregulation of src42A under the pannier (pnr) promoter resulted in reduced puc expression in the wing disc (Figure 7b) and produced a cleft phenotype in the adult thorax (Figure 7d). This cleft phenotype resembled that of JNK inactivation22 and could be partially rescued by deleting one copy of endogenous puc (Figure 7f), or expression of a wild-type Hep (HepWT) (Figure 7h), or co-expression of Ben and dUev1a (Figure 7j). Together, these results suggested that the Ben/dUev1a complex is required for the physiological function of Src42A in thorax development.

Figure 7
figure 7

Src42A modulates endogenous JNK activity upstream of Ben/dUev1a. (a–b) X-Gal staining of the puc-LacZ reporter gene in the developing wing are shown. The endogenous puc expression pattern in the notum region of wild-type wing disc (a) was reduced by expressing a src42A RNAi ((b) indicated by the arrow head). (c–j) Light micrographs of Drosophila adult thoraxes are shown. Compared with the controls (c), loss of src42A-produced strong thorax cleft phenotype (d) was restored by deleting on copy of puc (f), overexpression of Hep (h) or overexpression of Ben/dUev1a (j), whereas loss of puc, expression of Hep or Ben/dUev1a alone gave no obvious phenotype (e, g, i respectively). (k) A schematic diagram of Ben/dUev1a-JNK signaling in regulating Src42A-induced cell migration and death

Full size image

Conclusion

We have identified Src42A as a key modulator of both RasV12/lgl−/−-triggered tumor invasion and loss of cell-polarity-induced cell migration in Drosophila. Our genetic evidence established the Ben/dUev1a complex as an essential positive regulator that mediates Src42A-induced, JNK-dependent cell migration and death (Figure 7k). We showed that ectopic Ben/dUev1a expression not only induced MMP1 upregulation and cell migration but also cooperated with RasV12 to promote tumor growth and invasion, suggesting that Ben and dUev1a are candidate proto-oncogenes. Consistent with our findings, Uev1A, the mammalian homolog of dUev1a, is upregulated in many human cancer cell lines.44, 45 Our data suggested a link between Src activation and Ben/dUev1a-JNK signaling in regulating cell migration and death in development and disease, which provided beneficial information for mammalian studies and target therapies for cancer.

Materials and Methods

Drosophila strains and generation of clones

All stocks were raised on standard Drosophila media, and crosses were performed at 25 °C unless otherwise indicated. Fluorescently labeled invasive tumors were produced in the eye discs as previously described10 using the following strains: y, w, ey-Flp; tub-Gal80, FRT40A; Act>y+>Gal4, UAS-GFP (40A tester), lgl4 FRT40A UAS-RasV12 (40A tester) and ey-Flp, Act>y+>Gal4, UAS-GFP. Additional strains, including GMR-Gal4, ptc-Gal4, pnr-Gla4, UAS-p35, UAS-GFP and UAS-Src42ACA, were obtained from Bloomington stock center; UAS-ben-IR (no. 9413) and UAS-src42A-IR (no. 26019) were obtained from the VDRC center; UAS-dTAK1-IR, UAS-BskDN, UAS-hep-IR, UAS-dTRAF2-IR, UAS-Puc, pucE69,46 UAS-src42A-IR,25 UAS-src64B-IR,47 UAS-Src42A,35 UAS-BenT8, UAS-dUev1a, UAS-dUev1a-IR and UAS-scrib-IR12, 39 were previously described. Genotypes of flies used in each figure can be found in online Supplementary Information.

Immunohistochemistry

Antibody staining of imaginal discs was performed as previously described.16 The following antibodies were used: rabbit anti-phospho-JNK (1 : 200, Calbiochem, San Diego, CA, USA), mouse anti-MMP1 (1 : 200, Developmental Studies Hybridoma Bank, Iowa City, IA, USA). Secondary antibodies were anti-rabbit-Alexa (1 : 1000, Cell Signaling Technology, Danvers, MA, USA) and anti-mouse-Cy3 (1 : 1000, Jackson Immuno Research, West Grove, PA, USA).

X-gal staining

Eye and wing discs were dissected from the third instar larvae in PBT (1 × PBS pH 7.0, 0.1% Triton X 100) and stained for β-galactosidase activity as described.48

AO staining

Eye and wing discs were dissected from the third instar larvae in PBT and incubated in 1 × 10−5 M AO for 5 min at room temperature prior to imaging.