Elevated expression of the V-ATPase C subunit triggers JNK-dependent cell invasion and overgrowth in a Drosophila epithelium

Extra- and intracellular pH is tightly regulated by a number of proton transport systems. A prominent member is the vacuolar H⁺-ATPase (V-ATPase). The proton pump is required for several cellular processes depending on the cell type and intracellular localization (Hinton et al., 2009). Through the regulation of endolysosomal acidification, the V-ATPase participates in protein degradation as well as in the control of endosomal trafficking and sorting (Hurtado-Lorenzo et al., 2006). This function was recently also suggested to be required in a number of signaling pathways important in development and cancer (Niehrs and Boutros, 2010; Yan et al., 2009; Buechling et al., 2010; Cruciat et al., 2010; Hermle et al., 2010; Vaccari et al., 2010; Hermle et al., 2013). In tumor cells, enhanced plasma membrane activity of the V-ATPase and other proton transporter strongly correlates with increased cell proliferation and invasive cell migration (Webb et al., 2011).

V-ATPases are large protein complexes that are organized into two domains: the membrane-embedded V0 domain responsible for proton translocation and the cytoplasmic V1 domain carrying out ATP hydrolysis. As a multi-subunit complex, the activity of the V-ATPase depends on the reversible assembly of the complex as well as the dynamic expression of V-ATPase subunits at different cellular membranes. Whereas most of the 15 subunits are essential structural and functional components, others function as regulatory subunits controlling assembly and subcellular localization (Forgac, 2007). One important regulatory subunit is the C subunit, termed ATP6V1C1 in mammals and Vha44 in Drosophila. It is the only subunit that is released into the cytosol during disassembly of the proton pump (Kane, 2000; Merzendorfer et al., 2000). Its phosphorylation by protein kinase A has been suggested to promote (re-)assembly of the V0-V1 holoenzymes (Voss et al., 2007). Furthermore, biochemical analysis has revealed that its presence stabilizes complex assembly and increases pump activity (Puopolo et al., 1992).

Among the V-ATPase subunits, ATP6V1C1 is the most upregulated gene in oral squamous cell carcinoma, a highly metastatic human cancer (Otero-Rey et al., 2008). Expression is predominant at the periphery and invasive fronts of tumors (Garcia-García et al., 2012), suggesting an important role of the C subunit in invasive growth and migration (Pérez-Sayáns et al., 2009). Other V-ATPase subunits, including the a3 and c subunits, have also been implicated in cancer metastasis (Lu et al., 2005; Nishisho et al., 2011). Moreover, it has been shown that plasma membrane activity of the V-ATPase is strongly increased in highly metastatic breast cancer cells (Sennoune et al., 2004). The general conclusion of these and other studies has been that the V-ATPase contributes to the reduced extracellular pH of tumors (Robey et al., 2009; Webb et al., 2011). V-ATPase inhibitors have therefore been tested in human cancer trials with the rationale to lower tumor acidity and to improve patient outcome (Otero-Rey et al., 2008; Hinton et al., 2009; Fais, 2010; Webb et al., 2011). Most experimental data concerning the role of V-ATPase subunits in tumorigenesis has been obtained by using xenografts or cultured tumor cell lines separated from their tissue microenvironment (Sennoune et al., 2004; Lu et al., 2005; Nishisho et al., 2011). The question whether changes in cellular pH are cause or consequence of cancerous transformation therefore largely remains unanswered. Furthermore, it is not known whether altered V-ATPase function might impact cell behavior via the regulation of signaling pathways.

Here, we report on the oncogenic properties of the C subunit of the V-ATPase, Vha44, in the fruit fly Drosophila melanogaster. Drosophila imaginal discs are excellent model tissues for studying the function of individual oncogenes and tumor suppressors but also the cooperativity of genes in native cellular environments (Vidal et al., 2006; Halder and Mills, 2011). By contrast to xenografts or cultured cancer cell lines, the mutational load can be precisely controlled, allowing the genetic requirements of cell transformation to be analyzed within the tissue of origin. Moreover, because tumor cells are genetically produced, mechanical disruption of cells and of the extracellular matrix is avoided. Using the wing disc as a model epithelium, we show that the overexpression of Vha44 is sufficient to trigger invasive cell behavior in the epithelial wing disc tissue. Our data suggest that ectopic Vha44 expression increases vesicular acidification, impairs endolysosomal degradation and induces a TNF/Eiger- and JNK-dependent transformation program.

