The following stocks were used: UAS-Ras ^V12 (Lee et al., 1996); nub-Gal4 (Calleja et al., 1996); hid 5′F-GFP (Tanaka-Matakatsu et al., 2009); UAS-DER^DN, UAS-Diap1, UAS-puc, ap-Gal4, puc-LacZ, rpr-LacZ, and brk-LacZ (Bloomington Drosophila Stock Center); mys ^RNAi , UAS-diβ (Martin-Bermudo and Brown, 1999) and hid ^RNAi (Vienna Drosophila RNA-i Center). To generate mutant clones in the wing disc we used the FRT/FLP technique (Chou and Perrimon, 1992). Mutant clones were marked by the absence of GFP. The following mutant alleles and chromosomes were used: mys ¹¹ [also known as mys ^XG43 (Bunch et al., 1992)] and e22c-Gal4 UAS-flipase (Duffy et al., 1998). The e22c-Gal4 driver is expressed in some posterior wing disc cells and was therefore used in combination with UAS- flp to generate large mys clones. Flies were raised at 25°C.

Wing imaginal discs were stained using standard procedures and mounted in Vectashield (Vector Laboratories, Burlingame, CA, United States). The following primary antibodies were used: goat anti-GFP^FICT (Abcam, 1:500), rabbit anti-caspase Dcp1 (Cell Signaling; 1:100), rabbit anti-pJNK (Promega, 1:200), mouse anti-βGal (Promega, 1:1000), mouse anti-βPS (Developmental Studies Hybridoma Bank, DSHB, University of Iowa, United States, 1:50) and mouse anti-myc (Promega, 1:100). The secondary antibodies used were Alexa fluor 488, (Molecular Probes^TM) and Cy3 and Cy5 (Jackson ImmunoReseach Laboratories, Inc.) at 1: 200. DNA was labelled using Hoechst (Molecular Probes, 1:1000). Confocal images were obtained using a Leica SP5-MP-AOBS, equipped with a Plan-Apochromat 40X oil objective (NA 1.4).

To quantify fluorescent intensity of the different markers, fluorescent signaling was measured from maximum projections of 30 Z-confocal sections taken with 1 μm of interval per wing disc and genotype. Quantifications were made in the wing pouch region, which was manually selected using the FIJI-Image J line selection tool. Microscope settings were maintained between imaging sessions in each experiment. The background value taken from cell-free region was subtracted from all data series. Measurements of whole fluorescence intensity were done dividing the mean of all included pixels intensity by the outlined cell area, using the FIJI-Image J measure tool. Student’s t tests were used for statistical comparisons of fluorescence intensity values.

To deepen our understanding of the role of integrins as survival factors during epithelia morphogenesis, we tested whether removing integrins in Drosophila wing disc epithelial cells affected their survival. The Drosophila genome contains two integrin β subunits, βPS and βν [reviewed in (Brown, 1993; Yee and Hynes, 1993). βPS, encoded by the gene myospheroid (mys), is the only β chain present in the wing imaginal disc (Brabant et al., 1996); Supplementary Figure S1]. First, we generated mosaic wing disc epithelia containing clones of cells homozygous for the null allele mys ^XG43 (from now on mys cells, white arrow in Figures 1A–B”). To detect apoptosis, we used an antibody to cleaved effector caspase Dcp-1 (Yu et al., 2002). In control wing imaginal discs, Dcp-1 activity is very low, being detected only occasionally in a few cells scattered throughout the disc (Figure 1A, n = 20, where n represents the number of wing discs analyzed). In contrast, all wing discs containing mys clones showed apoptotic cells within the clone area (Figure 1B, n = 18). Second, we reduced integrin levels in larger areas of the disc using the nubbin (nub) Gal4 line to express a mys RNAi (nub>mys RNAi) in the wing pouch (inset in Figure 1C) region in the center of the disc that gives rise to the adult wing, indicated with a dotted white circle in all figures (Brand and Perrimon, 1993; Calleja et al., 1996). Immunostaining of control (n = 30) and experimental nub>mys ^RNAi wing imaginal discs (n = 36) with antibodies against the βPS protein and cleaved Dcp-1 showed that the expression of the mys specific RNAi in the wing pouch caused a strong reduction in βPS protein levels (Figures 1C,C’) and a robust increase in apoptosis (Figures 1C,C’’,E). Thus, integrin expression in wing disc epithelial cells prevent them from undergoing cell death. It is known that embryonic epithelial cells undergoing apoptosis are removed from the epithelium by extrusion rather than phagocytosis (Rosenblatt et al., 2001). Similarly, integrin-defective cells were extruded (Figures 1C–C’’) and accumulated between the basal surface of the disc epithelium and the ECM, as seen with an antibody against the ECM component Perlecan (Supplementary Figure S2).

