Over the past decade an intriguing connection between asymmetric cell division, stem cells and tumorigenesis has emerged. Neuroblasts, which are the neural stem cells of the Drosophila central nervous system, divide asymmetrically and constitute an excellent paradigm for investigating this connection further. Here we show that the simultaneous loss of the asymmetric cell division regulators Canoe (afadin in mammals) and Scribble in neuroblast clones leads to tumor-like overgrowth through both a severe disruption of the asymmetric cell division process and canoe loss-mediated Ras-PI3K-Akt activation. Moreover, canoe loss also interacts synergistically with scribble loss to promote overgrowth in epithelial tissues, here just by activating the Ras-Raf-MAPK pathway. discs large 1 and lethal (2) giant larvae, which are functionally related to scribble, contribute to repress the Ras-MAPK signaling cascade in epithelia. Hence, our work uncovers novel cooperative interactions between all these well-conserved tumor suppressors that ensure tight regulation of the Ras signaling pathway.
INTRODUCTION
A direct link between defects in the process of asymmetric cell division (ACD) and tumorigenesis was demonstrated for the first time in the neural stem cells of the Drosophila larval brain, where mutations in key ACD genes caused tumor formation and even metastatic growth (Caussinus and Gonzalez, 2005). Drosophila neural stem cells, called neuroblasts (NBs), have become one of the best paradigms in which to analyze the process of ACD (Doe, 2008; Knoblich, 2008). NBs divide asymmetrically to give rise to another NB that retains the self-renewal capacity of the mother stem cell, and a smaller daughter cell called the ganglion mother cell (GMC), which is committed to initiating a process of differentiation. The GMC will divide only once more to generate two neurons or glial cells. A complex protein network regulates this process, ensuring the correct orientation of the mitotic spindle and the basal sorting of cell fate determinants, such as Numb and Prospero (Pros). This regulatory network includes the Par complex, formed by the highly conserved partitioning defective proteins Par6 and Par3 [Bazooka (Baz) in Drosophila] and atypical protein kinase C (aPKC), which accumulates at the apical cortex of metaphase NBs (Petronczki and Knoblich, 2001; Schober et al., 1999; Wodarz et al., 2000, 1999). The Par complex associates, through the adaptor protein Inscuteable (Insc), with the Partner of Insc (Pins; LGN or Gpsm2 in mammals) complex, which thereafter orchestrates the proper orientation of the mitotic spindle (Parmentier et al., 2000; Yu et al., 2003). Crucial for this is the activation of the Warts kinase and consequent phosphorylation of the Pins partner Canoe (Cno) (Keder et al., 2015; Speicher et al., 2008). A Rap1-Rgl-Ral small GTPase complex also participates upstream of Cno, anchoring it to the membrane, in regulating asymmetric NB division in the Drosophila embryo (Carmena et al., 2011).
In the larval central brain, in the dorsoposterior region of each brain hemisphere, a particular type of NB lineage, called type II NBs (NBIIs), has been described (Bello et al., 2008; Boone and Doe, 2008; Bowman et al., 2008). These NBIIs divide to self-renew and to give rise to an intermediate neural precursor (INP), a transit-amplifying cell, instead of a GMC; the INP, after a period of maturation, divides asymmetrically generating another INP and a GMC. NBII lineages are especially susceptible to overgrowth after a defective asymmetric NB division (Bowman et al., 2008). In addition to ACD regulators having been revealed as tumor suppressors in the larval brain when they are compromised, other Drosophila genes originally identified as tumor suppressors have subsequently proven to be key regulators of ACD. This is the case for lethal (2) giant larvae [l(2)gl], discs large 1 (dlg1), scribble (scrib) and brain tumor (brat) (Albertson and Doe, 2003; Bello et al., 2006; Betschinger et al., 2006; Bilder et al., 2000; Bowman et al., 2008; Gateff, 1978, 1994; Lee et al., 2006; Ohshiro et al., 2000; Peng et al., 2000). All this evidence strongly supports a functional link between defects in the process of ACD and tumor formation. Remarkably, all these ACD regulators identified in Drosophila have homologs in vertebrates and most of them have been linked with various human cancers (Gómez-López et al., 2014). Moreover, a connection between impairment of ACD and tumorigenesis is also emerging in mouse models and human tumors (Chen et al., 2014; Cicalese et al., 2009; Hwang et al., 2014; Ito et al., 2010; Sugiarto et al., 2011; Wu et al., 2007).
In this study, using the neural stem cells of the Drosophila larval brain as a model system, we investigated a potential function of the ACD modulator Cno (afadin, or AF6 in mammals) in tumorigenesis (Miyamoto et al., 1995; Prasad et al., 1993; Speicher et al., 2008). Cno and afadin are cytoplasmic proteins normally associated with adherens junctions in epithelial cells, and they share a conserved modular structure including two Ras-associated (RA) domains (Kuriyama et al., 1996). We show that cno loss synergistically interacts with the loss of scrib (Scrib in mammals), a well-known tumor suppressor gene and ACD regulator (Albertson and Doe, 2003; Bilder et al., 2000). The simultaneous loss of cno and scrib in NBII clones led to tumor-like overgrowth through both a severe disruption of the ACD process and a cno loss-mediated Ras-PI3K-Akt activation. Moreover, in epithelial tissues cno also synergistically interacted with scrib to promote overgrowth, in this tissue by activating the Ras-Raf-mitogen activated protein kinase (MAPK) pathway. dlg1 and l(2)gl, which are functionally related to scrib, did not display such synergism with cno. However, they contributed to repress Ras-MAPK signaling in epithelia. The establishment of distinct protein interaction networks among all these tumor suppressor genes, cno and the scrib module, and the Ras pathway in different developmental contexts is discussed.
RESULTS
cno null mutant NBII clones show disrupted ACD but do not overgrow
We previously found that Cno is expressed in embryonic NBs, where, along with other ACD regulators, it ensures a correct asymmetric NB division. To investigate a potential function of cno as a tumor suppressor gene, we first analyzed its expression in larval brain NBs. Cno was detected in both NBI and NBII lineages, and we focused on the latter, which are more prone to manifest tumor-like overgrowth when ACD modulators are compromised (Bowman et al., 2008). Cno was observed along with the transcription factor Deadpan (Dpn), a marker of NBs and INPs, in the cytoplasm of these progenitors at interphase and in a cortical apical crescent at metaphase (Fig. 1A-B′). Cno was also detected at centrosomes throughout mitosis (Fig. 1B-C′).