Overexpression of Vha44 was targeted to the patched (ptc) expression domain of the larval wing disc epithelium using the GAL4/UAS expression system. This domain constitutes a narrow stripe at the anterior-posterior (AP) boundary, in which expression is highest at the boundary-facing side of the stripe (Fig. 1A–E). Expression of both HA-tagged and untagged Vha44 caused a striking unidirectional migration of the cells from the anterior ptc stripe into the posterior compartment (Fig. 1B,D; supplementary material Fig. S1A,B). Additional expression of monomeric red fluorescent protein (mRFP) confirmed the identity of the invading cells as originating from the ptc stripe (Fig. 1B,B′). In control discs, mRFP expression was always confined within the ptc stripe (Fig. 1A,C), consistent with the role of the AP boundary in preventing mixing of cells (Landsberg et al., 2009). x-z axis projection revealed that migration took place on the basal side of the wing epithelium (Fig. 1D). These results suggest that cells close to the boundary are extruded basally and then move towards the posterior compartment. Basal invasion into the neighboring compartment was also seen by Vha44 overexpression in other compartments of the wing disc such as the dorsal part using apterous (ap)-GAL4 (supplementary material Fig. S1A,D). Therefore, we conclude that Vha44 overexpression can trigger invasion in different regions of the disc epihelium.

Basal extrusion of epithelial cells in Drosophila tissue is often accompanied by apoptosis (Vidal et al., 2006; Igaki et al., 2009; Marinari et al., 2012). Accordingly, we observed that most of the extruded cells displayed pyknotic nuclei suggestive of apoptotic cell death (not shown). Further investigation revealed that, unlike mRFP-expressing control cells, a significant fraction of the invasive cells were positive for cleaved Caspase 3 (Cas3) as well as TUNEL labeling (Fig. 1F,G; supplementary material Fig. S2A,B). Apoptosis was highest at the AP boundary and in the invasive front. This suggests that cells were displaced into the basal extracellular matrix (ECM) at the boundaries due to apoptosis and then migrated into the posterior compartment. Consistent with previous results, we further verified that, unlike Vha44, the induction of apoptosis by overexpression of a pro-apoptotic protein is not sufficient to cause cell migration (supplementary material Fig. S2C,D; Vidal et al., 2006). This indicates that additional mechanisms are involved in the invasive behavior.

Next, we coexpressed the baculovirus protein p35, which acts as a strong suppressor of apoptosis by inhibiting Cas3 function (Hay et al., 1994). As a result, we observed no apoptosis and basal exclusion, but nevertheless the ptc stripe appeared broadened and irregular (Fig. 2A,B; supplementary material Fig. S2E,F). Staining for phospho-histone 3 (PH3) revealed increased proliferation in the ptc stripe in Vha44/p35- but not in mRFP/p35-expressing control cells (Fig. 2C,D). Expressing Vha44 in the ap or nubbin (nub) compartments, encompassing most of the wing disc or wing pouch, respectively, led to overgrowth even in the absence of p35, as shown by the increased size of the compartments and the protruding bulges at the wing disc edges (Fig. 2F,G; supplementary material Fig. S1C). We therefore conclude that Vha44 overexpression promotes proliferation, an effect that is masked by strong apoptosis at the AP boundary.

The phenotypes observed upon Vha44 overexpression are reminiscent of the cellular transformation occurring during tumorigenesis. Thus, we next investigated whether ectopic Vha44 expression activates oncogenic signaling pathways previously reported to mediate cell invasion and overgrowth in the wing disc (Vidal et al., 2006; Singh et al., 2010). Silencing the Src inhibitor Csk with a previously validated RNAi line did not significantly increase the level of Vha44-induced migration (Vidal et al., 2006; supplementary material Fig. S3A–D). Neither did we find a significant activation of Abl in the Vha44-expressing cells (supplementary material Fig. S3E). In contrast, the co-overexpression of Vha44 and RasV12, the oncogenic form of Ras, led to a strong increase in cell migration compared with overexpressing Vha44 and the control protein mRFP (Fig. 3B–D). Coexpressing RasV12 and Vha44 also caused a synergistic effect with regard to overgrowth (Fig. 3F–H). Because RasV12 expression alone did not lead to any invasion or overgrowth (Fig. 3A,E), these results indicate that Vha44 cooperates with oncogenic Ras in invasion and overgrowth.

