Oncogenic Ras stimulates Eiger/TNF exocytosis to promote growth

Abstract

Oncogenic mutations in Ras deregulate cell death and proliferation to cause cancer in a significant number of patients. Although normal Ras signaling during development has been well elucidated in multiple organisms, it is less clear how oncogenic Ras exerts its effects. Furthermore, cancers with oncogenic Ras mutations are aggressive and generally resistant to targeted therapies or chemotherapy. We identified the exocytosis component Sec15 as a synthetic suppressor of oncogenic Ras in an in vivo Drosophila mosaic screen. We found that oncogenic Ras elevates exocytosis and promotes the export of the pro-apoptotic ligand Eiger (Drosophila TNF). This blocks tumor cell death and stimulates overgrowth by activating the JNK-JAK-STAT non-autonomous proliferation signal from the neighboring wild-type cells. Inhibition of Eiger/TNF exocytosis or interfering with the JNK-JAK-STAT non-autonomous proliferation signaling at various steps suppresses oncogenic Ras-mediated overgrowth. Our findings highlight important cell-intrinsic and cell-extrinsic roles of exocytosis during oncogenic growth and provide a new class of synthetic suppressors for targeted therapy approaches.

INTRODUCTION

Activating mutations of the oncogenic Ras gene are highly prevalent in human cancers (Bos, 1989). However, targeted therapy strategies have not yielded the desired effects and the available chemotherapy regimens are not effective at treating the most aggressive Ras cancers (Asghar et al., 2010; Lièvre et al., 2010). Genetic studies in fruit flies and worms have led to the identification of Ras effectors and the characterization of the Ras/Raf/mitogen activated protein kinases (Ras/MAPK) signaling cascade during normal development (Nishida and Gotoh, 1993; Simon et al., 1991; Sternberg and Horvitz, 1991). This pathway was found to be conserved in mammals (Alessi et al., 1994; Cowley et al., 1994; Kolch et al., 1991; Mansour et al., 1994; Qiu et al., 1995). However, targeting components of the Ras/MAPK signaling cascade only partly inhibits overgrowth (Blum and Kloog, 2005), suggesting that oncogenic Ras drives tumorigenesis via additional signaling events. Moreover, blocking the Ras/MAPK pathway causes death of normal cells, thus making this approach not suitable for targeted therapy and placing an impetus on the need to identify the additional signaling events that oncogenic Ras specifically triggers to promote tumor development. Identification of mutations that synthetically suppress Ras tumor growth could not only broaden our understanding of cancer biology, but could also lead to the discovery of novel therapeutic targets. Indeed, several laboratories have conducted RNA interference (RNAi)-based synthetic suppressor screens in tissue culture settings and have identified important vulnerabilities of oncogenic Ras cells (Barbie et al., 2009; Luo et al., 2009; Sarthy et al., 2007; Scholl et al., 2009; Steckel et al., 2012; Strebhardt and Ullrich, 2006). Additional whole-animal synthetic suppressor screens could be particularly informative for revealing the role of oncogenic Ras in processes that are important for regulating growth in vivo.

Expression of oncogenic Ras (Ras^V12) in Drosophila imaginal discs gives rise to overgrowth (Karim and Rubin, 1998). Generating patches of labeled Ras^V12-expressing cells surrounded by wild-type cells allows putative host-tumor cell interactions to occur and permits genetic screens to identify mutations that can either enhance or suppress the growth of Ras^V12-expressing tumors (Chi et al., 2010; Pagliarini and Xu, 2003). Here, we report the characterization of one of the synthetic suppressor mutations, sec15, a mutation in a gene that encodes a component of the exocytosis machinery. Sec15 is a subunit of the evolutionarily conserved multiprotein complex termed the Sec6/Sec8 complex or the exocyst complex, which was originally identified in yeast (Finger et al., 1998; Finger and Novick, 1998; Guo et al., 1999; Heider and Munson, 2012; Novick et al., 1980; TerBush and Novick, 1995). The exocyst complex regulates secretion in eukaryotes by controlling the delivery of vesicles to the cell plasma membrane (Grindstaff et al., 1998; Guo et al., 1999; Heider and Munson, 2012; Jafar-Nejad et al., 2005; Mehta et al., 2005; TerBush and Novick, 1995). In contrast to core exocyst components, which are broadly required for normal exocyst function and cell viability, Sec15 regulates the delivery of specific cargo proteins and is dispensable for cell viability in Drosophila (Mehta et al., 2005).

In addition to being essential for cellular organization in all eukaryotes, vesicle transport has recently been found to play important roles in regulating signal transduction. For example, transport of endocytosed cell surface molecules to signaling targets on endosomes allows signal transduction to occur, whereas targeting these molecules to the lysosome for degradation attenuates or suppresses signaling (Seto et al., 2002). Transcytosis of vesicles facilitates the establishment of morphogen gradients, which are crucial for conveying proliferation and cell fate specification cues during development (Seto et al., 2002). Exocytosis has been previously found to mediate signal transduction by sending signaling molecules including neurotransmitters and ligands to neighboring cells (Li and Chin, 2003). By studying how sec15 suppresses Ras^V12, we show that exocytosis also regulates signal transduction within a cell by clearing signaling ligands. We found that Ras^V12 cells clear Eiger (also known as TNF) by exocytosis to downregulate pro-apoptotic Janus NH2-terminal kinase (JNK, also known as Bsk – FlyBase) signaling (Igaki et al., 2009; Moreno et al., 2002) and thus evade cell death. We have previously shown that JNK activation triggered by cell polarity defects could stimulate non-autonomous JAK-STAT signaling for proliferation (Wu et al., 2010). Here, we show that oncogenic Ras elevates exocytosis to hijack this process in order to promote overgrowth. Exocytosis-dependent accumulation of Eiger/TNF results in JNK activation in surrounding wild-type cells, which in turn, non-autonomously stimulates JAK-STAT signaling to promote the proliferation of Ras^V12 cells. These findings provide new mechanistic insights into the long known ability of oncogenic Ras cells to avoid cell death and promote growth, and also highlight the importance of exocytosis in signal transduction and cancer biology.