cno null NBII clones do not overgrow. (A,A′) GFP-labeled endogenous Cno is at low levels in interphase NBs (arrow) and INPs (arrowhead) of NBII lineages (dotted line); Dpn marks progenitors (NBs and INPs). (B-C′) At metaphase, Cno is enriched in cortical crescents in progenitors (green arrows); Cno is also detected in centrosomes (green arrowheads), as labeled by Centrosomin (Cnn); phospho-Histone H3 (PH3) marks mitotic cell DNA. B and C show the same clone at different confocal planes of focus. (D-F) NBs (dotted line indicates NB diameter) in cnoR2 versus WT clones (D,E) are significantly smaller (F; n, number of clones from 5 WT or 8 cnoR2 brains analyzed). (G-I) In WT clones, only one Dpn+ PntP1+ NB (dotted line) is present; the inset shows the NB expressing both factors (G); in cnoR2 clones two or three NBs are occasionally detected (H,I; n, number of clones of 9 WT and 11 cnoR2 brains). (J,K) Numbers of mature INPs (Dpn+ Ase+) (J) and GMCs (Ase+ Pros+) (K) are significantly lower in cnoR2 clones (n, number of clones from 5 or 6 WT and from 8 or 6 cnoR2 brains in J or K). (L,M) The progenitor cell mitotic index is significantly lower in cnoR2 clones (L), whereas there are no significant differences in the number of interphase progenitors (M); n, number of clones of 6 or 5 WT and 6 or 5 cnoR2 brains in L or M. Data in F,J-M were analyzed by Student's t-test. ***P<0.001, **P<0.01; n.s, not significant. Error bars indicate s.d. (N) Comparison of WT and cnoR2 clones. (O-Y) Defects and quantification of ACD regulator (red) localization (arrowheads) in cnoR2 clones; in WT clones, regulators are enriched in cortical crescents (arrows) of metaphase/anaphase progenitors. PCs, progenitor cells; NP, not present; CM, cortical mislocalization. Scale bars: 10 µm.
cno null NBII clones do not overgrow. (A,A′) GFP-labeled endogenous Cno is at low levels in interphase NBs (arrow) and INPs (arrowhead) of NBII lineages (dotted line); Dpn marks progenitors (NBs and INPs). (B-C′) At metaphase, Cno is enriched in cortical crescents in progenitors (green arrows); Cno is also detected in centrosomes (green arrowheads), as labeled by Centrosomin (Cnn); phospho-Histone H3 (PH3) marks mitotic cell DNA. B and C show the same clone at different confocal planes of focus. (D-F) NBs (dotted line indicates NB diameter) in cnoR2 versus WT clones (D,E) are significantly smaller (F; n, number of clones from 5 WT or 8 cnoR2 brains analyzed). (G-I) In WT clones, only one Dpn+ PntP1+ NB (dotted line) is present; the inset shows the NB expressing both factors (G); in cnoR2 clones two or three NBs are occasionally detected (H,I; n, number of clones of 9 WT and 11 cnoR2 brains). (J,K) Numbers of mature INPs (Dpn+ Ase+) (J) and GMCs (Ase+ Pros+) (K) are significantly lower in cnoR2 clones (n, number of clones from 5 or 6 WT and from 8 or 6 cnoR2 brains in J or K). (L,M) The progenitor cell mitotic index is significantly lower in cnoR2 clones (L), whereas there are no significant differences in the number of interphase progenitors (M); n, number of clones of 6 or 5 WT and 6 or 5 cnoR2 brains in L or M. Data in F,J-M were analyzed by Student's t-test. ***P<0.001, **P<0.01; n.s, not significant. Error bars indicate s.d. (N) Comparison of WT and cnoR2 clones. (O-Y) Defects and quantification of ACD regulator (red) localization (arrowheads) in cnoR2 clones; in WT clones, regulators are enriched in cortical crescents (arrows) of metaphase/anaphase progenitors. PCs, progenitor cells; NP, not present; CM, cortical mislocalization. Scale bars: 10 µm.
Given the expression pattern of Cno in NBII lineages, we next investigated the effect of eliminating its function by the mosaic analysis with a repressible cell marker (MARCM) technique (Lee et al., 1999). GFP-labeled wild-type (WT) and cnoR2 null mutant clones were subjected to a comparative analysis in which different parameters, such as the cellular composition of the clone, the mitotic index of progenitors and the distribution of ACD regulators, were studied. A very characteristic feature of cnoR2 clones was the reduced size of the NB in 100% of the clones analyzed (Fig. 1D-F). In addition, whereas only one NB, characterized by the simultaneous expression of the transcription factors Dpn and Pointed P1 (PntP1), is always observed in WT clones, extra NBs were occasionally detected in cnoR2 mutant clones (Fig. 1G-I,N). However, fewer mature INPs and GMCs were present in cnoR2 clones compared with WT clones (Fig. 1J,K,N). Concomitant with this observation, the mitotic index of progenitor cells was significantly lower in cnoR2 clones (Fig. 1L). Intriguingly, no significant differences were found in the number of progenitors at interphase (Fig. 1M). These results suggest that, even though in cnoR2 clones most INPs are entering the cell cycle, not all of them proceed to the mitotic phase or do so at a slower rate than WT INPs.
We next tested the localization of ACD regulators in NBs and INPs of cnoR2 clones. All of the apical regulators analyzed, the adaptor protein Miranda (Mira), which is required for the basal sorting of the determinants Pros and Brat, but not the cell fate determinant Numb, were mislocalized or absent in metaphase/anaphase NBs of cnoR2 clones (Fig. 1O-Y). Hence, disruption of asymmetric NB/INP division in cnoR2 clones might explain the abnormal cellular composition found within these clones.
scrib mutant NBII clones are eliminated by JNK pathway-mediated apoptosis
Despite the ACD defects observed in cnoR2 clones they did not show tumor growth. Similarly, the loss of other well-known tumor suppressor genes, such as scrib, in somatic clones does not lead to any massive overgrowth in epidermal cells; on the contrary, scrib mutant clones are eliminated by JNK pathway-mediated apoptosis induced by the surrounding WT tissue (Bilder et al., 2000; Brumby and Richardson, 2003; Igaki et al., 2006). We therefore analyzed the behavior of scrib clones in larval brain NBII lineages, a context in which it has not previously been described. scrib1 NBII clones were found in only 55% of brains (n=9), as compared with WT clones which appeared in 100% of the brains analyzed (n=10). In addition, scrib1 NBII clones were frequently very small (Fig. 2B,B″) compared with WT clones (Fig. 2A,A″). These observations suggested that scrib1 NBII clones are dying.
scrib NBII clones are eliminated by JNK-mediated apoptosis. (A-E) GFP-labeled NBII clones (green, single channel in A″-E″) of WT (A), scrib1 (B), bskDN (C) and bskDN in a scrib1 mutant background (D,E) stained for Dpn and diP-JNK (red, single channel in A′-E′). (A-B″) scrib1 clones are smaller than WT clones and upregulate diP-JNK. (C-C″) bskDN expression in a WT background has no apparent phenotype. (D-E″) bskDN expression partially rescues the scrib1 phenotype (two different examples are shown). Scale bar: 10 µm.
scrib NBII clones are eliminated by JNK-mediated apoptosis. (A-E) GFP-labeled NBII clones (green, single channel in A″-E″) of WT (A), scrib1 (B), bskDN (C) and bskDN in a scrib1 mutant background (D,E) stained for Dpn and diP-JNK (red, single channel in A′-E′). (A-B″) scrib1 clones are smaller than WT clones and upregulate diP-JNK. (C-C″) bskDN expression in a WT background has no apparent phenotype. (D-E″) bskDN expression partially rescues the scrib1 phenotype (two different examples are shown). Scale bar: 10 µm.