To understand the cellular mechanism by which Vha44 overexpression induces this complex tumor-like phenotype, we first examined the subcellular localization of overexpressed Vha44 in wing discs. Consistent with the reported localization of V-ATPase (Feng et al., 2009), endogenous and overexpressed Vha44 was mainly found in intracellular organelles but also at the plasma membrane of epithelial wing cells (supplementary material Fig. S4B–D). A substantial fraction of these organelles were positive for the late endosomal marker Rab7 (supplementary material Fig. S4D). To investigate the effects of Vha44 overexpression on V-ATPase function, we applied the acidotrophic fluorescent dye LysoTracker to isolated live wing discs overexpressing Vha44 under ptc-GAL4. We observed a strong accumulation of LysoTracker in the ptc stripe but not in the neighboring wild-type cells or in control cells (Fig. 4A,B). This indicates that organellar acidification is increased in Vha44-expressing cells. Blocking V-ATPase activity with the specific inhibitor Concanamycin A resulted in the complete loss of the LysoTracker labeling, indicating that the Vha44-mediated increase in organellar acidification is due to enhanced V-ATPase function (Fig. 4C). In addition, we found that overexpression of another V-ATPase subunit, Vha100-1, did not alter Lysotracker incorporation or cause invasion and overgrowth phenotypes (supplementary material Fig. S5A–C). In line with the proposed V-ATPase assembly function for Vha44, these results suggest that unlike Vha100-1, Vha44 is sufficient to cause association of V0 and V1 subcomplexes, thus increasing the amount of functional holoenzymes within membranes of intracellular organelles. Accordingly, using a GFP insertion in the genomic locus of Vha16-1, we found enhanced levels of the V0 component Vha16-1-GFP upon Vha44 overexpression (Fig. 4D). By contrast, a GFP insertion in VhaSFD was downregulated upon Vha44 overexpression (Fig. 4E). The reason for this downregulation is unclear. However, as VhaSFD (or H subunit) has been reported to inhibit ATP hydrolysis of the unassembled V1 sector (Jefferies and Forgac, 2008; Diab et al., 2009), its reduction upon Vha44 overexpression might be consistent with increased pump assembly.

To test whether increased proton transport is the cause of the observed tissue alterations, we overexpressed the sodium-proton exchanger Nhe2. This multi-transmembrane protein translocates protons across membranes in the same direction as the V-ATPase and has been shown to localize to the plasma membrane and endosomes (D’Souza et al., 1998). We found that Nhe2-YFP overexpression caused weaker but very similar phenotypes compared to Vha44: invasion, apoptosis and overgrowth and increased organellar acidification, (supplementary material Fig. S6A–D). By co-staining with Avl and Rab7, we confirmed that a pool of overexpressed Nhe2-YFP localized to early and late endosomal compartments (supplementary material Fig. S6E,F). Taken together, these results suggest that increased endolysosomal acidification might be sufficient for invasion and overgrowth phenotypes.

V-ATPase-dependent acidification has been linked to endosomal sorting and lysosomal degradation (Hurtado-Lorenzo et al., 2006). To study these membrane trafficking processes, we first expressed a GFP-Lamp1 fusion construct under the ubiquitous tubulin promoter. In Drosophila cells, the LAMP1-derived cytoplasmic tail is sufficient to target this fusion protein from the Golgi to late endosomes and lysosomes, where hydrolases degrade GFP (Rohrer et al., 1996; Pulipparacharuvil et al., 2005). Whereas GFP-Lamp1 was hardly detectable in the neighboring tissue and in mRFP-expressing control cells (Fig. 5A–C), overexpression of Vha44 (with and without p35) caused a severe accumulation of GFP-Lamp1 in the ptc stripe (Fig. 5B,C). GFP-Lamp1 partially colocalized in enlarged organelles with the late endosomal marker Rab7 and Lysotracker (Fig. 5D,E), giving rise to a ring-like structure with luminal GFP signal and Rab7 signal at the limiting membrane (inset in Fig. 5D). Generally, there seemed to be an irregular localization pattern for Rab7 in the Vha44-expressing ptc stripe with accumulations in enlarged compartments (Fig. 5F,G). Together, these data suggest that Vha44 overexpression impairs endolysosomal degradation leading to cargo accumulation in enlarged late endosomes and/or lysosomes.