RESULTS

sec15 synthetically interacts with oncogenic Ras

In Drosophila, GFP-marked mosaic clones of cells expressing Ras^V12 overgrow to develop into tumors (Pagliarini and Xu, 2003). The overgrowth phenotype can be readily ascertained by visualizing fluorescent signal intensity in third instar whole larvae (Fig. 1A,C) or by examining clone size in dissected eye-antenna imaginal discs (Fig. 1E,G). Furthermore, Ras^V12 tumors caused pupal lethality (98.4%, N=62; Fig. 1K). We induced sec15 or Ras^V12 single mutant clones or Ras^V12, sec15 double mutant clones and examined the growth of these mutant clones in similarly aged third instar eye-antenna discs. We found that the sec15 mutation did not disrupt cell proliferation (supplementary material Fig. S1A,B), and the size of sec15 mutant clones was comparable to that of wild-type clones (Fig. 1A,B,E,F), consistent with the reported cell viability of the sec15 null mutation (Mehta et al., 2005). In addition, sec15 null mutant cells persisted into the adult eye (Fig. 1I,J). The viability of sec15 mutant cells is not due to maternal protein deposition, as Sec15 protein level was dramatically reduced in mutant clone cells (supplementary material Fig. S2A). However, the sec15 mutation dramatically suppressed the overgrowth phenotype of Ras^V12 clones (Fig. 1C,D,G,H; 77.4% of the double mutants showed strong suppression similar to that shown in Fig. 1D,H; N=53). Furthermore, the sec15 mutation rescued the lethality of the animals bearing Ras^V12 tumors (76% viable, N=125, compared with 1.6% viable, N=62 for Ras^V12 animals; Fig. 1L). Moreover, RNA interference (RNAi)-mediated knockdown of Sec15 in Ras^V12, scrib⁻ cells suppressed tumor growth and invasion (supplementary material Fig. S8A-D). Finally RNAi knockdown of two core exocyst components, Sec6 and Sec8 showed a similar effect on Ras^V12-mediated overgrowth (supplementary material Fig. S1C-N). The sec15 mutant or Sec15 or Sec8 RNAi alone had no detectable effect on growth, whereas Sec6 RNAi alone showed a reduction in clone sizes (supplementary material Fig. S1C-N). Taken together, we conclude that the sec15 mutation synthetically suppresses Ras^V12 tumor growth.

Fig. 1.

Ras^V12 and sec15 synthetic lethal interaction. (A-D) Intact third instar larval cephalic regions showing wild-type (WT), sec15, Ras^V12 and Ras^V12, sec15 double mutant eye-antenna disc clones. Wild-type (A) and sec15 (B) clones are comparable in size, whereas Ras^V12 clones (C) overgrow to develop tumors. The sec15 mutation suppresses Ras^V12 tumor growth (D). (E-H) Similarly aged eye discs bearing wild-type, sec15, Ras^V12 and Ras^V12, sec15 double mutant clones. Wild-type (E) and sec15 (F) clones show a comparable sparse pattern, whereas Ras^V12 clones (G) overgrow to form large contiguous tumors. However, Ras^V12, sec15 double mutant clones are sparse and smaller (H). (I-L) Adult eyes showing lethality and contribution of larval clones to the eye field. Wild-type (I) and sec15 (J) clones similarly persist into the adult eye and do not cause animal lethality. Ras^V12 clones (K) are pupal lethal but animal lethality is suppressed in animals bearing Ras^V12, sec15 double mutant clones, and cells of this genotype now contribute to the adult eye (L).

Oncogenic Ras stimulates the exocyst

Interestingly, it was previously observed that RNAi depletion of exocytosis proteins also suppresses HRas^V12-mediated tumor growth in human cells (Issaq et al., 2010). We thus investigated the mechanism underlying this phenomenon. The observation that sec15 can selectively suppress Ras^V12 tumor growth suggested that oncogenic Ras could regulate the exocyst to promote growth. We first analyzed the abundance of exocyst proteins in Ras^V12 tumor clones relative to that of wild-type cells. We examined Sec15 protein levels in Ras^V12 tumors and extended this analysis to include Sec6, Sec8 and Rab11. Rab11 interacts with Sec15 to initiate assembly of the exocyst and the exocytosis of endocytosed molecules (Wu et al., 2005; Zhang et al., 2004). The previously published Sec15, Sec6 and Sec8 antibodies showed negligible staining in corresponding mutant and RNAi knockdown clones (supplementary material Fig. S2A-C). Sec15, Sec6 and Sec8 proteins were specifically upregulated in Ras^V12clones (Fig. 2A-C,E,G). Similarly, Rab11 was upregulated in Ras^V12 cells (Fig. 2D,F). The increased levels of exocyst proteins in Ras^V12 cells were subsequently confirmed by western blotting (Fig. 2H). Exocyst proteins displayed no obvious subcellular localization defects in Ras^V12 cells (Fig. 2E-G), indicating that oncogenic Ras stimulates exocyst proteins but does not affect their respective subcellular localization.

Fig. 2.

Ras^V12 cells upregulate exocyst proteins. (A-G) Ras^V12 clones stained with antibodies against exocyst proteins (Sec15, Sec6, Sec8 and Rab11) and co-stained with DAPI (blue) to mark nuclei. Lower panels show Sec15, Sec6, Sec8 and Rab11 channels alone. Sec15 is upregulated in Ras^V12 cells (A). Similarly, Sec6 (B), Sec8 (C) and Rab11 (D) are upregulated in Ras^V12 clones but not in neighboring wild-type cells. Green dotted lines denote clone boundaries. (E-G) x/z confocal projection images of Ras^V12 clones (green) stained with anti-Sec8, anti-Rab11 and anti-Sec15, and with DAPI. Sec8 (E), Rab11 (F) and Sec15 (G) proteins display no obvious localization defects but show increased abundance in Ras^V12 cells. (H) Lysate from discs bearing wild-type (WT) or Ras^V12 clones immunoblotted with anti-Sec6, anti-Sec8, anti-Sec15 or anti-Tubulin antibodies (loading control). Sec6, Sec8 and Sec15 protein abundance is increased in Ras^V12 lysates.