To determine whether, like in scrib epithelial clones, the JNK pathway is activated in scrib NBII clones and induces their apoptosis we stained them with an antibody to the activated form of JNK (diP-JNK). Compared with WT NBII clones, scrib NBII clones showed increased levels of diP-JNK (Fig. 2A-B″). Moreover, inhibiting the JNK pathway in scrib1 clones by expressing a dominant-negative form of Basket (BskDN), the Drosophila JNK homolog, increased the number of scrib1 NBII clones, which appeared in 71% of the brains analyzed (n=7); additionally, 44.4% of these clones (n=18 clones of six different brains) were larger than scrib1 clones and with lower levels of diP-JNK (Fig. 2D-E″). No effect was observed after overexpressing BskDN in otherwise WT clones (Fig. 2C-C″). Hence, all these data indicate that scrib NBII clones are eliminated by JNK pathway-mediated apoptosis.
cno scrib double-mutant NBII clones display tumor-like overgrowth
It is generally accepted that cancer is a multistep process, in which multiple genome alterations are required to render a carcinogenic phenotype (Hanahan and Weinberg, 2000, 2011). In Drosophila, the first demonstration of cooperative tumorigenesis came from studies in which mutations in tumor suppressor genes, with altered cell polarity, showed neoplastic tumor growth in combination with the constitutive activation of the oncogene Ras85D (Ras85DV12 or RasV12) (Brumby and Richardson, 2003; Pagliarini et al., 2003). Given that cooperation between two oncogenes or two tumor suppressor genes can also lead to hyperplasia or neoplasia (Briggs et al., 2008; Land et al., 1983; Liu et al., 2014), we considered whether the simultaneous loss of scrib and cno in clones could lead to a tumor-like overgrowth. Using the MARCM technique, we generated GFP-labeled cnoR2 scrib1 double-mutant clones in NBII lineages of late third instar larval brains. cnoR2 scrib1 clones showed a striking tumor-like overgrowth, with clone areas significantly larger than those of WT and cnoR2 or scrib1 single-mutant clones (Fig. 3A-E).
cno scrib NBII clones display tumor-like overgrowth. (A-E) Brain hemispheres showing a WT (A), cnoR2 (B), scrib1 (C) or cnoR2 scrib1 (D) clone, and quantification of the clone area (E; n, number of clones analyzed, each clone from different brains). (F-I′) WT NBII clones show a large Dpn+ cell, which is the NB (arrow), some smaller Dpn+ INPs and several Elav+ neurons within the clone (F,F′). cnoR2 and scrib1 clones show fewer progenitor cells (G-H′). cnoR2 scrib1 clones contain mainly Dpn+ progenitors and very few Elav+ neurons (I,I′). (J,K) Quantification of Dpn+ cells (n, number of clones of 19, 13, 9, 9 and 9 brains of the indicated genotypes). (L-Q) WT NBII clones show just one Dpn+ PntP1+ NB (arrow) and some PntP1+ immature INPs (L,L′). cnoR2 and scrib1 single clones show extra NBs (M-N′). cnoR2 scrib1 clone (confocal plane of the whole clone) contains multiple Dpn+ PntP1+ NBs (arrows in O,O′) and several PntP1+ immature INPs, as quantified in P,Q (n, number of clones of 10, 10, 8, 16, 7 and 9 brains of the indicated genotypes). (R-U′) Dpn+ Ase+ mature INPs in WT, cnoR2 and scrib1 NBII clones (arrows in R-T′). In cnoR2 scrib1 clones, fewer mature INPs are detected within the whole Ase+ cell population (U,U′), as quantified in V (n, number of clones of 9, 9, 11, 10 and 11 brains analyzed). cnoR2 scrib1(*), whole clone population; cnoR2 scrib1 (**), clones with massive overgrowth; cnoR2 scrib1(***), clones with fewer than 30 Dpn+ cells per clone. Note that cnoR2 scrib1 (**) clones show significant differences to cnoR2 and scrib1 single-mutant clones in all the analyses. Data (E,K,P,Q,V) were analyzed by Student's t-test. ***P<0.001, **P<0.01, *P<0.05; ns, not significant. Error bars indicate s.d. Scale bars: 10 µm, except 50 µm in A-D.
cno scrib NBII clones display tumor-like overgrowth. (A-E) Brain hemispheres showing a WT (A), cnoR2 (B), scrib1 (C) or cnoR2 scrib1 (D) clone, and quantification of the clone area (E; n, number of clones analyzed, each clone from different brains). (F-I′) WT NBII clones show a large Dpn+ cell, which is the NB (arrow), some smaller Dpn+ INPs and several Elav+ neurons within the clone (F,F′). cnoR2 and scrib1 clones show fewer progenitor cells (G-H′). cnoR2 scrib1 clones contain mainly Dpn+ progenitors and very few Elav+ neurons (I,I′). (J,K) Quantification of Dpn+ cells (n, number of clones of 19, 13, 9, 9 and 9 brains of the indicated genotypes). (L-Q) WT NBII clones show just one Dpn+ PntP1+ NB (arrow) and some PntP1+ immature INPs (L,L′). cnoR2 and scrib1 single clones show extra NBs (M-N′). cnoR2 scrib1 clone (confocal plane of the whole clone) contains multiple Dpn+ PntP1+ NBs (arrows in O,O′) and several PntP1+ immature INPs, as quantified in P,Q (n, number of clones of 10, 10, 8, 16, 7 and 9 brains of the indicated genotypes). (R-U′) Dpn+ Ase+ mature INPs in WT, cnoR2 and scrib1 NBII clones (arrows in R-T′). In cnoR2 scrib1 clones, fewer mature INPs are detected within the whole Ase+ cell population (U,U′), as quantified in V (n, number of clones of 9, 9, 11, 10 and 11 brains analyzed). cnoR2 scrib1(*), whole clone population; cnoR2 scrib1 (**), clones with massive overgrowth; cnoR2 scrib1(***), clones with fewer than 30 Dpn+ cells per clone. Note that cnoR2 scrib1 (**) clones show significant differences to cnoR2 and scrib1 single-mutant clones in all the analyses. Data (E,K,P,Q,V) were analyzed by Student's t-test. ***P<0.001, **P<0.01, *P<0.05; ns, not significant. Error bars indicate s.d. Scale bars: 10 µm, except 50 µm in A-D.