Endocytic activation and/or impaired degradation of signaling receptors have been implicated in sustained tumor growth and invasion. In Drosophila, oncogenic Notch and c-jun N-terminal kinase (JNK) signaling has been demonstrated to occur in endosomes (Vaccari et al., 2008; Igaki et al., 2009; Vaccari and Bilder, 2005; Wilkin et al., 2008; Rodahl et al., 2009; Hori et al., 2011). Therefore, we investigated the role of Notch and JNK signaling in Vha44-induced transformation. Consistent with alterations of protein trafficking and degradation upon ectopic Vha44 expression, we found an accumulation of the transmembrane receptor Notch in the lumen of enlarged late endosomes (Fig. 5I,J). In addition, we found a significant colocalization of Notch-YFP with LysoTracker (Fig. 5M,N). To monitor directly whether Notch is degraded in Vha44-expressing cells, we applied an antibody against the extracellular domain of Notch (NECD) onto live wing discs. Notch was found at the cell surface at t=0 hours in Vha44-expressing cells (Fig. 5K). In contrast, after 5 hours, the receptor accumulated in intracellular compartments in Vha44-cells (Fig. 5L). These results indicate that, similarly to GFP-Lamp1, Notch accumulates in Vha44-overexpressing cells due to reduced endolysosomal degradation.

Previous studies have shown that Notch activation requires V-ATPase activity (Yan et al., 2009; Vaccari et al., 2010; Hermle et al., 2013). We therefore tested the status of Notch signaling activation in cells overexpressing Vha44. We found that in Vha44-expressing cells expression of the Notch signaling reporter E(Spl)mβeta-lacZ was well detectable in areas in which Notch signaling is normally low, such as the presumptive vein regions of the disc. However, it was not as high as in peak signaling portions of the disc, such as the presumptive wing margin (supplementary material Fig. S7A–D). Accordingly, overexpression of Vha44 resulted in both Notch gain- and loss-of-function phenotypes in adult wings (supplementary material Fig. S7G–J). Finally, quantitative real-time PCR (qPCR) demonstrated no significant change in E(Spl)mβeta expression in wing discs expressing Vha44 with nub-GAL4 (supplementary material Fig. S7K). This suggests that despite Notch accumulation in acidified organelles, there is no major alteration of Notch pathway activation in Vha44-overexpressing cells compared with wild-type cells.

In contrast to Notch signaling, the qPCR revealed increased expression of the JNK signaling inhibitor and target gene puckered (puc; supplementary material Fig. S7K). We therefore monitored JNK activity in the Vha44-expressing cells by immunostaining for the activated phosphorylated form of JNK (pJNK) and by assessing puc expression using the transcriptional reporter puc-lacZ. Both, pJNK and puc-lacZ levels were increased in Vha44-expressing cells but not in mRFP-expressing control cells (Fig. 6A,B; supplementary material Fig. S8A,B), indicating that JNK signaling is activated. Similar results were obtained for Nhe2 (supplementary material Fig. S6B). Accordingly, we found that the expression of Hep^CA, the activated form of JNKK kinase, is sufficient to drive invasion (supplementary material Fig. S8D). A further readout for upregulated JNK activity is increased expression of Matrix Metalloproteinase-1 (MMP1). MMPs are essential downstream targets of JNK that control invasive migration via basal membrane degradation (Uhlirova and Bohmann, 2006; Vidal et al., 2006). Consistent with JNK activation, we found increased MMP1 levels basally in the domain of Vha44 but not mRFP expression (Fig. 6C,D). We also observed a clear loss of the basal membrane components collagen IV and laminin (Fig. 6E; supplementary material Fig. S8F). These effects were abolished by the coexpression of p35 (Fig. 6F), indicating that the degradation of the basal membrane is linked to apoptosis and basal extrusion of cells. Taken together, these results suggest that excess Vha44 leads to JNK activation, resulting in apoptosis as well as MMP1 upregulation and ECM degradation, which might be a prerequisite for basal extrusion and invasion.

