lgl mutant clones die by caspase-dependent apoptosis when induced in an lgl^+/-background

In order to investigate the proliferation and polarity phenotypes of lgl mutant cells during development, we induced clones of two different lgl null mutations in Drosophila wild-type wing discs, larval proliferating epithelia where competitive mechanisms between genetically different cell populations have been extensively characterised [14, 15]. As can be observed in Figure 1A and Figure S1A, B in Additional File 1, lgl⁴ [21] or lgl^27S3 [22] mutant clones grew preferentially in the proximal-distal direction, in a pattern similar to that of wild-type clones [23]. At two days after induction lgl^-/- clones and their twin lgl^+/+ clones were similar in size (Figure S1A in Additional file 1), while at three days lgl^-/- clones were about one half the size of their twins (bright white, compare Figure S1B and Figure S1A in Additional file 1, n = 40 each) and many of them (37%, n = 287) disappeared by the end of larval development, leaving only the twin clone (Figure S1B in Additional file 1, arrow). However, some persisted to the end of development, resulting in scarring of the adult wing (Figure S1C in Additional file 1).

Staining for active-Caspase 3 revealed that lgl^-/- cells die by apoptosis (Figure 1A) and in basal sections many pycnotic nuclei were visible within lgl^-/- clones (Figure S1E in Additional file 1), confirming the presence of dying cells. Caspase activation occurred mostly in mutant cells at clonal boundaries, where mutant and normal tissues were in close contact (outlined, magnified in Figure 1A*). The anti-apoptotic protein Drosophila Inhibitor of Apoptosis 1 (dIAP1), ubiquitously expressed in imaginal tissues [24], was downregulated in lgl^-/- cells (Figure 1B, arrow), whereas the JNK pathway was activated (as detected by phospho-JNK staining) in the mutant clones (Figure 1C, arrow). Possible crosstalk among these pathways in inducing cell death in lgl mutant clones will be discussed in detail in Section 5.

lgl mutant clones showed slight defects in disc folding, as revealed by F-actin staining (Figure S1D in Additional file 1, arrow); however, mutant cells did not seem to be strongly compromised in apical-basal polarity (as detected by aPKC and Scribble staining), at least up to 72 hrs after clone induction (Figures 2C and S1F, G in Additional file 1), and no discontinuities in basement membrane were visible (not shown), in agreement with previous observations [22, 25].

lgl mutant cells express low levels of dMyc oncoprotein compared with the adjacent epithelium, and the latter grows at the expense of the lgl^-/-clones

To test whether competitive interactions were involved in the elimination of lgl^-/- cells from the normal tissue, we performed a double clonal assay in which lgl^-/- clones were induced in an lgl^+/- background in parallel to wild-type clones induced in a wild-type background (Figure 1D). As expected, while no differences were observed between wild-type control clones and their twins (right panel, black and white bars respectively, P = 0.86), lgl mutant clones were smaller than their wild-type twins (left panel, black and white bars respectively, P < 0.001), as well as smaller than wild-type clones induced in control discs (right panel, black bars, P < 0.001). The wild-type twins of lgl^-/- clones (left panel, white bars) were instead much larger than the wild-type twins from control discs (right panel, white bars, P < 0.001). Since no dominant effects on cellular growth rate have been reported for lgl null mutations, and developmental stages of lgl^+/- animals are of the same duration as those of lgl^+/+ individuals, even in clonal assays, this result suggests that some non-autonomous mechanisms are at work in lgl^-/- clone elimination and that mutant cells are being replaced by the normal surrounding tissue.

Since it is known that differences in dMyc abundance among adjacent cells can trigger competitive behaviour [14, 15], we next analysed dMyc protein levels and found that dMyc was weakly expressed within lgl mutant clones with respect to the surrounding lgl^+/- tissue (Figure 2). We therefore speculated that cells bearing precancerous lesions that do not confer a survival or proliferative advantage relative to neighbouring cells, as with lgl^-/- cells that lack dMyc protein, can be eliminated by apoptosis. In the wing disc, dMyc protein accumulates mainly in the distal region, the wing pouch, and is expressed only weakly in the proximal regions, hinge and pleura (see Figure S2A in Additional File 1). In these proximal regions, no differences in dMyc levels were visible between lgl^-/- clones and the adjacent cells; such clones were larger than those in the wing pouch, although they never formed tumours (Figure S1G in Additional File 1. Proximal: 3,254 ± 1,129 pixels vs distal: 2,197 ± 756 pixels, n = 35 each; P < 0.001) and showed low levels of cell death (not shown). Thus, low dMyc levels in lgl^-/- cells seem to affect growth and viability of these cells mainly in the distal region (wing pouch), where the surrounding tissue expresses high levels of dMyc.