Compared with WT clones, the cell population of cnoR2 scrib1 clones appeared very homogeneous. To investigate the composition of these clones, we used antibodies against the transcription factors Dpn and Elav, which mark progenitor cells (NBs and INPs) and differentiated cells (neurons), respectively. WT clones never contained more than 30 (average of 21) progenitor (Dpn+) cells (n=23). Then, taking this number as a reference, we distinguished in cnoR2 scrib1 clones (n=37) two main groups: one (49%, n=18) that did not overgrow (i.e. contained fewer than 30 Dpn+ cells); and one (51%, n=19) that did overgrow (i.e. contained more than 30 Dpn+ cells). Within this latter group, some mutant clones (27%) showed massive overgrowth (more than 100 Dpn+ cells per clone) (Fig. 3F-K). Remarkably, independently of the number of Dpn+ cells, all cnoR2 scrib1 clones were composed almost exclusively of progenitor (Dpn+) cells, with very few differentiated (Elav+) cells.
Next, we analyzed in more detail whether the progenitor cells present in the double-mutant clones were NBs, INPs or a mix of both. We observed that, whereas in WT clones there was always only 1 NB (Dpn+ PntP1+) and an average of 4.7 immature INPs (iINPs; Dpn− PntP1+) (n=13), cnoR2 scrib1 clones showed significantly more NBs and iINPs, with a mean value of 7.5 NBs and 7.7 iINPs (n=18); this value increased to 12.4 NBs and 10.1 iINPs when considering the overgrowing population of double-mutant clones (n=7), and even in the non-overgrowing population the mean value was 4.3 NBs and 6 iINPs (n=11) (Fig. 3L-Q). To further characterize cnoR2 scrib1 clones, we tested the number of mature INPs (mINPs). Whereas in WT clones (n=14) mINPs (Dpn+ Ase+) represented 53% of the whole Ase+ cell population, in cnoR2 scrib1 clones (n=12) mINPs represented 44%, decreasing to 38% in the double-mutant clone population that showed massive overgrowth (n=8) (Fig. 3R-V). Given the greater number of NBs and iINPs in cnoR2 scrib1 clones compared with WT clones, this result (the decrease in the number of mINPs in double-mutant clones) might suggest that INPs do not mature properly and tend to revert to a more undifferentiated state. The numbers of each of these cell types (NBs, iINPs and mINPs) in cnoR2 or scrib1 single-mutant clones were significantly different from those in the overgrowing population of cnoR2 scrib1 clones (Fig. 3K,P,Q,V).
Thus, in summary, the loss of cno and of scrib synergistically interact in NBII lineages, leading to the formation of a tumor mass that is mainly composed of progenitor cells, with very few differentiated cells.
ACD is substantially altered in cno scrib double-mutant NBII clones
Like Cno, the tumor suppressor Scrib participates in regulating ACD (Albertson and Doe, 2003). Hence, we considered how the simultaneous loss of both modulators in cno scrib clones could be affecting this process. We analyzed the localization of the apical protein aPKC and the basal cell fate determinant Numb in progenitors of these mutant clones. The localization of aPKC and Numb failed in 85% and 0% in cnoR2 clones and in 42.8% and 40% in scrib1 clones, respectively (Fig. 4A-B′,D-E′,G; see also Fig. 1O-P′,Y). In cnoR2 scrib1 double-mutant clones, the asymmetric cortical distribution of both ACD regulators at metaphase/anaphase failed in 100% of the progenitors analyzed, with both regulators distributed throughout the cell cortex in most cells of mutant clones (Fig. 4C,C′,F-G). These defects suggest that ACD is severely affected in cnoR2 scrib1 clones and that this could account for the substantial increase in progenitor cells observed in these clones.
cno scrib NBII clones show severe defects in ACD and the upregulation of Ras activity. (A-F′) WT, scrib1 and cnoR2 scrib1 NBII clones stained for aPKC and PH3 (A-C′) or Numb and PH3 (D-F′). In WT clones, aPKC and Numb are asymmetrically enriched in dividing progenitors (arrows in A,A′,D,D′). scrib1 clones show defects in the localization of both aPKC and Numb (arrows in B,B′,E,E′); arrowhead points to a Numb crescent correctly formed in the NB, which is better seen in the inset (another focal plane of the same clone). In cnoR2 scrib1 NBII clones, both aPKC and Numb are always in extended crescents at the cortex of clone cells (arrows in C,C′,F,F′). (G) Quantification of aPKC and Numb localization defects in the specified genotypes. PCs, progenitor cells; NP, not present; CM, cortical mislocalization. (H-Z) Dpn labels progenitor cells and Elav labels differentiated cells (neurons). The tumor-like overgrowth shown by cnoR2 scrib1 NBII clones (J,O,W) is significantly suppressed by reducing Ras activity (K,L,P,Q) and by expressing a cno WT form (X,Z), but it is not altered by expressing a cno form lacking the Ras-associated (RA) domains (cnoΔN) (Y,Z). Eliminating or reducing Ras activity in WT or cnoR2 clones does not have any significant effect on progenitor cell number (H,I,L-N,Q-V). n, number of clones of 18, 8, 14 and 9 (L), 19, 6, 15 and 14 (Q), 4, 8, 8 and 5 (V) or 16, 9 and 11 (Z) brains. Data in L,Q,V,Z were analyzed by Student's t-test. ***P<0.001, **P<0.01; ns, not significant. Error bars indicate s.d. (AA-DD) WT NBII clones always contain one NB (AA arrow, BB). In UAS-RasV12 scrib2 NBII clones extra NBs are detected (CC arrows, DD). n, number of clones analyzed of 8 WT and 8 mutant clone brains. Scale bars: 10 µm.
cno scrib NBII clones show severe defects in ACD and the upregulation of Ras activity. (A-F′) WT, scrib1 and cnoR2 scrib1 NBII clones stained for aPKC and PH3 (A-C′) or Numb and PH3 (D-F′). In WT clones, aPKC and Numb are asymmetrically enriched in dividing progenitors (arrows in A,A′,D,D′). scrib1 clones show defects in the localization of both aPKC and Numb (arrows in B,B′,E,E′); arrowhead points to a Numb crescent correctly formed in the NB, which is better seen in the inset (another focal plane of the same clone). In cnoR2 scrib1 NBII clones, both aPKC and Numb are always in extended crescents at the cortex of clone cells (arrows in C,C′,F,F′). (G) Quantification of aPKC and Numb localization defects in the specified genotypes. PCs, progenitor cells; NP, not present; CM, cortical mislocalization. (H-Z) Dpn labels progenitor cells and Elav labels differentiated cells (neurons). The tumor-like overgrowth shown by cnoR2 scrib1 NBII clones (J,O,W) is significantly suppressed by reducing Ras activity (K,L,P,Q) and by expressing a cno WT form (X,Z), but it is not altered by expressing a cno form lacking the Ras-associated (RA) domains (cnoΔN) (Y,Z). Eliminating or reducing Ras activity in WT or cnoR2 clones does not have any significant effect on progenitor cell number (H,I,L-N,Q-V). n, number of clones of 18, 8, 14 and 9 (L), 19, 6, 15 and 14 (Q), 4, 8, 8 and 5 (V) or 16, 9 and 11 (Z) brains. Data in L,Q,V,Z were analyzed by Student's t-test. ***P<0.001, **P<0.01; ns, not significant. Error bars indicate s.d. (AA-DD) WT NBII clones always contain one NB (AA arrow, BB). In UAS-RasV12 scrib2 NBII clones extra NBs are detected (CC arrows, DD). n, number of clones analyzed of 8 WT and 8 mutant clone brains. Scale bars: 10 µm.