To test whether inhibition of JNK signaling is not only necessary but also sufficient to drive invasive behavior, we coexpressed Puc and Vha44. Inhibition of JNK signaling rescued apoptosis as well as the migration and proliferation phenotypes (Fig. 6G,G′). A similar suppression of apoptosis and proliferation was observed when coexpressing Basket (Bsk)-DN, a dominant-negative form of the Drosophila orthologue of JNK, with ap-GAL4 (Fig. 6H,H′). Consistent with these results, we found that the activation of JNK signaling resulting from expression of puc-lacZ, a hypomorphic allele of the inhibitor Puc, increased the invasion phenotype (Fig. 6B, compare with 6D). Finally, we tested the contribution of Eiger, the only Drosophila orthologue of the TNF-α ligand and potent activator of JNK signaling (Igaki et al., 2002; Moreno et al., 2002), by expressing Vha44 in a mutant eiger background. Loss of both eiger copies strongly suppressed apoptosis, invasion and overgrowth of Vha44-expressing cells (Fig. 6I,J). Together, these results demonstrate that Eiger/JNK signaling is both necessary and sufficient for Vha44-dependent invasion and overgrowth (for a model see Fig. 7).

Increasing evidence suggests that pH alterations significantly affect cancer features including sustained growth, tissue invasion, metastatic potential and chemoresistance (Stock and Schwab, 2009; Webb et al., 2011). Whether the effect of dysregulated pH on cancer progression is a cause or consequence for the development of these tumor traits has been difficult to address using traditional approaches such as xenografts or cultured cancer cell lines. In addition, because pH alterations ultimately affect multiple cellular processes, it is unclear what process is required for tissue transformation.

Drosophila imaginal discs are excellent model tissues for studying the impact of transforming events in the native epithelium (Vidal et al., 2006; Halder and Mills, 2011). Here, we have shown that overexpression of the V-ATPase subunit Vha44 in different compartments of the epithelial wing disc is sufficient to induce tumor-like transformations. Cells overexpressing Vha44 overproliferate, giving rise to protruding cell masses. In addition, they display increased apoptotic rates and basal invasion, which is facilitated by ECM degradation. Both overgrowth and invasion are increased by oncogenic Ras and inhibited by downregulation of Eiger/JNK signaling, demonstrating that the latter pathway is a key driver of cellular transformation in our tumor model.

How does Vha44 induce such a complex tumor-like phenotype? In line with the proposed role of the C subunit as an assembly and disassembly regulator (Puopolo et al., 1992; Huss et al., 2011), we have found that Vha44 overexpression leads to enhanced uptake of LysoTracker in wing epithelial cells. This effect is reversed by pharmacological V-ATPase inhibition and mimicked by overexpression of a different proton exchanger, suggesting that Vha44-overexpressing cells possess increased proton pump assembly and activity at intracellular membranes. Although the overexpression of another V1 subunit also increased vacuolar acidity in yeast (Hughes and Gottschling, 2012), the overexpression of a transmembrane V0 subunit, Vha100-1, did not lead to additional LysoTracker uptake. This suggests that excess cytosolic V1 components are more effective in enhancing V-ATPase assembly and, thus, activity.

We expected that increased pump activity would result in more lysosomal degradation. In contrast, we found the opposite. GFP-Lamp1 and Notch accumulated in enlarged late endosomal compartments in which Notch degradation is impaired. A similar combination of increased acidification and decreased proteolysis has been observed in cells deficient for the lysosomal transporter Spinster (Rong et al., 2011). How endosomal sorting and transport become blocked upon increased acidification is currently unclear. An interesting possibility is that changes in luminal pH might alter the binding properties of proteins and lipids at the cytosolic side of the endosomes (Hurtado-Lorenzo et al., 2006). Indeed, in mammalian cells, the acidification by the V-ATPase was shown to promote transport towards lysosomes by recruiting the GTPase ARF6 (and possibly Rab7 as suggested here) to early endosomes (Hurtado-Lorenzo et al., 2006). Therefore, excessive acidification might alter endosomal maturation and cargo progress towards functional lysosomes.