dMyc overexpression within lglmutant clones unleashes their neoplastic potential

To investigate if the reduced level of dMyc protein in lgl mutant cells could be a cause of their elimination from the epithelium, we took advantage of the MARCM system [26] to express the UAS-dmyc construct [27] in lgl^-/- clones. As can be observed in Figure 3A, lgl^-/-; UAS-dmyc (dmyc^over) clones were round in shape and massively overgrown, with the largest clones located in the proximal regions of the disc. A 10-fold increase in clone area was observed with respect to lgl^-/- control clones induced through the same genetic system (average area of clones sampled in the hinge/pleura: 24,518 ± 4,237 pixels vs 2,725 ± 1,146 pixels respectively, n = 68 each). A gain in cell size has been reported for dmyc^over cells [28], raising the question of whether this could contribute in a significant way to lgl^-/-; dmyc^over clone area. Our analysis confirmed that cells in lgl^-/-; dmyc^over clones were larger than in lgl^-/- control clones (83 ± 41 pixels vs 67 ± 32, n = 80 each), but this gain in cell size accounts only for a small fraction of clone expansion, indicating that the massive overgrowth of lgl^-/-; dmyc^over clones is due mainly to a substantial increase in cell number.

dMyc ectopic expression is associated with an autonomous increase in apoptosis [29], so we expected to find high levels of cell death within lgl^-/-; dmyc^over clones. However, while small clones showed many active-Caspase 3 positive cells (Figure 3B, arrows), the majority of the large clones showed little or no active-Caspase 3 staining (Figure 3B, clone outlined). We indeed found a significant negative correlation between clone size and active-Caspase 3 staining (Figure 3C, Spearman's rank correlation coefficient 0.716, P < 0.001). Furthermore, active-Caspase 3 staining was evident immediately outside the overgrown clones (Figure 3B, arrowheads), indicating that dMyc overexpression conferred lgl^-/- cells the ability to out-compete neighbouring cells. dIAP1 protein was expressed in the overgrowing mutant clones (Figure 3B), but was undetectable in the smallest clones, where the active-Caspase 3 signal is the strongest (Figure 3B, arrows).

dmyc overexpression in lgl^-/- cells therefore had a different outcome depending on the region of the disc where clones were located. In the proximal regions (hinge and pleura), where the endogenous dMyc levels are very low and competitive interactions between lgl^-/- and wild-type cells are relaxed, some lgl^-/-; dmyc^over cells survived and formed clonal progeny that expressed dIAP1 and was therefore able to take advantage of dMyc's role in promoting biosynthesis and proliferation to out-compete surrounding cells (Figure 3B, arrowheads; see also the large clone in the hinge region in Figure 3D). lgl^-/-; dmyc^over clones located in the distal region (pouch) instead grew poorly, did not express dIAP1 protein and underwent untimely cell death (Figure 3B, arrows); the explanation for this behaviour could be because endogenous dMyc is expressed at high levels here, which is likely to prevent these lgl^-/-; dmyc^over clones from acquiring a competitive advantage. Other mechanisms could also be involved in generating such regional diversity in clonal growth, and further work is required to address this complex issue.

A staining for the apical marker aPKC indicated that lgl^-/-; dmyc^over cells displayed impaired apical-basal polarity (Figure 3D, see in particular the Z projection), a typical hallmark of epithelial neoplasias [30]. Staining for Laminin A, a major component of the basement membrane in both Drosophila and mammalian epithelia, revealed signs of discontinuity across the mutant clones (Figure 3E, arrows), suggesting that lgl^-/-; dmyc^over cells can acquire invasive properties, as has been demonstrated with other cooperative oncogenic models in Drosophila [25, 31, 32]. No pharate adults were recovered from larvae carrying lgl^-/-; dmyc^over clones (not shown), indicating that the differentiation of adult structures was impaired.