cno loss in cno scrib mutant NBII clones promotes Ras activation
Despite the strong ACD defects observed in cnoR2 scrib1 clones, it was still unclear how the JNK-mediated apoptosis observed in scrib clones was overcome by the absence of Cno. In epithelial tissues, constitutive activation of the oncogene Ras (RasV12) in scrib clones entails a change in the JNK pathway outcome from a proapoptotic to a progrowth effect. As a consequence, RasV12 scrib clones massively overgrow (Brumby and Richardson, 2003; Igaki et al., 2006). Intriguingly, Cno binds directly to Ras through its RA domains (Kuriyama et al., 1996), and we have previously shown that Cno represses the Ras signaling pathway during muscle and heart progenitor specification (Carmena et al., 2006). Hence, we considered whether Ras would be upregulated in cno scrib NBII clones, contributing in combination with scrib loss to the tumor-like overgrowth observed.
To test this hypothesis, we expressed a dominant-negative form of Ras (RasDN) in cnoR2 scrib1 clones. In this genetic background, the total number of progenitor (Dpn+) cells significantly decreased to 24.4 (n=14 clones), as compared with the 48.6 progenitor cells observed in cnoR2 scrib1 clones (n=20) (Fig. 4H-L). We also performed this experiment expressing RasRNAi to knock down Ras in the cnoR2 scrib1 mutant background with a similar result (Fig. 4M-Q). The loss of Ras in an otherwise WT NBII clone did not have any consequences for clone morphology, size or the number of progenitor cells (Fig. 4I,N,S,L,Q,V). Reducing Ras levels in cnoR2 clones did not significantly affect the number of progenitors, but the morphology and size of the clones/NBs were overall more similar to those of WT clones (Fig. 4T-V).
To further support an effect of cno loss on Ras upregulation in cnoR2 scrib1 clones, we expressed in this mutant background either a WT form of Cno (CnoWT) or a form that lacks the two RA domains (CnoΔN). Whereas the expression of CnoWT rescued the cnoR2 scrib1 overgrowth phenotype (mean of 17.7 Dpn+ cells per clone, as compared with 51.8 in cnoR2 scrib1 clones), no significant effect was observed after expressing CnoΔN (Fig. 4W-Z).
All these data strongly support that (1) cno loss in cnoR2 scrib1 clones leads to Ras activation, which suppresses scrib loss-induced cell death, and (2) that the severe defects in ACD observed in cnoR2 scrib1 clones without Ras activation are insufficient to cause tumor-like overgrowth in cnoR2 scrib1 clones.
RasV12 scrib mutant NBII clones survive but do not overgrow
As mentioned above, RasV12 scrib clones promote overgrowth in epithelial tissues (Brumby and Richardson, 2003; Pagliarini et al., 2003). Given that in NBII lineages cnoR2 loss induced Ras activation and tumor-like overgrowth in a scrib1 background, we tested the effect of directly expressing RasV12 on scrib NBII clones in larval brains. Whereas scrib clones were eliminated by apoptosis and were rarely detected (see above), RasV12 scrib mutant clones survived, appearing at a similar rate to WT clones. Whereas in WT clones there was always only 1 NB (Dpn+ Ase−) per clone (n=12), in RasV12 scrib clones 2 or 3 NBs were found in 18.2% or 36.3%, respectively, of the clones analyzed (n=11) (Fig. 4AA-DD). Despite these defects in the number of NBs, and contrary to what occurs in epithelial tissues, RasV12 scrib NBII clones did not show overgrowth.
Altogether, from these and the previous results we conclude that the tumor-like overgrowth of cnoR2 scrib1 NBII clones does not rely only on cno loss-mediated Ras derepression or on the severe ACD defects observed, but on both of them (see Discussion).
cno loss in cno scrib mutant NBII clones activates the Ras-PI3K-Akt pathway
In epithelia, RasV12 signals through the Raf-MAPK pathway to promote overgrowth (Brumby and Richardson, 2003). Thus, to confirm the effect of cno loss on Ras activation, we assessed the levels of diP-MAPK (a Ras-Raf-MAPK pathway activity readout) in cnoR2 scrib1 NBII clones. No diP-MAPK was detected in these or control clones, and it was at low levels or undetectable in RasV12 NBII clones (Fig. S1). These results indicate that the Ras-Raf-MAPK pathway is not normally active in WT NB clones and that the ectopic activation of Ras in cnoR2 scrib1 NBII clones is barely able to induce this pathway. In fact, the expression of RafRNAi in a cnoR2 scrib1 mutant background had no significant effect on reducing the cnoR2 scrib1 overgrowth phenotype, as apparent from analyzing the number of progenitor (Dpn+) cells (P=0.12) (Fig. 5A-C).
cno loss in cno scrib NBII clones activates the Ras-PI3K-Akt pathway. (A-G) The tumor-like overgrowth shown by cnoR2 scrib1 NBII clones is not altered by reducing the Ras effector Raf (A-C; n, number of clones of 15 and 10 brains analyzed), but it is significantly suppressed by downregulating the Ras effector Akt (E-G; n, number of clones of 10 and 9 brains). Data in C,G were analyzed by Student's t-test. ***P<0.001; ns, not significant. Error bars indicate s.d. (D) In Drosophila, Ras normally triggers the Raf-MAPK pathway but can also activate the PI3K-Akt signaling pathway in particular conditions (Prober and Edgar, 2002). (H) Western blot of larval brain lysates with NBII clones of the indicated genotype showing pAkt levels. Two isoforms of Akt are detected by the antibody (66 and 85 kDa); the 66 kDa isoform is activated in cnoR2 scrib1 clones (arrow). Scale bar: 10 µm.
cno loss in cno scrib NBII clones activates the Ras-PI3K-Akt pathway. (A-G) The tumor-like overgrowth shown by cnoR2 scrib1 NBII clones is not altered by reducing the Ras effector Raf (A-C; n, number of clones of 15 and 10 brains analyzed), but it is significantly suppressed by downregulating the Ras effector Akt (E-G; n, number of clones of 10 and 9 brains). Data in C,G were analyzed by Student's t-test. ***P<0.001; ns, not significant. Error bars indicate s.d. (D) In Drosophila, Ras normally triggers the Raf-MAPK pathway but can also activate the PI3K-Akt signaling pathway in particular conditions (Prober and Edgar, 2002). (H) Western blot of larval brain lysates with NBII clones of the indicated genotype showing pAkt levels. Two isoforms of Akt are detected by the antibody (66 and 85 kDa); the 66 kDa isoform is activated in cnoR2 scrib1 clones (arrow). Scale bar: 10 µm.