The observation that Rab11 is upregulated along with core exocyst proteins suggested that oncogenic Ras elevates the secretion of endocytosed molecules. To begin to address this possibility, we sought to examine the transport of endocytosed molecules in the mutant cells by performing Dextran dye pulse-chase experiments. After the dye pulse phase we found that Ras^V12 cells had no problem taking up the dye as they contained more dye (dye-positive punctae) than control cells [4.05±1.47 for Ras^V12 versus 2.29±0.65 for wild-type cells (means±s.d.); Fig. 3A,B,G], indicating that endocytosis is generally not inhibited in Ras^V12 cells. Examining dye abundance in mutant versus control cells after the dye chase period provided a crude measure of secretion within these cells. Interestingly, Ras^V12 cells were significantly more efficient at dye clearance after the dye chase phase. Ras^V12 cells showed fewer dye-positive punctae compared with wild-type cells (the pulse:chase ratios of dye-positive punctae for Ras^V12 and wild-type cells were 4.63±0.27 and 1.53±1.12, respectively; Fig. 3D,E,G). The sec15 mutation abrogated the increased dye clearance effect of oncogenic Ras [2.74±1.98 (after the pulse), 1.26±0.29 (after chase) and 2.18±0.34 (pulse:chase ratio); Fig. 3C,F,G]. Taken together, the above data indicate that oncogenic Ras stimulates the exocyst.

Fig. 3.

Ras^V12 cells stimulate secretion through Sec15. (A-F) Projections of confocal images showing wild-type (WT), Ras^V12 or Ras^V12, sec15 mutant clones (green, middle panels) after pulse (A-C) and followed by a chase period (D-F) of Dextran dye (red, lower panels). Samples are co-stained with DAPI (blue) to mark nuclei. Middle panels show the clone channel alone. The dotted green lines in the lower panels denote clone boundaries. Ras^V12 cells (B) show no problem in dye uptake. Cellular clearance of the dye is accelerated in Ras^V12 cells (E, lower panel) compared with that of similarly treated wild-type control cells (D, lower panel). The sec15 mutation suppresses the ability of Ras^V12 cells to clear out the dye (C,F). (G) Quantification of A-F. The number of dye-positive vesicles after each experimental phase (pulse and chase) was scored in multiple clones for each genotype. Pulse:chase ratios were derived from averages of dye-positive vesicle numbers after each experimental phase. Error bars represent standard deviation from the mean for each genotype analyzed. In this and all subsequent figures, numbers within bars indicate the number of cells analyzed per corresponding genotype.

Oncogenic Ras promotes the interaction of Eiger/TNF with the exocyst

We next sought to investigate the significance of exocyst stimulation vis-à-vis tumor growth. One way in which Ras activation contributes to tumorigenesis is by allowing the cell to evade cell death (Cox and Der, 2003; Downward, 1998; Wolfman et al., 2002; Wu et al., 2009). In Drosophila, the tumor necrosis factor (TNFα) homolog Eiger triggers cell death (Igaki et al., 2009; Moreno et al., 2002). Similar to vertebrate TNFs, Eiger is produced as a transmembrane protein and is subsequently cleaved to yield a soluble ligand (Black et al., 1997; Brandt et al., 2004; Jue et al., 1990; Kauppila et al., 2003; Moss et al., 1997). Eiger/TNF is endocytosed to activate JNK signaling on endosomes, which triggers cell death (Igaki et al., 2009; Moreno et al., 2002). Preventing endocytosis impinges on Eiger/TNF signaling (Igaki et al., 2009). Because endocytosis is not generally inhibited in Ras^V12 cells and sec15 suppresses oncogenic Ras-mediated overgrowth, we hypothesized that oncogenic Ras could promote an interaction between the exocyst and Eiger/TNF in order to accelerate cellular clearance of Eiger/TNF, hence preventing Eiger/TNF signaling from directly killing tumor cells. We first examined whether Eiger/TNF and Sec15 proteins interact in Ras^V12 cells using two independent and complementary approaches, immunostaining and biochemistry. We found that Eiger/TNF colocalized with Sec15 in Ras^V12 cells (Fig. 4A-C). Sectioning of isosurface-rendered images from confocal projections revealed that indeed Sec15 coats Eiger/TNF-positive vesicles (Fig. 4D), suggesting that Eiger/TNF is a Sec15 cargo protein. Moreover, we performed anti-Eiger or anti-GFP control pulldown experiments on lysate prepared from eye-antenna discs bearing Ras^V12 clones and found that Sec15 and other exocyst proteins, Sec5 and Sec8, specifically co-precipitated with Eiger/TNF (Fig. 4M). We conclude that oncogenic Ras promotes the interaction of the exocyst with Eiger/TNF.

Fig. 4.

Sec15 interacts with Eiger/TNF. (A-D) Ras^V12 cells counterstained with anti-Sec15 (red) and anti-Eiger (green). Sec15 and Eiger colocalize (A; side panels represent magnified view from white box). B and C are single channels for Sec15 and Eiger from A. (D) Cross-section images of isosurface-rendered confocal projection images showing Eiger (green) inside Sec15-coated (red) vesicles. (E-H) Ras^V12 cells counterstained with anti-Rab11 (red) and anti-Eiger (green). Rab11 and Eiger colocalize (E; side panels represent magnified view from white box). F and G are single channels for Rab11 and Eiger from E. (H) Cross-section of isosurface-rendered confocal projection showing Eiger (green) inside Rab11-coated (red) vesicles. (I-L) Ras^V12 cells co-expressing Rab5-GFP (red) counterstained with anti-Eiger (green). Rab5-GFP and Eiger colocalize (I; side panels represent magnified view from white box). J and K are single channels for Rab5-GFP and Eiger, respectively. (L) Cross-section of isosurface-rendered confocal projections showing Eiger (green) inside Rab5-GFP-coated (red) vesicles. (M) Anti-Eiger or anti-GFP (control) immunoprecipitation experiments with lysate derived from tissues bearing Ras^V12 clones. Immunoprecipitates (IP) were blotted with antibodies against Sec15, Sec8, Sec5 and α-Tubulin (loading control).

Oncogenic Ras promotes Eiger/TNF exocytosis to prevent tumor cell death

Endocytosed molecules destined for secretion are shuttled to the cell cortex in Rab11-positive endosomes (Dollar et al., 2002; Sonnichsen et al., 2000; Ullrich et al., 1996). We thus tested whether Eiger/TNF localizes to early (Rab5-positive) endosomes and to recycling (Rab11-positive) endosomes. We found that Eiger/TNF colocalized with both Rab5 and Rab11 endosomes (Fig. 4E-L). Together with the Eiger/TNF-exocyst interaction data above, these observations supported the notion that Eiger/TNF is being cleared out of Ras^V12 tumors by exocytosis. To directly test this, we examined Eiger/TNF protein distribution in similarly aged Ras^V12 or Ras^V12, sec15 mutant clones. We found that Eiger/TNF accumulated at the margins of Ras^V12 clones (Fig. 5A-C), consistent with the Eiger/TNF protein being exocytosed.