Our results are in accordance with previous findings demonstrating that the inhibition of lysosome formation in dor, car, tsg101 and vps25 mutants can cause tumor growth and invasion in the fly (Moberg et al., 2005; Thompson et al., 2005; Vaccari and Bilder, 2005; Chi et al., 2010). These mutations block the maturation of late endosomes to lysosomes at different levels, similarly causing signaling receptor accumulation. Furthermore, the progression of vps25-dependent tumors was shown to be suppressed by mutants for V-ATPase subunits, suggesting that oncogenic signaling requires low intraluminal pH (Vaccari et al., 2010). It is therefore possible that a decreased luminal pH might promote clustering or conformational changes of receptors that alter recruitment of cytosolic signal transducers. The activation of Eiger/JNK signaling, the key pathway in our tumor model, has also been linked to endosomes (Igaki et al., 2009). However, at this point it is unclear to what extent acidity contributes to the activation of Eiger, its receptor Wengen and/or the phosphorylation of JNK (Igaki et al., 2011). For the canonical Wnt [or Wingless (Wg)] pathway, it has been shown that the receptor complex, consisting of Frizzled and LRP6, can directly bind to a V-ATPase subunit and that phosphorylation of LRP6, a prime event in the signaling cascade, occurs upon entry into acidic endosomes (Buechling et al., 2010; Cruciat et al., 2010; Hermle et al., 2010). In the case of the Notch receptor, acidification might promote the intramembraneous S3 cleavage by presenilins (Vaccari et al., 2010). Alternatively, the reduced proteolysis rate in a disturbed endolysosomal pathway might increase the time of transit of receptors in signaling-competent endosomes, leading to sustained signaling. Further investigation is therefore needed to elucidate the precise mechanisms by which signaling receptors and their downstream transducers, particularly within the JNK signaling pathway, become activated in acidified compartments.

The induction of cellular transformation in a patch of cells within an epithelium, as described here, elicits a myriad of cell-autonomous and non-cell-autonomous responses. Epithelial neighbors and recruited inflammatory cells release both proliferative and apoptotic stimuli that strongly influence tumor growth (Cordero et al., 2010). Transformed cells that become apoptotic might further stimulate proliferation of the surviving ones by releasing growth-promoting morphogens such as Wg and Dpp (Huh et al., 2004; Pérez-Garijo et al., 2004; Ryoo et al., 2004; Morata et al., 2011). Accordingly, we observed that Wg strongly accumulates in parts of the Vha44-expressing tissue and that Wg signaling is moderately upregulated upon Vha44 overexpression (supplementary material Fig. S7F,K). Increased proliferation at tissue boundaries is particularly interesting because mechanical strains at the boundary can lead to overcrowding, which has recently been shown to cause delamination and apoptosis (Marinari et al., 2012). These effects possibly play a role when Vha44 is expressed in the ptc domain. Here, cells become strongly apoptotic and are basally excluded close to the boundary. Upon inhibition of apoptosis, the proliferation effect induced by Vha44 becomes visible and might be further supported by the release of growth factors from the transformed cells themselves or from the tissue environment. Beyond developmental boundaries, such as the AP boundary, it is conceivable that in situ carcinomas can generate similar border effects (Vidal et al., 2006; Vidal et al., 2010).

A major advantage of Drosophila for studying cancer-related processes is the introduction of multiple genetic alterations. Genetic interaction experiments are very useful for assessing the cooperativity of oncogenic pathways. Such studies have highlighted the importance of Ras signaling in eliciting cancerous transformation upon defects in epithelial polarity and membrane trafficking (Pagliarini and Xu, 2003; Chi et al., 2010). Although poorly understood, the JNK signaling pathway seems to play a central role in this process (Igaki et al., 2006; Uhlirova and Bohmann, 2006; Leong et al., 2009; Wu et al., 2010). One example is the RasV12-induced shift of the JNK activating ligand, Eiger/TNF-alpha, from a tumor suppressor into a tumor promoter (Cordero et al., 2010). Whether the increase of invasion and overgrowth by RasV12 in our tumor model involves JNK signaling and whether the activation of JNK by Vha44 requires Ras remains to be further explored. It also needs to be determined how Eiger/JNK inhibition rescues the pathological aspects of our model. The inhibition of apoptosis might be the prime reason for the rescue, but more direct influences on the endolysosomal or associated pathways, such as autophagy, cannot be excluded at this stage (Wu et al., 2009).