As discussed above, lgl^-/-; dmyc^over clones in the wing pouch behave differently from those in the wing hinge/pleura. In Figure 3D, the arrow indicates a clone in the wing pouch that does not exhibit tumourous properties. Notably, despite expression of UAS-dmyc transgene, clones do not show the high levels of dMyc protein observable in the clone located in the hinge. Since dmyc expression was induced using a heterologous promoter, the fact that dMyc protein is low in the lgl mutant clones in the wing pouch suggests that in this region an lgl-dependent regulation of dmyc at a post-transcriptional level is at work. Moreover, this result suggests that dMyc protein must be stably upregulated in lgl mutant clones in order to trigger overgrowth. This observation is similar to that of many human cancers, where Myc protein is highly and stably overexpressed [33]. Altogether, these data highlight a key role for dmyc in cooperating with the loss of a tumour suppressor to promote proliferation and to unleash the invasive potential of mutant cells.

dp110 kinase expression fails to rescue the defective growth of lglmutant cells

To understand whether dMyc could be substituted in rescuing the defective growth of lgl^-/- cells in an lgl^+/- background by other growth-promoting molecules, we expressed in lgl^-/- clones an activated form of the catalytic subunit of the Phosphoinositide-3 Kinase, dp110^CAAX [34], whose well-known role in cell growth and proliferation does not involve CC [14]. As can be observed in Figure S3B in Additional File 1, dp110 is not able to rescue the defective growth of lgl^-/- cells. Indeed, UAS-dp110^CAAX; lgl^-/- clones are comparable in size to lgl^-/- clones induced through the same technique (average area 1,974 ± 753 pixels vs 1,894 ± 916 pixels respectively, n = 40 each, P = 0.67); moreover, no significant impairments in cell polarity were observed (see Figure S3B in additional file 1, Z projection). As a control, a disc bearing dp110^CAAX clones is shown (Figure S3A in Additional file 1). Since it has been demonstrated that the ectopic expression of genes in the Insulin pathway, such as dp110, has no effect on ribosome biogenesis in Drosophila [35], it is plausible that the competitive properties of dMyc are due mainly to an increase in ribosome biogenesis [13, 15]; its effect in increasing growth, proliferation and survival of lgl mutant cells could be due to the boosting of their biosynthetic rates.

Inhibition of cell death is not sufficient for driving tumourous growth of lglmutant clones in the wing pouch region

Since several molecules associated with cell death were found to be deregulated in lgl mutant clones generated in an lgl^+/- background (see Figure 1), we tried to rescue lgl mutant clones viability by co-expressing dIAP1 or a dominant negative form of the Drosophila JNK, basket (bsk). As can be seen in Figure 4A, lgl^-/-; UAS-dIAP1 clones in the wing pouch still showed active-Caspase 3 signals (arrow). A statistical analysis performed on clones found in the wing pouch indeed demonstrated that they are smaller than their wild-type twins (average area 7,721 ± 944 pixels vs 11,583 ± 1,248 pixels respectively, n = 28; P < 0.001). Since the JNK pathway is activated in lgl mutant clones in the wing pouch (Figure 1C, arrow), it could significantly contribute to their death. It has however previously been reported that JNK activation also occurs in response to dIAP1 inactivation [36, 37]; therefore if the reverse interaction occurred it may be expected that JNK signalling would be absent in lgl^-/-; UAS-dIAP1 cells but, as shown in Figure 4B, there is strong upregulation of active JNK, as detected by pJNK staining (arrow), indicating that it contributes to the elimination of lgl mutant clones independently of dIAP1. Both pJNK and active-Caspase 3 were also visible in the mutant clones in the proximal regions of the wing disc (Figure 4A, B, hinge and pleura respectively, white arrowheads), but their phenotype was quite different from that of clones in the wing pouch; they were large, round-shaped and, as can be seen in Figure 4A, active-Caspase 3 signal was also present in several cells surrounding the mutant clone (grey arrowheads). A statistical analysis showed that such clones were larger than their wild-type twins (average area 15,243 ± 1,228 pixels vs 9,855 ± 1,991 pixels respectively, n = 22; P < 0.01) These phenotypes were associated with high dMyc levels, as shown in Figure 4C (arrow); thus JNK signalling may here be subverted from a pro-apoptotic to a pro-growth function, as has been demonstrated to occur in genetic contexts in which alterations in the polarity genes lgl, scrib and dlg are accompanied by an ectopic expression of activated (oncogenic) Ras or Notch [32, 38, 39].