Integrins can mediate both signaling and adhesion [reviewed in (Adams and Watt, 1993)]. To address which integrin function is required to facilitate cell survival in the wing imaginal disc, we used a chimeric integrin (diβ) that lacks integrin-adhesive function but maintains its ability to signal (Martin-Bermudo and Brown, 1999; Dominguez-Gimenez et al., 2007). We had previously shown that expression of diβ in the wing disc inhibited the basal localization of endogenous integrin [(Supplementary Figures S3B,B’; (Dominguez-Gimenez et al., 2007)]. Here, we found that it caused an increase in apoptosis (Supplementary Figures S3B,B”), strongly suggesting that integrin signaling is not sufficient to prevent cell death in detached wing disc cells.

Altogether, these results propose that integrin-mediated adhesion to the ECM promotes cell survival in the wing disc epithelium by regulating caspase activity. To further support this hypothesis, we tested whether overexpression of one of the Drosophila IAP (inhibitor of apoptosis) proteins, Diap1, suppressed the cell death caused by expression of mys RNAi and found that it did (Figures 1D-D’’,E, n = 35).

Previous experiments have shown that overexpression of hid or rpr induces downregulation of Diap1 levels and apoptosis in Drosophila wing imaginal discs (Milan et al., 1997; Bergantinos et al., 2010; Yoo et al., 2002). As Diap1 overexpression rescues cell death due to integrin downregulation, integrins could promote cell survival through the regulation of hid and/or rpr expression. To test this, we analysed hid and rpr expression in control and nub>mys ^RNAi wing discs, using the hid 5′F-GFP [from now on hidGFP, (Tanaka-Matakatsu et al., 2009)] and rprLacZ (Nordstrom et al., 1996) transcriptional reporters. hidGFP levels in control discs (hidGFP; nubGal4) are hardly detectable, consistent with a low occurrence of cell death in this tissue (Tanaka-Matakatsu et al., 2009); Figures 2A-A’’’,E]. This expression was dramatically upregulated in nub > mys ^RNAi discs (n = 25, Figures 2B-B’’’). In contrast, the normal expression pattern of rprLacZ, confined to a small region of the notum and two stripes in the D-V and A-P boundaries [(Nordstrom et al., 1996), Supplementary Figures S4A,A’], was not altered in nub > mys ^RNAi discs (n = 28, Supplementary Figures S4B,B’). These results suggest that integrins block hid, but not rpr expression in normal wing disc epithelial cells to promote cell survival. Furthermore, consistent with the idea that Drosophila hid acts upstream of Diap1 (Wang et al., 1999; Goyal et al., 2000), we found that expression of Diap1 was not able to rescue the increase in hidGFP levels observed in mys ^RNAi cells (n = 24, Figures 2C–C’’’).

To further confirm hid involvement in integrin loss of function-dependent cell death, we tested whether reducing hid levels could rescue apoptosis in nub > mys ^RNAi discs. We found that expression of an RNAi against hid (hidGFP; nub > mys ^RNAi ; hid ^RNAi ) resulted in a significant reduction of mys RNAi-induced apoptosis (n = 20, Figures 2A-A’’’, B-B’’’, D-D’’’, E). These findings demonstrate that integrins promote cell survival in wing imaginal discs through the negative regulation of hid expression.

Altering the levels of Decapentaplegic (Dpp), a Drosophila transforming growth factor β homologue), in wing discs results in upregulation of brinker (brk), a transcriptional repressor that prevents apoptosis (Adachi-Yamada et al., 1999; Moreno et al., 2002). However, reducing integrin function does not seem to affect Dpp signalling, as the expression pattern of the brk reporter brkLacZ was not altered in nub > mys ^RNAi wing discs (n = 26, Supplementary Figure S5B).