Here, we introduce a novel tumor model in Drosophila based on V-ATPase subunit C overexpression. We describe the induction of cellular transformation in the native epithelium, with high similarity to human cancers. Genetic dissection of our model revealed increased endosomal acidification, disrupted endolysosomal trafficking and degradation as well as a requirement for JNK signaling. All these effects have previously not been associated with V-ATPase overexpression in tumors. A recent study by Vidal and colleagues showed that sections of human squamous cell carcinoma boundaries exhibited similarities with the wing disc invasion model, including the upregulation of MMPs (Vidal et al., 2010). Due to the elevated levels of ATP6V1C1 in oral squamous cell carcinomas (Otero-Rey et al., 2008), our study suggests that clinical cancer trials involving V-ATPase inhibitors could be expanded to include this type of tumor (Fais, 2010). As demonstrated by the genetic interaction with Ras signaling, our model could also be used for studying the cooperativity of oncogenic pathways and for the evaluation of combinatorial therapies.

For Vha44 expression in the wing disc, we used ptc-Gal4, dpp-Gal4, nub-GAL4 (all Bloomington) or ap-Gal4 (a gift of Marek Mlodzik) drivers. Controls for UAS-Vha44-HA or UAS-Vha44 were UAS-myr-mRFP or UAS-GFP (both Bloomington). Apoptosis inhibition was performed by coexpression of UAS-p35 (Bloomington). As JNK reporter we used puc-lacZ (E69), which also acts as a mild JNK activator. The basal membrane was visualized with viking-GFP (Flytrap). Other flystocks were: egr³ (a gift of Marcos Vidal), UAS-Nhe2-YFP (Simons et al., 2009), UAS-puc2A, UAS-Bsk-DN (both a gift of Stephane Noselli), E(Spl)β -lacZ (a gift of Eric Lai) and UAS-Vha100-1 (a gift of Robin Hiesinger). Flies were raised and crossed at 25°C, except for p35 or ap-GAL4 crosses at 29°C. For the generation of the UAS-Vha44 and the UAS-Vha44-HA fly strains we cloned Vha44 with the following primers: forward, 5′-CACCATGATGTCGGAATACT-3′ and reverse, 5′ -TTAGAC-CTTGGCCTGCTCCA-3′ (with STOP codon for Vha44) or reverse, 5′-GACCTTGGCCTGCTCCACCA-3′ (without STOP codon for Vha44-HA). The amplicon was first inserted into the pENTR^™/SD/D-TOPO^® Gateway vector (Invitrogen) and then cloned into the pUASg-attB or pUASg-HA-attB destination vectors (kindly provided by Konrad Basler). Transgenesis was performed by Bestgene, Inc.