Apart from the Raf-MAPK pathway, other effectors of Ras have been described, such as the phosphatidylinositol-3 kinase (PI3K)-Akt pathway (Fig. 5D). Hence, we examined whether this pathway was exerting the effects of ectopic Ras activation in cnoR2 scrib1 NBII clones. The expression of AktRNAi in that mutant background dramatically rescued the overgrowth phenotype of cnoR2 scrib1 clones, with an average of 20 Dpn+ cells per clone as compared with 51.8 in the double mutant (Fig. 5E-G). The loss of Akt in an otherwise WT NBII clone did not have any effect on clone size, morphology or progenitor cell number (Fig. S2). We also directly measured Akt signaling activation in cnoR2 scrib1 clones by analyzing phospho-Akt (pAkt) levels. Western blotting of WT brains versus those with cnoR2 scrib1 clones showed higher levels of the pAkt 66 kDa isoform in the latter (Fig. 5H). Hence, the ectopic Ras induced by cno loss in cnoR2 scrib1 NBII clones leads to PI3K-Akt pathway activation (see Discussion).
cno scrib mutant clones in epithelia overgrow and activate the Ras-Raf-MAPK pathway
To determine a potential context-dependent effect of cno loss on Ras activation and tumor induction in a scrib mutant background, we analyzed cno scrib clones in epithelial tissues, which grow by symmetric cell divisions. cnoR2 scrib1 clones induced in antennal discs showed a clear tumor overgrowth in 28% of the clones analyzed (n=14), whereas no overgrowth was observed in WT (n=12) or cnoR2 (n=10) clones, and scrib1 clones overgrew in 6.7% of cases (n=15). RasV12 scrib clones showed tumor overgrowth in 100% of the clones analyzed (n=20) (Fig. 6A-L). Hence, the cnoR2 scrib1 phenotype was similar, although with less penetrance and expressivity, to that described for RasV12 scrib in epithelial tissues.
cno scrib clones in epithelial tissues overgrow and activate the Ras-MAPK pathway. (A-F′) Confocal sections of eye-antennal discs showing GFP-labeled antennal disc clones of the specified genotypes stained for L(2)gl and diP-MAPK (A-F). e.d., eye disc; a.d., antennal disc. Whereas WT (A), cnoR2 (B) or scrib1 (C) clones do not overgrow, cnoR2 scrib1 clones in antennal discs show overgrowth (arrow in D); RasV12 (E) and RasV12 scrib1 (F) clones also show overgrowth, the latter to a greater extent than cnoR2 scrib1 clones. diP-MAPK is not detected in WT (A′), cnoR2 (B′) or scrib1 (C′) antennal disc clones but it is present in cnoR2 scrib1 clones (arrows in D′), in RasV12 (E′) and to a higher extent in RasV12 scrib1 (F′). (G-L) Antennal disc sagittal views showing the altered rounded cell morphology in mutant clones (arrow in J), compared with the columnar cell morphology in WT clones (arrow in G). (M) Western blot of eye-antennal imaginal disc lysates of the indicated genotypes showing diP-MAPK and MAPK levels. (N) Fold increase of MAPK activation in the indicated genotypes relative to WT. Error bars indicate s.d. (n=3 independent experiments). Scale bars: 50 µm (A-F′); 10 µm (G-L).
cno scrib clones in epithelial tissues overgrow and activate the Ras-MAPK pathway. (A-F′) Confocal sections of eye-antennal discs showing GFP-labeled antennal disc clones of the specified genotypes stained for L(2)gl and diP-MAPK (A-F). e.d., eye disc; a.d., antennal disc. Whereas WT (A), cnoR2 (B) or scrib1 (C) clones do not overgrow, cnoR2 scrib1 clones in antennal discs show overgrowth (arrow in D); RasV12 (E) and RasV12 scrib1 (F) clones also show overgrowth, the latter to a greater extent than cnoR2 scrib1 clones. diP-MAPK is not detected in WT (A′), cnoR2 (B′) or scrib1 (C′) antennal disc clones but it is present in cnoR2 scrib1 clones (arrows in D′), in RasV12 (E′) and to a higher extent in RasV12 scrib1 (F′). (G-L) Antennal disc sagittal views showing the altered rounded cell morphology in mutant clones (arrow in J), compared with the columnar cell morphology in WT clones (arrow in G). (M) Western blot of eye-antennal imaginal disc lysates of the indicated genotypes showing diP-MAPK and MAPK levels. (N) Fold increase of MAPK activation in the indicated genotypes relative to WT. Error bars indicate s.d. (n=3 independent experiments). Scale bars: 50 µm (A-F′); 10 µm (G-L).
As mentioned above, in epithelia RasV12 signals through the Raf-MAPK pathway to promote overgrowth (Brumby and Richardson, 2003). Thus, we tested whether cno loss induces the Ras-MAPK pathway in this epithelial context. By immunofluorescence, neither cnoR2 nor scrib1 single-mutant clones showed any apparent diP-MAPK upregulation compared with WT clones (Fig. 6A′-C′). However, both cnoR2 scrib1 and RasV12 scrib antennal clones exhibited diP-MAPK upregulation, the former to a lesser extent than the latter (Fig. 6D′,F′). Hence, in epithelial tissues the cnoR2 scrib1 and RasV12 scrib mutant clone overgrowth phenotypes are more similar than in NBs (see Discussion). Intriguingly, quantification of diP-MAPK levels in cnoR2 or scrib1 single-mutant disc extracts showed, on average, about a fivefold or sevenfold increase, respectively, in MAPK activation relative to WT discs (Fig. 6M,N). This result strongly supports a repressive effect of both Cno and Scrib on Ras-MAPK signaling pathway in epithelia (see also Dow et al., 2008).