Fig. 5.

Sec15-dependent Eiger exocytosis suppresses JNK-induced cell death in Ras^V12 cells but promotes JNK activation in the surrounding wild-type cells. (A-Y) Comparable z-sections showing Ras^V12 or Ras^V12, sec15 double mutant clones or Ras^V12, eiger-RNAi clones (green) stained with anti-Eiger (red) or anti-phospho-JNK (pJNK, red) to detect JNK signaling activity or anti-active caspase 3 to detect apoptotic cells, and with DAPI (blue) to mark DNA. Gray panels show Eiger, phospho-JNK or caspase 3 channels alone. The dotted green lines denote clone boundaries. C,J and Q are z-section views. Eiger is concentrated at the boundary of Ras^V12 clones (A-C) but not at the boundary of Ras^V12, sec15 (D,E) or Ras^V12, eiger-RNAi (F,G) clones. Eiger is moderately retained in sec15 clones (V,W) but is present at higher levels in Ras^V12, sec15 double mutant clones (D,E). Wild-type cells surrounding Ras^V12 clones (H-J), but not wild-type cells surrounding Ras^V12, sec15 (K,L) or Ras^V12, eiger-RNAi (M,N) clones show elevated phospho-JNK levels. Ras^V12, sec15 double mutant clones show elevated phospho-JNK levels (K,L). Correspondingly, cell death can be detected in wild-type cells surrounding Ras^V12 clones (O-Q) but not in wild-type cells surrounding Ras^V12, sec15 double mutant clones. Instead, cell death is concentrated inside the double mutant clones (R,S). (T,U) Confocal projection showing discs containing Ras^V12, sec15, bsk-DN triple mutant clones stained with DAPI to mark nuclei and stained for active caspase-3 to detect apoptotic cells. The caspase channel alone is shown in U. (X,Y) sec15 mutant cells stained with DAPI and anti-phospho-JNK. (Z) Quantification of caspase-positive cells in Ras^V12 single or Ras^V12, sec15 double or Ras^V12, sec15, bsk-DN triple mutant clones. Error bars denote standard deviations from the mean for each genotype analyzed. P values are derived from a two-tailed distribution with two-sample equal variance t-test. Scale bars: 5 μm.

Importantly, examination of Ras^V12, sec15 double mutant clones revealed that Eiger/TNF no longer accumulated at the boundary, but rather accumulated within the clones (Fig. 5D,E). The sec15 mutant clones showed a moderate retention of Eiger/TNF (Fig. 5V,W). We then depleted Eiger/TNF specifically in the Ras^V12 cells and found that this prevented the accumulation of Eiger/TNF that is otherwise seen as a ring around the clones (Fig. 5F,G).

Furthermore, the cellular release of soluble TNF from its precursor transmembrane form and its resulting activity are mediated by the TNF-converting enzyme, Tace (Black et al., 1997; Blobel, 1997; Moss et al., 1997). Eiger/TNF contains a Tace cleavage site equivalent to TNFα cleavage site in vertebrates (Kauppila et al., 2003; Narasimamurthy et al., 2009). We first verified that Drosophila Tace plays a similar role in Eiger/TNF signaling. First, we found that RNAi knockdown of the sole Drosophila Tace in the developing eye suppressed the Eiger/TNF-mediated small-eye phenotype (supplementary material Fig. S3A-D). Similar results were obtained in wing imaginal discs (data not shown). We subsequently tested for a role for Tace in Eiger/TNF non-autonomous signaling using the hanging-eye phenotype model (Narasimamurthy et al., 2009). We found that coexpression of eiger and Tace in cells rescued by expression of the dominant-negative JNK allele bsk-DN in the developing eye (GMR>egr,Tace,bsk-DN) produced a hanging-eye phenotype (supplementary material Fig. S3F,F′). The majority of GMR>egr,Tace,bsk-DN animals die in late pupal stages (73%, N=157) and show a ring of necrotic tissue around the eyes (supplementary material Fig. S3F″). Interestingly, the GMR>egr,bsk-DN animals did not have a dramatic hanging-eye phenotype (supplementary material Fig. S3E). These data indicate that Tace plays an important role in the secretion and signaling of Eiger. Finally, we knocked down Tace by RNAi in the Ras^V12 cells and found that this abolished Eiger/TNF accumulation around the Ras^V12 clones (supplementary material Fig. S4). Collectively, these data indicate that the pool of Eiger/TNF seen around Ras^V12 clones originates from Ras^V12 cells, and Sec15 plays an important role in Eiger/TNF protein export.

Eiger/TNF normally activates JNK signaling to trigger cell death (Igaki et al., 2009; Moreno et al., 2002). Correspondingly, we found that JNK activity was elevated inside the Ras^V12, sec15 double mutant clones compared with that of Ras^V12 or sec15 mutant clones (Fig. 5H,L,X,Y). Increased JNK signaling correlated with increased ectopic cell death inside the Ras^V12, sec15 double mutant clones compared with Ras^V12 single mutant clones, as measured by activated caspase-3 immunoreactivity, a cell death indicator (58.74±19.48%; N=257 versus 6.99±4.32%; N=699 caspase-positive cells, respectively; mean±s.d.; Fig. 5O-Q versus Fig. 5R,S; Fig. 5Z). Furthermore, concomitant suppression of JNK signaling, by expressing the dominant-negative JNK allele bsk-DN, restored viability to Ras^V12, sec15 clones (3.85±5.8%; N=408 caspase-positive cells; Fig. 5T,U,Z). We conclude that Sec15-dependent exocytosis of Eiger/TNF downregulates JNK to protect Ras^V12 cells from apoptosis.