Wing discs were dissected from L3 larvae in PBS and immunostained according to standard procedures (Hermle et al., 2010). Antibodies and dyes used in this study were: mouse anti-Dlg (1:25), rat anti-DE-cadherin (1:25), mouse anti-Notch (NICD) (1:50), mouse anti-Notch (NECD) (1:50), mouse anti-Wg (1:50), mouse anti-MMP1 (1:25), mouse anti-b-gal (1:25; all from DSHB), rat anti-HA (1:200; from Roche), rabbit anti-cleaved Caspase 3 (1:200; from Cell Signaling), rabbit anti-Laminin (1:25; from Abcam), rabbit anti-Rab7 (1:1000; kindly provided by Akira Nakamura), rabbit anti-Avl (1:250; kindly provided by David Bilder), guinea pig anti-v100 (1:2000; kindly provided by Robin Hiesinger), mouse anti-pJNK (G7) (1:50; from Santa Cruz Biotechnology); rabbit anti-pAbl[pY412] (1:100; from Invitrogen). A polyclonal antibody against Vha44 was raised in guinea pig against the following peptide: QIGQIDGDLKTKSQA (Eurogentec). For F-actin and nuclei visualization, AlexaFluor 488- and 555-Phalloidin (1:1000; from Invitrogen) and HOE33342 (1:1000; from Invitrogen) were used, respectively. For TUNEL staining, wing discs were dissected in PBS, fixed in 4% paraformaldehyde and permeabilized. The Fluorescein In Situ Cell Death Detection Kit (Roche) was used according to the instruction manual. To assess intravesicular acidification, dissected L3 larval discs were incubated for 2 minutes at room temperature with LysoTracker green DND26 or red DND99 (1μM; Invitrogen) in PBS, then mounted and immediately analyzed. V-ATPase activity was blocked with Concanamycin A (1μM; Sigma) treatment for 3 hours at room temperature in PBS. Notch internalization assays were performed as described (Vaccari and Bilder, 2005). Discs were analyzed by Zeiss LSM 510 laser confocal microscopy. Images were processed with Adobe Photoshop and ImageJ software. For quantification, the area of the Cas3- or mRFP-positive cells was determined and normalized to the area of the ptc stripe. The mean pixel intensities of the immunoreactive Rab7 signal were measured in square areas of equal size in Vha44-expressing and neighboring wild-type tissue.

A total of 15 wing discs for each genotype were dissected and immediately lysed in cold RIPA buffer. The whole disc extract was loaded on 12% SDS-PAGE gel and western blot was performed following standard procedures. The Vha44 antibody was used at 1:500. Other antibodies were anti-β-tubulin (1:1000; from DSHB) and horseradish peroxidase (HRP)-conjugated anti-guinea pig and anti-mouse (1:5000; from Santa Cruz Biotechnology and DAKO, respectively).

Total RNA from wing imaginal discs expressing (40 discs per sample) was extracted using TRIZOL Reagent (Invitrogen) and RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. Concentration and purity was determined by measuring optical density at 260 and 280 nm using a Nanodrop spectrophotometer. 500 μg of total RNA was reverse transcribed using a SuperScript VILO cDNA Synthesis kit (Invitrogen) according to the manufacturer’s protocol. 5 ng of cDNA was amplified (in triplicate) in a reaction volume of 15 μl containing the following reagents: 7.5 μl of TaqMan PCR Mastermix 2× No UNG (Applied Biosystems, Foster City, CA), 0.75 μl of TaqMan Gene expression assay 20× (Applied Biosystems, Foster City, CA). 300 nM of primers and 100 nM of Roche probes were used for each sample. RT-PCR was carried out on the ABI/Prism 7900 HT Sequence Detector System (Applied Biosystems), using a pre-PCR step of 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 60 seconds at 60°C.

The following primers were used: forward, 5′-GAGTG-CCTGACCCAGGAG-3′ and reverse, 5′-CGGTCAGCTCCA-GGATGT-3′ [for E(spl)mβ]; forward, 5′ -GTCACACCAATC-AGTGGAG-3′ and reverse, 5′-CGAGCAGCCGGATTCTATTA-3′ (for fz3); forward, 5′-GCCACATCAGAACATCAAGC-3′ and reverse, 5′-CCGTTTTCCGTGCATCTT-3′ (for puc); forward, 5′-CGGATCGATATGCTAAGCTGT-3′ and reverse, 5′-CGAC-GCACTCTGTTGTCG-3′ (for rpL32-RA).

Data were analyzed using GraphPad Prism 5.0d (GraphPad Software). Values were normalized by the amount of rpl32 in each sample. Statistical analysis was performed using the Student’s t-test.

We thank Susanne Helmstädter for excellent technical support. We thank G. Pyrowolakis, M. Mlodzik, M. Vidal, T. Xu, S. Noselli, R. Rousset, Flytrap, the Bloomington Stock Center and the VDRC for fly strains. We thank the DHSB for antibodies, as well as A. Nakamura and D. Bilder for antibodies. We acknowledge Roland Nitschke from the Life Imaging Center Freiburg for help with confocal microscopy. We thank M. Vidal and M. Norman for helpful suggestions and the critical reading of the manuscript.