Dlg1 and L(2)gl repress Ras-MAPK signaling in epithelia without synergistically interacting with Cno
Scrib functions along with the tumor suppressors Dlg1 and L(2)gl in a common pathway to regulate cell polarity and growth in epithelial tissues (Bilder et al., 2000; Elsum et al., 2012). Moreover, not only scrib but also dlg1 and l(2)gl cooperate with RasV12 to promote neoplastic overgrowth (Brumby and Richardson, 2003; Pagliarini et al., 2003). Thus, we considered whether cno loss would also synergistically interact with dlg1 and l(2)gl mutants in imaginal discs to induce tumor formation. dlg1RNAi; cnoR2 and l(2)glRNAi; cnoR2 double-mutant clones in antennal discs showed consistent overgrowth and upregulation of diP-MAPK in 100% of the clones analyzed (Fig. 7A,A′,E-F′). The localization of Scrib was also disturbed (Fig. 7G′,K′-L′). Intriguingly, dlg1RNAi and l(2)glRNAi single-mutant clones already displayed this phenotype (Fig. 7C-D′,I-J′). Hence, the loss of cno in both dlg1RNAi and l(2)glRNAi genetic backgrounds seemed to slightly exacerbate the phenotype of each single-mutant condition, especially in terms of altered cellular morphology and tissue disorganization (Fig. 7G-L), although the effect did not seem to be synergistic, as it was clearly observed in cnoR2 scrib1 antennal disc clones (see Discussion).
Dlg1 and L(2)gl repress Ras-MAPK signaling in epithelia. (A-F′) Confocal sections of eye-antennal discs showing GFP-labeled clones of the indicated genotypes stained for Scrib and diP-MAPK (A-F). e.d., eye disc; a.d., antennal disc. dlg1 and l(2)gl clones show overgrowth (arrows in C,D) and an upregulation of diP-MAPK (arrows in C′,D′), a phenotype that it is not exacerbated in dlg1, cnoR2 or l(2)gl cnoR2 double-mutant clones (E-F′). (G-L) Antennal disc sagittal views showing the altered rounded cell morphology in mutant clones. (G′-L′) Defects in Scrib expression and localization are found in dlg1 and l(2)gl clones (arrows and arrowheads in I′,J′) compared with WT clones (G′). Green dotted lines delineate the clones. (M-R) In WT NBII clones, only one large NB (Dpn+ Ase−) is detected (arrow in M). In dlg1 clones, extra NBs are occasionally detected (arrows in N; the inset shows a deeper confocal plane of focus). In l(2)gl clones, several extra NBs are present (arrows in O) and clones are smaller than WT clones. (P-R) dlg1RNAi cnoR2 (Q) and l(2)glRNAi cnoR2 (R) clones present a small NB and sometimes extra NBs (arrows), like cno single-mutant clones (P), as quantified in S (n, number of clones of 13, 19, 15, 22 and 26 brains analyzed of the indicated genotypes). Data were analyzed by Student's t-test. *P<0.05; ns, not significant. Error bars indicate s.d. Scale bars: 50 µm (A-F′); 10 µm (G-R). (T) Diagrams depicting distinct protein interaction networks among Cno, the Scrib module [Scrib, Dlg1 and L(2)gl] and the Ras pathway in different developmental contexts (NBIIs versus epithelial cells) and in distinct genetic backgrounds (WT versus cno scrib double mutants). Arrows indicate epistatic relationships. In WT NBIIs, Cno directly represses Ras by physically interacting with Ras-GTP. While Cno is epistatic to Dlg1 and L(2)gl, Cno and Scrib act in parallel or partially convergent pathways. In WT epithelial cells, Cno, Scrib, Dlg1 and L(2)gl cooperate to repress the Ras-MAPK pathway: Cno and Scrib repress it by directly binding Ras-GTP and MAPK, respectively; Dlg1 and L(2)gl repress it indirectly by contributing to Scrib localization and by other (direct or indirect) unknown mechanism(s) (dashed line). In cno scrib double-mutant NBIIs, both the upregulation of the Ras-PI3K-Akt pathway and the strong defects in ACD lead to a tumor-like overgrowth; in cno scrib mutant epithelia, the upregulation of the Ras-Raf-MAPK pathway promotes the tumor-like overgrowth.
Dlg1 and L(2)gl repress Ras-MAPK signaling in epithelia. (A-F′) Confocal sections of eye-antennal discs showing GFP-labeled clones of the indicated genotypes stained for Scrib and diP-MAPK (A-F). e.d., eye disc; a.d., antennal disc. dlg1 and l(2)gl clones show overgrowth (arrows in C,D) and an upregulation of diP-MAPK (arrows in C′,D′), a phenotype that it is not exacerbated in dlg1, cnoR2 or l(2)gl cnoR2 double-mutant clones (E-F′). (G-L) Antennal disc sagittal views showing the altered rounded cell morphology in mutant clones. (G′-L′) Defects in Scrib expression and localization are found in dlg1 and l(2)gl clones (arrows and arrowheads in I′,J′) compared with WT clones (G′). Green dotted lines delineate the clones. (M-R) In WT NBII clones, only one large NB (Dpn+ Ase−) is detected (arrow in M). In dlg1 clones, extra NBs are occasionally detected (arrows in N; the inset shows a deeper confocal plane of focus). In l(2)gl clones, several extra NBs are present (arrows in O) and clones are smaller than WT clones. (P-R) dlg1RNAi cnoR2 (Q) and l(2)glRNAi cnoR2 (R) clones present a small NB and sometimes extra NBs (arrows), like cno single-mutant clones (P), as quantified in S (n, number of clones of 13, 19, 15, 22 and 26 brains analyzed of the indicated genotypes). Data were analyzed by Student's t-test. *P<0.05; ns, not significant. Error bars indicate s.d. Scale bars: 50 µm (A-F′); 10 µm (G-R). (T) Diagrams depicting distinct protein interaction networks among Cno, the Scrib module [Scrib, Dlg1 and L(2)gl] and the Ras pathway in different developmental contexts (NBIIs versus epithelial cells) and in distinct genetic backgrounds (WT versus cno scrib double mutants). Arrows indicate epistatic relationships. In WT NBIIs, Cno directly represses Ras by physically interacting with Ras-GTP. While Cno is epistatic to Dlg1 and L(2)gl, Cno and Scrib act in parallel or partially convergent pathways. In WT epithelial cells, Cno, Scrib, Dlg1 and L(2)gl cooperate to repress the Ras-MAPK pathway: Cno and Scrib repress it by directly binding Ras-GTP and MAPK, respectively; Dlg1 and L(2)gl repress it indirectly by contributing to Scrib localization and by other (direct or indirect) unknown mechanism(s) (dashed line). In cno scrib double-mutant NBIIs, both the upregulation of the Ras-PI3K-Akt pathway and the strong defects in ACD lead to a tumor-like overgrowth; in cno scrib mutant epithelia, the upregulation of the Ras-Raf-MAPK pathway promotes the tumor-like overgrowth.