Eiger/TNF activates JAK-STAT signaling non-autonomously

Although blocking JNK signaling in Ras^V12, sec15 double mutant clones suppresses cell death, it fails to fully restore Ras^V12 tumor growth (supplementary material Fig. S5A,B). This suggests that Sec15 contributes to Ras^V12 tumor growth by an additional mechanism. Similar to vertebrate TNF, Eiger can act non-autonomously (Cordero et al., 2010; Pérez-Garijo et al., 2013). In addition, it has been shown that JNK signaling can non-autonomously stimulate the proliferation of adjacent cells (Ryoo et al., 2004; Wu et al., 2010), thus we wondered whether the Sec15-dependent accumulation of Eiger/TNF in the surrounding wild-type cells activates JNK to promote Ras^V12 tumor growth. We tested whether JNK is activated in the wild-type cells surrounding Ras^V12 clones. JNK activity was elevated in wild-type cells surrounding Ras^V12tumors (Fig. 5H-J), but not in wild-type cells surrounding the Eiger/TNF-secretion-defective Ras^V12, sec15 double mutant clones or the Eiger/TNF RNAi-depleted Ras^V12 clones (Fig. 5K-N). In agreement with this and previous findings (Karim and Rubin, 1998), we could detect activated caspase-positive cells around Ras^V12 tumors (Fig. 5O-Q). Therefore, increased Eiger/TNF secretion by Ras^V12 cells leads to Eiger/TNF accumulation and activation of JNK signaling in surrounding wild-type cells. This is consistent with earlier reports of Eiger/TNF activating JNK non-autonomously (Cordero et al., 2010; Pérez-Garijo et al., 2013).

We have previously shown that JNK activation triggered by cell polarity defects can induce the expression of JAK-STAT ligands encoded by the unpaired (upd) genes. Upd ligands cooperate with Ras^V12 through non-autonomous JAK-STAT activation (Wu et al., 2010). Blocking JNK signaling in Ras^V12 cells has no obvious effect on their proliferation (Igaki et al., 2006), supporting the possibility that the observed JNK activation in surrounding wild-type cells could promote Ras^V12 tumor growth by a non-cell autonomous mechanism. We thus wondered whether Eiger/TNF-exocytosis-mediated JNK activation in the surrounding cells promotes the growth of Ras^V12 cells. We first examined Upd expression levels using the Upd-lacZ reporter line (Sun et al., 1995) and found that Upd was particularly up-regulated in wild-type cells surrounding Ras^V12 clones (supplementary material Fig. S6D). Next, we used a JAK-STAT activity GFP reporter (Bach et al., 2007), which includes the promoter fragment of Socs36E, a transcriptional target of JAK-STAT signaling (Karsten et al., 2002), to examine JAK-STAT activity level in Ras^V12 cells. JAK-STAT was upregulated in Ras^V12 cells and in the surrounding Upd-producing wild-type cells (Fig. 6A). This increase was more robust in larger Ras^V12 clones than in smaller ones (supplementary material Fig. S6B, arrow versus boxed area). Similarly, Stat92E immunoreactivity, which is an indicator of JAK-STAT pathway activation (Chen et al., 2002; Johansen et al., 2003), was elevated in Ras^V12 discs (supplementary material Fig. S6F). In a complementary experiment, we found that the transcription level of Socs36E was elevated in Ras^V12 discs compared with that of wild-type control discs (Fig. 6F), consistent with the STAT-GFP and Stat92E results. This indicates that JAK-STAT signaling is activated in Ras^V12 cells.

Fig. 6.

Eiger/TNF exocytosis activates JAK-STAT signaling through JNK. (A) The JAK-STAT activity GFP-reporter line shows pathway activation in Ras^V12 clones. The lower panel shows STAT-GFP alone. (B) dTAK1-rescued Ras^V12 clones (red) induced in dtak1 hemizygous mutant discs do not show JAK-STAT GFP (green) induction. Lower panel shows STAT-GFP alone. The red outlines show clone boundaries in A and B. (C,D) Ras^V12 or Ras^V12, dom-RNAi double mutant clones stained with DAPI to mark nuclei and with antibody against the mitotic marker, phospho-histone-3 (PH3). The majority of Ras^V12 clones are phospho-histone-3 positive (C). The lower panel of C represents a magnified view of DAPI and phospho-histone-3 from the boxed area in the upper panel. Ras^V12, dom-RNAi clones (D) show reduced phospho-histone-3 staining. The lower panel shows the phospho-histone-3 channel alone. The green outlines show clone boundaries in D. Scale bars: 5 μm. (E) Quantification of C and D. (F) RT-PCR experiment measuring Socs36E levels relative to expression levels of the housekeeping gene Rp49. Normalized fold differences in Socs36E expression levels between Ras^V12 and wild-type (WT) or between Ras^V12, eiger-RNAi (eiger RNAi-knockdown specifically in RasV12 clones) and Ras^V12 or between Ras^V12, sec15 double mutant and Ras^V12 clones are shown. Experiments were performed in triplicate and standard deviations were derived from the coefficient variations of experimental and control samples.

To test whether the observed activation of JAK-STAT in Ras^V12 cells is due to increased Eiger/TNF exocytosis to neighboring wild-type cells and the resulting JNK activation there, we first blocked Eiger/TNF exocytosis to the surrounding wild-type cells and examined JAK-STAT activation. We found that depleting Eiger/TNF protein specifically in Ras^V12 cells (Ras^V12, egr-RNAi clones) or preventing the secretion of Eiger/TNF (Ras^V12, sec15 double mutant cells) or reducing the protein dosage of the Eiger/TNF receptor (Wengen) all suppressed JAK-STAT activity compared to that of Ras^V12 cells (Fig. 6F; supplementary material Fig. S6E-G). Consistent with these results, wild-type cells surrounding Ras^V12, sec15 double mutant clones failed to up-regulate Upd (supplementary material Fig. S6H). Next, we tested whether directly blocking JNK activation specifically in the surrounding wild-type cells would suppress JAK-STAT activation in Ras^V12 clones. To prevent JNK activation specifically in wild-type cells surrounding Ras^V12 clones, we induced Ras^V12 clones in JNKKK dtak1 (Tak1 – FlyBase) mutant discs, but rescued JNKKK/dTAK1 function with a UAS-dTAK1 transgene only in the Ras^V12 clones (i.e. clones of cells co-expressing Ras^V12 and dTAK1 in dtak1-null mutant discs). These clones, hereafter referred to as dtak1//Ras^V12 clones, did not show STAT-GFP expression (Fig. 6B), mimicking Ras^V12, egr-RNAi clones. Finally, we removed the surrounding wild-type cells by expressing Ras^V12 throughout the developing eye discs [Eyeless-Gal4>Ras^V12, (ey>Ras^V12)] and examined JAK-STAT activation. These discs did not show broad STAT-GFP induction. Instead, we detected isolated patches of STAT-GFP-positive cells near caspase-positive cells (supplementary material Fig. S6C). Taken together, these findings indicate that it is the activation of JNK signaling specifically in the surrounding wild-type cells that non-cell autonomously stimulates JAK-STAT signaling in Ras^V12 cells.