We then blocked the JNK pathway inside the lgl^-/- clones by using a bsk^DN transgene [39] and found that the active-Caspase 3 signal was no longer observable in lgl^-/- clones, regardless of the region in which they were located (not shown), indicating that JNK signalling-induced cell death is the main pathway by which lgl mutant cells die. As can be seen in Figure 4D, UAS-bsk^DN; lgl^-/- clones in the wing pouch (outlined) expressed low levels of dMyc and did not form tumours; statistical analysis performed in this region showed that, despite the fact that UAS-bsk^DN; lgl^-/- clones no longer died, their size was smaller than that of their wild-type twins (average area 12,697 ± 8,491 pixels vs 21,957 ± 8,372 pixels respectively, n = 15, P < 0.001), demonstrating that even when cell death is blocked UAS-bsk^DN; lgl^-/- cells have a proliferative disadvantage with respect to the wild-type tissue.

In contrast to the wing pouch, UAS-bsk^DN; lgl^-/- clones in the proximal regions (hinge and pleura) showed high dMyc protein levels, lost polarity and overgrew (Figure 4E), indicating that the pro-growth role of the JNK pathway observed in polarity-compromised clones overexpressing activated Ras or Notch [32, 38, 39] appears not to be necessary in the wing disc for the tumourous growth induced by the cooperation between lgl mutation and dMyc oncoprotein. Altogether, these data show that the difference in clonal growth reported in section 2 for lgl^-/- cells in the wing pouch versus the hinge/pleura regions also occurs for lgl^-/- clones in which cell death has been inhibited and is also associated with the different levels of dMyc protein in the mutant versus the adjacent normal tissue.

The Intrinsic Tumour Suppressor pathway does not appear to be involved in lgl^-/-cells elimination in the wing pouch region

A recent study demonstrated that clones of scrib and dlg tumour suppressor mutants generated in wild-type imaginal discs are eliminated by the Intrinsic Tumour Suppressor (ITS) pathway, involving JNK-dependent apoptosis induced by an endocytic accumulation of the TNF homologue, Eiger (Egr) [8]. Since scrib and dlg are well-known lgl partners in regulating apical-basal cell polarity and proliferation and show similar neoplastic phenotypes in a homotypic background [2], it could be expected that lgl^-/- clones could be eliminated by a similar mechanism. To investigate this, we first looked for alterations in endocytosis in the lgl mutant clones by using an early-endosome reporter, Rab5, since it was observed by Igaki et al. [8] that Rab5-positive endosomes accumulated in scrib mutant clones in the eye disc correlating with pJNK staining. However, in the wing pouch region we did not observe changes in Rab5 levels within lgl^-/- clones with respect to the neighbours (see Figure 5A, arrowhead, and respective magnification), whereas outside the wing pouch a moderate increase in Rab5 levels in lgl^-/- clones was observed (see Figure 5A, arrow, and respective magnification). Since Igaki et al. [8] demonstrated that ITS depends on an autocrine TNF signalling, we then silenced the TNF homologue, egr, in lgl mutant cells by expressing a UAS-egrRNAi construct (for validation see Figure S4A, B in Additional File 1) and scored for changes in clone morphology, but no alterations were observed relative to the lgl^-/- clonal phenotype (Figure 5B-B"). lgl^-/-; UAS-egrRNAi clones in the wing pouch were comparable in size to lgl^-/- clones induced through the same system (average area pixels 2,954 ± 876 vs 3,177 ± 1,028 pixels respectively, n = 19 each; P = 0,58). Similar effects were seen in lgl mutant clones in discs in which the UAS-egrRNAi construct was expressed under the control of the hedgehog promoter in the whole posterior compartment, thereby also removing Egr protein in the lgl^+/- background (not shown). In Figure 5B and 5B', the apical and basal sections of a wing pouch are shown in which an lgl mutant clone is being basally extruded (in Figure 5B the position of the mutant clone is outlined), which also showed increased pJNK staining (Figure 5B"). The fact that JNK signalling is increased here suggests that, in this case, JNK pathway activation is not triggered by egr-mediated ITS. We next expressed in lgl mutant clones a dominant negative form of Rab5 [40] to block endocytosis, since in the Igaki et al. study [8] it was shown that blocking endocytosis decreased JNK signalling and cell death of scrib mutant clones. Indeed, we found that some lgl^-/- mutant clones expressing Rab5^DN in the proximal regions overgrew (13 out of 27 scored, Figure 5C), while clones in the wing pouch never did. Again, lgl^-/-; UAS-Rab5^DN clones (arrows) in the wing pouch were much smaller than the wild-type twins (see Figure S4C in Additional File 1) and expressed both active-Caspase 3 (Figure 5D, arrow) and pJNK (Figure 5E, arrow), distinct from what has been observed for scrib mutant clones in the eye disc, where Rab5^DN expression increased their growth [8]. Notably, all the overgrowing clones observed in the proximal regions of the wing disc were characterised by high dMyc protein levels (Figure 5C). Taken together, these data exclude egr-mediated ITS as the main mechanism responsible for lgl mutant clones elimination, at least in the wing pouch.