As mentioned in the introduction, the role of JNK in anoikis is controversial, as, depending on the cell type and cellular context, it can be either essential (Frisch et al., 1996) or dispensable (Khwaja and Downward, 1997). In Drosophila wing discs, JNK activation triggers apoptosis in response to several stimuli, including distortion of positional information (Adachi-Yamada et al., 1999), TNF signalling (Igaki et al., 2002; Moreno et al., 2002) or irradiation (Luo et al., 2007). To study the role of the JNK pathway in apoptosis induced by integrin loss of function, we analyzed JNK activation using an antibody against phosphorylated JNK (pJNK, Figure 3). Compared to controls (n = 16, Figures 3A,A’,D), pJNK levels were upregulated in nub > mys ^RNAi discs (n = 18, Figures 3B,B’,D), suggesting that integrins control JNK activity in the wing disc. This was further confirmed using a reporter that monitors the transcription of a target of the JNK pathway, the JNK inhibitor puckered (pucLacZ) (Adachi-Yamada et al., 1999). We found that while in wild type wing discs, pucLacZ expression was restricted to a small proximal region (Adachi-Yamada et al., 1999), Supplementary Figures S6A,A’), in nub > mys ^RNAi wing discs, pucLacZ levels were strongly upregulated in the pouch (n = 28,Supplementary Figures S6B,B’).

In some contexts, JNK signalling can be activated as a consequence of apoptosis (Kuranaga et al., 2002). To test whether activation of JNK lies upstream or downstream of loss of integrin function, JNK activation was assessed in nub > mys ^RNAi ; Diap1 wing discs (Figure 3). We found that pJNK (n = 20, Figures 3C,D) and pucLacZ (n = 25, Supplementary Figures S6C,C’) levels were still upregulated despite rescue of apoptosis (Figure 1D), suggesting that JNK activation is upstream of cell death due to integrin loss of function in wing disc cells. Stress induced cell death in the wing disc activates JNK upstream of caspase activator proteins, which, in turn, reinforce JNK activation in a positive feedback loop to amplify the apoptotic response (Shlevkov and Morata, 2012). Here, we found that, in fact, the increase in JNK activity of nub > mys ^RNAi ; Diap1 discs was lower than the one observed in nub > mys ^RNAi discs (Figures 3B–D). This result suggests that the feedback loop between JNK activity and apoptosis observed in stress-induced apoptosis may also operate in integrin loss of function-dependent apoptosis.

To test whether the JNK pathway mediates cell death due to integrin loss of function in wing imaginal discs, we analyzed if overexpression of the JNK inhibitor puc suppressed the cell death caused by mys RNAi (nub > mys ^RNAi ;puc) and found that it did (Figures 4A–D, n = 24).

The relationship between JNK signalling, pro-apoptotic proteins and caspase activation is context-dependent. Thus, the JNK pathway can act upstream or downstream of pro-apoptotic genes such as rpr or hid. For instance, experiments in Drosophila S2 cells suggested that Rpr could stimulate JNK activation through DIAP1 degradation (Kuranaga et al., 2002). In contrast, JNK signalling promotes hid expression (and apoptosis) in eye discs, and it is required for rpr-reporter induction in response to radiation in wing discs (Kuranaga et al., 2002; Moreno et al., 2002). In our system, inhibition of JNK signalling via puc overexpression was able to suppress hid upregulation in nub > mys ^RNAi wing discs to a large extent (n = 17, Figures 4C-C’’’,E). Altogether, these results allow us to conclude that reduction of integrin function in wing discs leads to apoptosis by stimulating hid transcription through JNK activation.

The EGFR activates a network of signaling pathways promoting survival in many different cellular contexts [reviewed in (Henson and Gibson, 2006)]. In addition, experiments mainly from cell culture have revealed synergistic cooperation between growth factors and integrins in cell survival (Reginato et al., 2003; Miranti and Brugge, 2002). For instance, EGFR promotes survival through regulation of the Ras/MAPK pathway and consequent inhibition of the proapoptotic gene hid in the eye disc (Bergmann et al., 1998; Kurada and White, 1998; Yu et al., 2002). Moreover, wing disc mutant clones for EGFR or for some of their dowmstream effectors, such as Ras, are smaller than their twin spot, suggesting a role for the EGFR pathway in cell survival (Diaz-Benjumea and Garcia-Bellido, 1990; Diaz-Benjumea and Hafen, 1994; Zecca and Struhl, 2002). To directly address a role for EGFR in cell survival in the wing disc, we expressed an EGFR dominant negative construct (UAS-DER^DN) under the control of the nub-Gal4 line and found it led to increase cell death in the pouch region (nub > DER^DN, n = 21, Figures 5A, A’, C, C’, E). However, in contrast to what happens in nub > mys ^RNAi wing discs (Figures 5B, B’, E), where cell death was found distributed all over the pouch, expression of DER^DN restricted apoptosis to regions of high EGFR activity (Gabay et al., 1997), such as the wing margin and the presumptive vein territories (yellow arrow and asterisks in Figures 5A, A’, C, C, respectively). In addition, and contrary to what happens in the eye, EGFR did not seem to exert its pro-survival function through regulation of hid or rpr transcription, as hidGFP or rpr-lacZ expression were not altered in nub > DER^DN wing discs (Figures 5A, A’, C, C”; Supplementary Figure S7). We thus conclude that the EGFR pathway has a limited role in the control of cell survival in the wing disc that is independent of the transcriptional regulation of the pro-apoptotic genes hid and rpr. Finally, we examined a possible cooperation between integrins and the EGFR pathway to mediate cell survival. We found that the simultaneous co-expression of mys ^RNAi and DER^DN in wing discs (nub > mys ^RNAi ; DER^DN) enhanced the cell death phenotype caused by the expression of any of them on their own (Figures 5B–E), strongly suggesting that integrins and the EGFR pathway act in paralell to promote cell survival in wing imaginal discs.