JAK-STAT signaling promotes oncogenic Ras-mediated overgrowth

Finally, we sought to investigate whether this Eiger/TNF exocytosis-triggered JNK-JAK-STAT signaling relay contributes to Ras^V12 tumor growth. If this is the case, then it should be expected that (1) depleting Eiger/TNF in Ras^V12 cells (Ras^V12, egr-RNAi), (2) inducing Ras^V12 clones in eiger mutant animals, (3) directly blocking JNK in wild-type cells surrounding Ras^V12 cells (Ras^V12//dtak1, juxtaposed dtak1 and Ras^V12 mutant cells) or (4) preventing JAK-STAT activation in Ras^V12 cells would result in suppression of overgrowth. Indeed, Ras^V12, egr-RNAi or Ras^V12 clones induced in eiger mutant animals showed reduced growth compared with that of mosaic Ras^V12 clones (Fig. 7A-C,G). Moreover, Ras^V12, Tace-RNAi clones, which display no Eiger/TNF accumulation around the clones, showed reduced growth compared with that of Ras^V12 clones. Tace-RNAi alone showed no effect on growth (supplementary material Fig. S3C,G-J). Similarly, Ras^V12//dtak1 mutant cells showed reduced growth compared to that of controls (Fig. 7E-G).

Fig. 7.

Eiger/TNF exocytosis activation of JAK-STAT signaling promotes Ras^V12 tumor overgrowth. (A-F) Clones (green) from similarly aged intact third instar larvae (A-D), dissected cephalic regions (E,F) or dissected eye discs (lower panels). Ras^V12 clones overgrow (A). eiger/Tnf RNAi knockdown specifically in the Ras^V12 clones (B) or generating Ras^V12 clones in eiger mutant animals (C) or blocking JAK-STAT signaling by using Domeless RNAi (dom-RNAi) (D) suppresses the overgrowth of Ras^V12 clones. Dissected cephalic regions (E,F) or dissected eye discs (lower panels) show that dTAK1-rescued Ras^V12 clones induced in dtak1 hemizygous males (Ras^V12//dtak^−/−) (F), exhibit reduced growth compared with dTAK1-rescued Ras^V12 clones induced in the siblings (dtak1 heterozygous females; Ras^V12//dtak^+/−) (E). (G) Quantification of A-F. Confocal stacks from similarly aged animals of the indicated genotypes and the image analysis software Imaris were used to measure clone (green) sizes. Error bars show s.d. The numbers in the bars indicate the sample size for the corresponding genotype.

Furthermore, removing the surrounding wild-type cells by expressing Ras^V12 throughout the eye discs using eyFLP, Act>y⁺>Gal4, UAS-Ras^V12 or ey>Ras^V12 results only in moderate growth compared to the more pronounced overgrowth of mosaic Ras^V12 tissues (supplementary material Fig. S6I,J and see figure S1 in Wu et al., 2010). Consistent with reduced growth, ey>Ras^V12 discs do not cause animal lethality, whereas the overgrowth of mosaic Ras^V12 clones kills the animal during pupal stages (Fig. 1K; supplementary material Fig. S6K). We then tested whether preventing JAK-STAT activation in Ras^V12 cells, by co-expressing an RNAi transgene against the JAK-STAT receptor, Domeless, (Ras^V12, dom-RNAi clones) suppresses Ras^V12 tumor growth. Ras^V12, dom-RNAi clones showed a reduced mitotic potential as determined by the percentage of phosphorylated-histone-3-positive cells (6.3±5%, N=214 for Ras^V12, dom-RNAi clones and 49.5±29.5%, N=214 for Ras^V12; mean±s.d.; Fig. 6C,D,E) and correspondingly showed reduced tumor growth (Fig. 7A,D,G). Similarly, blocking JAK-STAT activity in ey>Ras^V12 (ey>Ras^V12, dom-RNAi) suppresses growth (supplementary material Fig. S5K,L). This suggests that the patches of dying cells contribute to the growth of Ras^V12 cells. Taken together, the above findings indicate that the accumulation of Eiger/TNF and the resulting activation of JNK signaling in the surrounding wild-type cells non-autonomously promote Ras^V12 tumor growth through JAK-STAT activation (Fig. 8).

Fig. 8.

Oncogenic Ras promotes overgrowth by stimulating Eiger/TNF exocytosis and activating JNK-JAK-STAT non-autonomous proliferation signaling. Eiger normally activates JNK to trigger cell death. JAK-STAT signaling on the other end promotes cell proliferation (black arrows). Sec15 blocks Eiger-induced cell death (green inhibition arrow) by promoting (top green arrow) the exocytosis of Eiger to the surrounding wild-type cells (top red arrow). This results in JNK activation in the surrounding wild-type (WT) cells, which non-autonomously activates the JAK-STAT pathway (bottom red arrow) to promote oncogenic Ras tumor growth.

DISCUSSION

Earlier efforts to understand Ras signaling during normal development in model organisms led to the elucidation of the Ras/MAPK signaling cascade (Nishida and Gotoh, 1993; Simon et al., 1991; Sternberg and Horvitz, 1991). However, the signaling events that are elicited by oncogenic Ras to drive overgrowth are complex and not fully understood (reviewed by Mitin et al., 2005). Here, we used a Drosophila model and searched for mutations that can block Ras^V12 overgrowth in an in vivo synthetic screen. We identified sec15, a mutation in a gene encoding a component of the exocytosis machinery, as a synthetic suppressor of Ras^V12 tumor growth. Studying the underlying cause of this phenomenon led to the unexpected discovery of both cell-intrinsic and cell-extrinsic mechanisms that promote Ras^V12 overgrowth.