lgl^-/-cells grown in a Minute background overexpress dMyc, display a competitive behaviour and form malignant tumours

As we showed in section 5, lgl^-/- cells grew slower than wild-type cells in a clonal context, which is possibly the main reason why they are eliminated from the epithelium. To confirm the hypothesis for an active role of CC in restraining lgl^-/- clonal growth, we induced lgl mutant clones in the slow-dividing Minute background, to give them a proliferative advantage. As can be seen in Figure 6A-C"' and in the Additional File 2, lgl mutant clones in these larvae were able to overgrow (average area of clones sampled all across the wing disc: 12,974 ± 3,427 pixels vs 2,648 ± 1,211 pixels of lgl mutant clones in a wild-type background, n = 40 each) and cells became rounded, indicating that apical-basal polarity was lost, as can be seen in the Z projection where aPKC apical determinant spreads cortically. lgl^-/- clones showed high levels of dMyc protein (Figure 6A, arrow), which accumulated mainly in those outside of the wing pouch (see also Figure S5 in Additional File 1). dMyc upregulation is attributable to lgl loss of function because wild-type cells did not show any changes in dMyc levels when growing in a M/+ background (Figure S6 in Additional file 1; in particular, no dMyc accumulation was visible in a large clone in the hinge region, S6A, arrow). lgl^-/- clones in the wing pouch did not overgrow, showed a level of dMyc similar to that found in the M/+, lgl^+/- adjacent cells (Figure S5A in Additional File 1, clone outlined) and did not undergo apoptosis (not shown). In that region, dMyc endogenous expression is high (see Figure S5A in Additional File 1), so it is plausible that its upregulation by the loss of function of lgl was not sufficient to trigger the cell growth advantage necessary for driving clonal expansion at the expense of the adjacent tissue.

In contrast, in regions where dMyc abundance is very low, such as the hinge and the pleura, dMyc upregulation by lgl mutation might be sufficient for clone overgrowth. In large lgl^-/- clones in the hinge or pleura, dIAP1 was also expressed (Figure 6B') and active-Caspase 3 signal was evident in few cells in the mutant clones, but marked mainly groups of surrounding cells (Figure 6B, B', arrows), which were deficient in dIAP1 (Figure 6B, B', arrowheads). Moreover, cells lost polarity (Figure S5 in Additional file 1) and the basement membrane showed signs of discontinuity (arrows in Figure 6C'-C"', compared with the wild-type disc shown in C), indicating invasive behaviour. No pharate adults were recovered from mosaic larvae (not shown). Altogether, these data show that lgl^-/- cells grown in a M/+, lgl^+/- background can acquire a competitive advantage, which correlates with high dMyc expression and polarity loss. The upregulation of dMyc protein inside lgl^-/- clones induced in the M/+ background may be a consequence of an increase in dmyc transcript levels, since an upregulation of dmyc transcription, assayed by a dm>LacZ construct [41] (dm stands for diminutive, the gene encoding dMyc protein) was observed (Figure S7B in Additional File 1). In contrast, there was no correlation between dmyc transcription and protein abundance in lgl^-/- clones grown in the lgl^+/- background; the reduced levels of dMyc protein observed (Figure 2) did not result from a decrease in dmyc transcription (Figure S7A in Additional File 1), therefore post-transcriptional mechanisms must be at work in this case.