Suppression of anoikis after detachment of cancer cells from the ECM is a key step in tumor metastasis [reviewed in (Simpson et al., 2008)]. In fact, the ability of oncogenic Ras to suppress anoikis has long been considered critical to Ras transformation. However, recent evidence suggests that Ras and other oncogenes can induce both pro- and anti-apoptotic signals depending on the cell type and context [reviewed in (Cox and Der, 2003)]. In Drosophila, Ras overactivation renders cells refractory to stress- and irradiation-induced apoptosis (Bergmann et al., 1998; Kurada and White, 1998; Pinal et al., 2018). To test whether oncogenic Ras could protect cells from anoikis in the wing imaginal disc, we overexpressed an activated form of DRas, Ras^V12 (Karim and Rubin, 1998) and found it substantially suppressed the cell death phenotype caused by integrin depletion (nub > mys ^RNAi, ras ^V12; Figures 6A-A’’, 6B–B’’, C–C’’, D, n = 18). These results demonstrate that oncogenic Ras is able to suppress anoikis in wing imaginal discs. In epithelial cell lines, oncogenic Ras confers resistance to anoikis partly by downregulating the expression of proapoptotic Bcl-2 members (Rak et al., 1995) and by preventing downregulation of antiapoptotic proteins (Rosen et al., 2000). In Drosophila, it appears that the Ras pathway regulates Hid activity at the level of both, protein phosphorylation and gene expression, in hid-induced apoptosis (Bergmann et al., 1998; Kurada and White, 1998). However, we found that, in the Drosophila wing disc epithelium, the ability of oncogenic Ras to confer resistance to anoikis does not involve transcriptional regulation of hid, as hidGFP levels were not altered in nub > mys ^RNAi; Ras^V12 (n = 21, Figures 6A’’’, C’’’) discs compared to nub > mys ^RNAi (n = 20, Figure 6B’’’).

Ectopic expression of Ras^V12 in the dorsal compartment of wing imaginal discs, by means of the apterous Gal4 (ap) Gal4 line, causes tissue overgrowth and cell shape changes, which results in the formation of ectopic folds [ap > Ras^V12, n = 20, Figures 7A,A’,C,C’, (Prober and Edgar, 2000; Soler Beatty et al., 2021). Reducing integrin function also alters cell shape causing a mild folding phenotype (Dominguez-Gimenez et al., 2007), Figures 7B, B’ E, E’]. Here, we found that co-expression of mys ^RNAi and Ras^V12 enhanced the cell shape and folding phenotype caused by the sole expression of any of them (ap > mys ^RNAi ; Ras^V12, n = 20, Figures 7D,D’).

In this work, we show that reducing integrin expression ensued basal extrusion of the dead cells (Figures 1C–C’’, Figures 7F,G). However, even though oncogenic K-Ras MCDK cells survive and extrude basally, a mechanism proposed to initiate invasion into surrounding tissues (Slattum et al., 2014), we found that expression of Ras^V12 on its own in wing disc cells (ap > Ras^V12) did not result in cell extrusion (Figure 7H). In striking contrast, the expression of Ras^V12 in integrin mutant cells not only allowed their survival (Figure 6) but also induced their basal extrusion beneath the surface of the wing disc epithelia (ap > mys ^RNAi ; Ras^V12, n = 20, Figure 7I). Furthermore, because depletion of both integrins and hid (ap > mys ^RNAi ; hid ^RNAi ) did not result in cell extrusion (Figure 7J), the basal extrusion of mys; Ras^V12 wing disc cells was not due only to the ability of Ras^V12 to rescue apoptosis. From the above results we conclude that integrins prevent extrusion of transformed Ras^V12 cells.