Oncogenic Ras stimulates Eiger/TNF exocytosis

We found that Ras^V12 tumors elevate exocytosis to promote rapid clearance of Eiger/TNF and hence avert JNK-mediated cell death. Oncogenic Ras could stimulate the exocyst through the Ral-specific guanine nucleotide exchange factors (RalGEFs). It has been shown that oncogenic Ras activates RalGEFs, which in turn activate the Ras-like small GTPases RalA and RalB (Lim et al., 2005; Urano et al., 1996; White et al., 1996; Wolthuis et al., 1998). Moreover, RalGEF-Ral signaling stimulates gene expression (Neel et al., 2011). It is therefore likely that Ral mediates the observed increased levels of exocyst proteins in Ras^V12 cells. Consistent with this, Sec15 transcription is increased in Ras^V12 tissues (supplementary material Fig. S7A). Furthermore, activated Rals interact with and stimulate the exocyst through Exo84 and Sec5 (Hamad et al., 2002; Moskalenko et al., 2002), suggesting that RalGEF-Ral signaling could be the underlying means by which Ras^V12 cells generally stimulate exocytosis. We cannot rule out the possibility that the exocytosis of other molecules is elevated.

Eiger/TNF exocytosis is mediated by the Sec15-Eiger/TNF interaction. Consistent with this, sec15 mutant cells retain Eiger/TNF. In contrast to Ras^V12, sec15 double mutant cells, sec15 mutant cells do not undergo apoptosis. This is likely owing to the fact that oncogenic Ras stimulates Eiger/TNF expression (supplementary material Fig. S7B) in addition to promoting Eiger/TNF exocytosis. The Sec15-Eiger/TNF interaction is interesting as it might represent a novel role of the exocyst in cargo selection. It is tempting to speculate that the cytosolic domain of Eiger/TNF could mediate its interaction with the exocyst. In the future it will be interesting to investigate the nature of this interaction, map out the relevant domains and elucidate a sorting mechanism. The finding that oncogenic Ras hijacks the exocytosis machinery to clear up pro-apoptotic ligands provides a new way by which tumor cells can evade cell death and an additional explanation of how exocytosis can modulate signal transduction.

Oncogenic Ras usurps the JNK-JAK-STAT non-autonomous signal to promote overgrowth

We have previously shown that JNK activation triggered by cell polarity defects can stimulate non-autonomous JAK-STAT signaling for proliferation and that JAK-STAT signaling can cooperate with Ras^V12 (Wu et al., 2010). Here, we discovered that one of the effects of oncogenic Ras itself is to hijack this non-autonomous JNK-JAK-STAT proliferation signaling cascade. We propose that Eiger/TNF, including Eiger/TNF from the nearby fast-secreting Ras^V12 cells, is endocytosed but rapidly cleared out of Ras^V12 cells by the exocyst. Elevated exocytosis in Ras^V12 cells creates an imbalance such that the amount of secreted Eiger/TNF is higher than that being internalized. This not only causes a Ras^V12 cell to retain less Eiger/TNF, but the endocytosis and exocytosis cycle also permits neighboring Ras^V12 cells to relay Eiger/TNF, resulting in the eventual accumulation of Eiger/TNF in the surrounding wild-type cells. Consistent with this, depleting Eiger/TNF or interfering with the exocyst specifically in Ras^V12 cells abolished Eiger/TNF accumulation around Ras^V12 clones. Eiger/TNF from Ras^V12 cells activates JNK signaling in the surrounding wild-type cells. JNK activation stimulates JAK-STAT signaling in Ras^V12 cells to promote tumor overgrowth. Congruent with this model, JNK activity and the JAK-STAT ligand Upd are upregulated in wild-type cells surrounding Ras^V12 clones. Preventing Eiger/TNF secretion (Ras^V12, sec15 double mutant clones or Ras^V12, egr-RNAi clones) suppresses JNK activation, Upd stimulation and JAK-STAT activation in the surrounding wild-type cells. Moreover, blocking JNK signaling specifically in wild-type cells surrounding Ras^V12 clones prevents JAK-STAT activation in Ras^V12 cells and suppresses tumor growth.

Interestingly, oncogenic Ras or wound-induced JNK activation does not cause invasion, in contrast to JNK induced by the cell polarity mutation scrib. This could be due to context-dependent JNK activation. However, we found that Sec15-RNAi suppresses the growth and invasion phenotypes of Ras^V12, scrib double mutant clones (supplementary material Fig. S8A-D) and rescues animal lethality (supplementary material Fig. S8C-E), highlighting the importance of exocytosis in promoting tumor growth and invasion.

Another striking result of our study is the realization of the importance of the interaction between Ras^V12 and the surrounding non-Ras^V12 cells in oncogenic Ras-mediated growth. This is best illustrated by the fact that a tissue in which oncogenic Ras is uniformly expressed grows less than a tissue containing both wild-type and oncogenic Ras-expressing cells. This is consistent with accumulating data from rodents and patients indicating that signaling events emanating from host/stromal cells promote tumor development (Bremnes et al., 2011; Lu et al., 2013; Mueller and Fusenig, 2004; Sounni and Noel, 2012). Additional significance is derived from recent studies of various solid tumors showing heterogeneity within each tumor (Diaz et al., 2012; Gerlinger et al., 2012; Gerstung et al., 2011; Marusyk and Polyak, 2010).

Our study reveals that an important effect of oncogenic Ras is the increased exocytosis of Eiger/TNF. Indeed, the fact that blocking exocytosis of Eiger/TNF can suppress oncogenic Ras-mediated growth highlights the importance of Eiger/TNF exocytosis in cancer biology. Exocytosis proteins are conserved and have been implicated in diverse human cancers types, including Ras cancers (Cheng et al. 2004, 2005; Issaq et al., 2010; Neel et al., 2011; Palmer et al., 2002), suggesting that a similar tumor-promoting mechanism could be at play in vertebrates. Because sec15 null mutation does not cause cell lethality and the gene is evolutionarily conserved, it suggests a new type of potential therapeutic targets.