To determine the functional importance of dMyc upregulation in lgl^-/- clones induced in the M/+, lgl^+/- background, we silenced dmyc inside lgl^-/- clones using a UAS-dmRNAi construct (for validation, see Figure S2C in Additional File 1). Knockdown of dmyc expression in lgl^-/- clones prevented the cell polarity defects observed with lgl^-/- clones alone (Figure 6D, compared with Figure 6A-C, Z projections), even when the lgl mutant tissue occupied a larger proportion of the disc (Figure 6D'); no degradation of basement membrane occurred (not shown) and clones did not form tumourous masses, regardless of the region in which they were located. Adults were indeed recovered at the expected frequency (not shown). Concerning the cell death pattern in these clones, it ranged from a complete absence of active-Caspase 3 signal in the majority of discs analysed (Figure 6E) to scattered signals, either in lgl^-/- cells, surrounding cells, or in regions distant from mutant clones (Figure 6E', arrows). These results indicate that lgl^-/-; UAS-dmRNAi and M/+, lgl^+/- cells coexist in the same tissue without undertaking competitive interactions, possibly because both populations show a similar impairment in cell proliferation. Similar results were obtained by Wu and Johnston upon the induction of dm loss-of-function clones in a M/+ background [42]. M/+ tissue could also intrinsically possess a low level of dMyc protein, but we were not able to detect differences in dMyc protein levels between wild-type and M/+ clones throughout the wing disc (see Figure S6 in Additional File 1). Thus, the silencing of dmyc in lgl^-/- clones and the reduced ribosomal pool in the M/+, lgl^+/- background may make their levels of biosynthesis comparable. The ability of lgl mutant patches to grow despite dMyc deprivation might be due to the upregulation of other growth-promoting factors in the M/+, lgl^+/- context, which are however per se unable to provide lgl^-/- cells with tumourigenic features.

The oncogenic cooperation between lglmutation and dMyc protein is not tissue-specific

With the aim of assessing whether the oncogenic cooperation between lgl mutation and dMyc protein could be conserved in other Drosophila tissues, we investigated lgl^-/- clonal behaviour in the ovarian follicular epithelium, a monolayered adult tissue of somatic origin that surrounds the germ line of the egg chamber. Interestingly, the phenotype of lgl^-/- clones in the follicular epithelium appeared to be rather different from that observed in the wing imaginal epithelia, but similar to that previously reported in the whole mutant animal; in females bearing a homozygous temperature-sensitive lgl mutation, egg chambers invariantly show hyperproliferation of follicular cells that display loss of apical-basal polarity and migrate between the nurse cells [43]. We found that lgl mutant clones overproliferated and formed multilayers near the chamber poles (Figure 7A), consistent with previous studies [2, 44]. Active-Caspase 3 staining revealed that apoptosis was occasionally observable in egg chambers from Stage 8 onward in lgl^+/- cells at the clone border (Figure 7B). This cell death pattern is reminiscent of a mechanism of dMyc-induced CC. CC, however, has never been described in the follicular epithelium, so we induced dmyc^over Flp-out clones [45] in adult females and observed the cell death pattern in their egg chambers. As can be seen in Figure S8 in Additional File 1, the wild-type tissue underwent massive cell death, particularly when it was completely surrounded by dmyc^over cells (arrows), suggesting that dmyc is able to induce apoptosis in the surrounding cells also in this tissue. Indeed, lgl^-/- clones expressed dMyc protein in early egg chambers (Figure 7C) as well as at later stages (Figure 7C'), where endogenous protein is normally absent (see Figure S2B in Additional file 1). dmyc was also transcriptionally activated, as can be seen from dm>LacZ expression in Figure S7C in Additional file 1. Thus, lgl^-/- clonal behaviour in the follicular epithelium positively correlates with dmyc expression. dIAP1 levels in lgl^-/- follicular clones were similar to those seen in the lgl^+/- adjacent tissue (see Figure 7E), and no autonomous apoptosis was visible in lgl^-/- follicular cells (not shown). Further, we allowed clones to grow for four additional days in order to obtain stronger phenotypes. As can be seen in Figure 7D-D", dMyc protein was detectable in all lgl^-/- cells; clonal phenotype varied from larger areas of multilayered tissue (Figure 7D), which sometimes extended away from the poles (Figure 7D'), to clones that invaded into the nurse cells territory (Figure 7D", arrow). To determine the contribution of dmyc to the lgl mutant phenotype in this tissue, we lowered dMyc levels inside lgl^-/- cells using the UAS-dmRNAi construct. We observed a pronounced decrease both in clone size and number (visible clone area: 44% decrease in 60 clones at chamber poles analysed for each genotype; clone number: 74% decrease on a total of 20 pairs of ovaries for each genotype) and no multilayered tissue was seen, demonstrating that also in this epithelium dMyc protein is required for lgl mutant clones to grow as malignant tumours.

Mosaic analyses

For the list of mutant and transgenic flies utilised, please see Additional File 1.

lgl^27S3 and lgl⁴ mutant phenotypes are fully rescued by expression of UAS-lgl transgenes, resulting in normal larval and adult structures [22, 5].

For mosaic analyses, either the Flp-FRT [54] or MARCM [26] systems were used. For larval staging, freshly hatched larvae were collected in a one-to-four hour time window. For the twin analysis, larvae were heat shocked 48 hours After Egg Laying (AEL) for one hour at 37°C and adult females were heat shocked one to two days after eclosion for one hour at 37°C. For the MARCM system, larvae were heat shocked 48 hours After Egg Laying (AEL) for 20 minutes at 37°C. Minute individuals were always heat shocked at 72 hours AEL to compensate for developmental delay. For the Flp-out system [45], adult females were heat shocked one to two days after eclosion for eight minutes at 37°C. For all the systems used, tissues were collected 72 hours after the heat shock both for larvae and adults, unless otherwise specified. All crosses were kept at 25°C.

Immunofluorescence

Imaginal discs and ovaries were fixed and stained according to standard protocols. The following antibodies and dilutions were used: rabbit anti-active-Caspase 3 (1:200, (9664S) Cell Signaling Technology, Inc., Danvers, MA, USA); mouse anti-dIAP1 (1:100, BA Hay); rabbit anti-Laminin A (1:200, Y Kitagawa); rabbit anti-Scribble (1:100, CQ Doe); mouse anti-dMyc (1:5) [55]; mouse anti-βGal (1:25, [40-1a] DSHB, University of Iowa, Iowa City, IA, USA); rabbit anti-βGal (1:50, F Graziani); rabbit anti-aPKCζ (1:200, [sc-216] Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA); mouse anti-phospho-JNK (1:100, [G9 clone] Cell Signaling Technology Inc., Danvers, MA, USA); rabbit a-Egr (1:500, M Miura). Alexa Fluor 555 or 568 goat anti-mouse and anti-rabbit (1:200, Invitrogen Corporation, Carlsbad, CA, USA) and Cy5-conjugated goat anti-mouse and anti-rabbit (1:100, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) were used as secondary antibodies. DAPI and TOPRO staining were used to detect DNA and phalloidin staining to detect F-actin. Samples were analysed by laser confocal microscopy (Leica TSC SP2, Leica Microsystems SpA, Milan, Italy) and entire images were processed with Adobe Photoshop software. ImageJ free software from NIH, Bethesda, MD, USA was used for rebuilding the projections along the Z axis starting from 25-45 Z stacks.

Measurements and statistical analysis

To measure clone area (in pixel²) and fluorescence intensity (in calibrated units), ImageJ free software from NIH, Bethesda, MD, USA, was used. Fluorescence intensity (signal intensity) was calculated as the average gray value within the selection. The average and the standard deviation alone or the t-Student test P value are given in the text. For the correlation between clone size and active-Caspase 3 staining in Figure 3, the Spearman's rank correlation coefficients were calculated using the software XLSTAT (available at http://www.xlstat.com). To measure average cell area in wing disc clones, TOPRO stained nuclei and clone area were scored in confocal series and the relative cell area (in pixels²) was calculated.