MATERIALS AND METHODS

Fly lines

Animals were aged at 25°C on standard medium. The following fly lines were used in this study: (1) y w; FRT82B; (2) y w; FRT82B, UAS-Ras^V12/TM6B; (3) y w; FRT40A, UAS-Ras^V12; (4) y w; FRT82B, sec15¹ (H. Bellen, Baylor College of Medicine, Houston, TX, USA); (5) y w; UAS-RasV¹²; FRT82B, sec15¹/TM6B; (6) UAS-Sec15-RNAi (VDRC); (7) FRT40,UAS-Ras^V12;UAS-eiger-RNAi/TM6B; (8) w;UAS-Ras^V12; FRT82B,Stat92E⁰⁶³⁴⁶/TM6B; (9) w;UAS-Ras^V12; FRT82B,UAS-dom (IR)/TM6B; (10) y w,ey-Flp1;act>y+>GAL4,UAS-GFP.S65T;FRT82B,tub-GAL80; (11) y w,ey-Flp1;act>y+>GAL4,UAS-myrRFP;FRT82B,tub-GAL80; (12) y w,ey-Flp1;tub-GAL80,FRT40A;act>y+>GAL4,UAS-GFP.S65T; (13) w;10XSTAT-GFP.1; (14) UAS-UAS-BskDN;eyFLP5,Act>y+>Gal4, UAS-GFP/Cyo;FRT82B, Tub-Gal80/TM6B; (15) Eyeless-GAL4 (Bloomington); (16) w,dTak1; ey-Flp5,act>y+>GAL4,UAS-GFP;FRT82B,tub-GAL80; (17) w,dTak1; ey-Flp5, act>y+>GAL4,UAS-myrRFP;FRT82B,tub-GAL80; (18) UAS-dTAK1; FRT82B, UAS-Ras^V12; (19) upd-lacZ; UAS-Ras^V12; sec15¹, FRT82B/Sb; (20) UAS-Ras^V12, UAS-Sec15-RNAi; 10XSTAT-GFP.1/TM6B; (21) ey-FLP5, act>y+>GAL4,UAS-GFP; (22) Sec6-RNAi (VDRC); (23) Sec8-RNAi (VDRC); (24) Wgn²² (P. Barker, McGill University, Montreal, Canada); (25) UAS-TACE-RNAi; (26) UAS-TACE; (27) UAS-Rab5:GFP (Bloomington); (28) GMR-Gal4; (29) UAS-Eiger^w; and (30) UAS-Bsk-DN.

Staining and imaging

Eye-antenna discs were dissected, fixed and stained as described previously (Igaki et al., 2006; Pagliarini and Xu, 2003; Wu et al., 2010). Tumor and adult eye size analyses were carried out on a Leica MZ FLIII fluorescence stereomicroscope equipped with a camera. Samples were examined by confocal microscopy with a Zeiss LSM510 Meta system. Images were analyzed and processed with Imaris (Bitplane) and Illustrator (Adobe) software, respectively. The following primary antibodies were used: guinea pig anti-Sec6 at 1:1000 [U. Tepass (University of Toronto, Canada) and H. Bellen]; guinea pig anti-Sec8 at 1:1000 (U. Tepass and H. Bellen); guinea pig anti-Sec15 at 1:1000 (H. Bellen); rat anti-Rab11 at 1:1000 (S. Cohen, University of Kansas, USA); rabbit anti-Rab11 at 1:1000 (D. Ready, Purdue University, IN, USA); rabbit anti-Stat92E at 1:1000 (S. Hou); rabbit anti-phospho-histone-3 at 1:1000; rabbit anti-cleaved caspase-3 at 1:500; rabbit anti-Eiger polyclonal antibody R1 at 1:50 (T. Igaki, Kobe University, Japan); mouse anti-phospho-JNK monoclonal antibody G9 at 1:250. Secondary antibodies were from Invitrogen. TUNEL staining was performed using the Apoptag Red kit from Chemicon.

Isosurface rendering

Projections of serial confocal sections were used to generate three-dimensional graphical images using the image analysis software Imaris. The clipping plane function was used to section across the resulting graphical images.

Vesicle trafficking studies

The whole central nervous system, including the attached eye discs, was dissected in Schneider medium and incubated in Schneider medium supplemented with 0.5 mg/ml Dextran-Alexa-546 for 2 h. Half of the samples were immediately washed and fixed. Eye discs were separated and mounted in DAPI-Vectashield. The remaining half of the samples were washed and incubated in Schneider medium for an additional 2 h (chase), rinsed in PBS, fixed and mounted as above.

Western blots and immunoprecipitation

Third instar imaginal discs bearing wild-type or Ras^V12-expressing cells were homogenized in lysis buffer (50 mM HEPES pH 7.5, 150 mM KCl, 5 mM MgCl₂, supplemented with protease inhibitor tablets; Roche). Samples were separated by SDS-PAGE, transferred onto nitrocellulose membrane and blotted with anti-Sec15, anti-Sec6, anti-Sec8. Lysates were diluted 1:100 and blotted with anti-α-Tubulin antibodies as a loading control. For immunoprecipitation experiments, 500 µl of lysate obtained from discs containing Ras^V12 clones was pre-cleared with Protein A-agarose beads for 1 h at 4°C, split equally into two tubes and incubated either with 2 μl of anti-Eiger or 2 μl of anti-GFP antibodies for 4 h at 4°C. Lysates were then incubated with Protein A-agarose beads for 1 h at room temperature. For pull downs, beads were precipitated and washed three times in modified lysis buffer containing 0.5% Triton X-100. Samples were separated by SDS-PAGE, transferred onto nitrocellulose membrane and blotted with antibodies against Sec15, Sec8, Sec5 and α-Tubulin.

Real-time polymerase chain reaction

Total RNA from eye-antenna imaginal discs containing wild-type or mutant clones was isolated using a Trizol RNA extraction method. The SuperScript III First-Strand Synthesis System kit was used to synthesize cDNA. Real-time PCR was carried out on an Applied Biosystems machine using the SYBR Green fast kit following the manufacturer's instructions. Relative gene expression was obtained from triplicate runs normalized to Rp49 (RpL32 – FlyBase) as the endogenous control. The following primers were used: Rp49, 5′-GGCCCAAGATCGTGAAGAAG-3′ and 5′-ATTTG-TGCGACAGCTTAGCATATC-3′; Socs36E, 5′-GTCGCTGCCAGTCAG-CAA-3′ and 5′-TGCTCCCATTGAAAGTGCTTT-3′; Sec15, 5′-CATTA-TGCTGTCCATCAACACCTT-3′ and 5′-CAGCGTTGCAGTAAAACCT-CATT-3′; eiger, 5′-GATGGTCTGGATTCCATTGC-3′ and 5′-TAGTCT-GCGCCAACATCATC-3′.

Clones of mutant cells in the eye-antennal discs were generated using standard MARCM system (Lee and Luo, 2001) at 25°C. Hanging-eye phenotype experiments were carried out at 29°C. Immunostaining and immunoprecipitation experiments were performed as described previously (Chabu and Doe, 2008; Pagliarini and Xu, 2003).