Finally, we wanted to determine the behavior of dlg1RNAi; cnoR2 and l(2)glRNAi; cnoR2 double-mutant NBII clones. Both dlg1RNAi and l(2)glRNAi single-mutant NBII clones had extra NBs compared with control NBII clones, but they did not overgrow (Fig. 7M-O,S). In fact, in the case of l(2)glRNAi clones, and despite the extra NBs (up to 7), the clones appeared smaller than control clones. Intriguingly, both dlg1RNAi; cnoR2 and l(2)glRNAi; cnoR2 double-mutant NBII clones displayed a phenotype that was more similar to that of cnoR2 single-mutant clones (i.e. 100% of the clones showed an undersized NB, which is a very characteristic feature of cnoR2 NBII clones) with no apparent overgrowth (Fig. 7P-R). We tested the levels of Dlg1 and L(2)gl proteins in NBII clones after overexpressing their RNAi, and observed strong reduction or elimination (Fig. S3). Hence, as in epithelia, in NBII lineages cno did not synergistically interact with dlg1 or l(2)gl to promote tumor-like overgrowth, this synergistic interaction being specifically and exclusively established only between cno and scrib loss in both systems (see Discussion).
DISCUSSION
In this work we have uncovered a synergistic interaction between cno and scrib mutants that leads to tumor-like overgrowth via Ras activation in both NBII lineages and epithelia. In NBII lineages, cno scrib mutant clones showed a much stronger phenotype than RasV12 scrib clones. In the latter, the overexpression of Ras was able to overcome the JNK-mediated cell death in scrib NBII clones; however, RasV12 scrib clones, despite displaying extra NBs, did not overgrow. This would suggest that the loss of cno in cno scrib clones has additional consequences other than activating Ras. Given that both cno and scrib modulate ACD, it is very plausible that the simultaneous loss of both regulators in NBII lineages contributes to exacerbating the phenotype by severely compromising this process. However, downregulating Ras signaling in cno scrib mutant clones suppressed their tumor-like overgrowth, strongly supporting the contention that both the upregulation of Ras activity and the accumulated ACD defects in cno scrib clones contribute to the tumor-like overgrowth phenotype. In epithelia, in antennal imaginal discs, both cno scrib and RasV12 scrib clones showed upregulation of diP-MAPK and a concomitant overgrowth, although cno scrib clones to a lesser extent than RasV12 scrib clones. The lower penetrance of the cno scrib phenotype is probably due to the much higher thresholds of Ras activity provided by RasV12 (see also below). No apparent diP-MAPK upregulation or overgrowth was detected in cno or scrib single-mutant clones compared with control clones, indicating a synergism between these mutants also in imaginal discs.
Although the Raf–diP-MAPK signaling pathway was upregulated in response to Ras activation in imaginal discs, as previously shown (Brumby and Richardson, 2003), we were unable to detect diP-MAPK in NBII lineages in control or cno scrib clones, and it was at low levels or undetectable in RasV12 clones. However, we found by different approaches that Ras is activated in cno scrib NBII clones, suggesting that other Ras effectors might be acting in those clones. Whereas in mammals Ras (K-, H- or N-Ras) can signal through different effectors, including Raf, PI3K and RALGDS, in Drosophila Ras acts mainly through the Raf-MAPK signaling cascade (Neuman-Silberberg et al., 1984). Interestingly, even though Ras does not normally act through PI3K in Drosophila, in wing imaginal discs RasV12 is able to activate this pathway, which in turn leads to the activation of its downstream effector Akt1 (Prober and Edgar, 2002; Willecke et al., 2011). We have found that this scenario also occurs in cno scrib NBII clones, where the Ras-PI3K-Akt1 signaling pathway was aberrantly activated, and the overgrowth phenotype shown by these clones was suppressed by compromising this pathway.
Scrib functions along with Dlg1 and L(2)gl, forming part of what has been called the Scrib polarity module (Elsum et al., 2012). In fact, not only scrib but also dlg1 and l(2)gl mutants cooperate with oncogenic Ras to induce tumor growth and metastasis in Drosophila imaginal discs (Brumby and Richardson, 2003; Igaki et al., 2006; Pagliarini et al., 2003). Hence, we examined whether cno loss would also synergistically interact with dlg1 and l(2)gl to induce overgrowth. dlg1RNAi cno and l(2)glRNAi cno double-mutant clones in antennal discs showed a marked overgrowth along with a clear upregulation of diP-MAPK. However, this phenotype was very similar to that shown by dlg1RNAi and l(2)glRNAi single-mutant clones. This is intriguing because it suggests that both Dlg1 and L(2)gl are, directly or indirectly, impinging on the Ras-MAPK signaling pathway, promoting its repression. RAS is aberrantly activated in most human cancers. Hence, it is of great relevance to understand how it is normally regulated to avoid its oncogenic activity. To our knowledge, only Scrib, from the Scrib polarity module, has been described as inhibiting the Ras-MAPK pathway both in Drosophila and in mammalian cells by directly interacting with MAPK (Dow et al., 2008; Nagasaka et al., 2010). In zebrafish, lgl(2) (llgl2) loss has also been associated with an Erb-Ras signal upregulation (Reischauer et al., 2009). Altogether, we propose the existence of a coordinate and cooperative action in epithelial tissues between Cno and the Scrib/Dlg1/L(2)gl module to avoid aberrant Ras-MAPK signaling activation in normal conditions (Fig. 7T). The strongest effect on Ras-MAPK derepression found in both dlg1 and l(2)gl single-mutant clones, as compared with cno and scrib single-mutant clones, might be due to the defects in Scrib localization or levels observed in the former mutants (Fig. 7G′,I′-J′). Given the high evolutionary conservation of these proteins, the cooperation between Cno and the Scrib module might also be functioning in humans (see below).
Like Scrib, Dlg1 and L(2)gl are also involved in regulating ACD in NBs. Thus, looking for potential synergisms between cno and dlg1 or l(2)gl mutants in NBII lineages, we found that dlg1RNAi cno and l(2)glRNAi cno double-mutant clones did not overgrow. In fact, they displayed defects characteristic of the cno single-mutant phenotype, such as the undersized NB. This indicates that cno is epistatic to both dlg1 and l(2)gl during ACD in NBII lineages, which might explain the lack of synergisms between them. Conversely, Cno and Scrib must function in parallel or in at least partially independent pathways during ACD in order to explain the strong synergism seen when both genes are compromised (Fig. 7T). Despite the extra NBs present in all cno, l(2)gl, dlg1 and scrib NBII clones in the larval brain, the loss of each of these ACD regulators did not lead to a tumor-like overgrowth. On the contrary, these clones tended to be smaller and, in the case of scrib clones, they died by apoptosis. These observations might appear counterintuitive as l(2)gl, dlg1 and scrib were originally identified as tumor suppressor genes due to the massive overgrowth they produced when they were compromised in the larval brain (Bilder et al., 2000; Gateff, 1994). It is important to remark, though, that these studies were not done by clonal NB analysis but in the homozygous mutant condition, and individual NB lineages were not analyzed in detail (NBII lineages had not even been described at that time). Contrary to what happens in homozygous mutant brains, NB mutant clones are surrounded by WT cells, which may counteract or even eliminate the clonal mutant cells, a phenomenon known as cell competition (Morata and Ballesteros-Arias, 2015). In the context of cancer biology, where tumors start with mutation(s) in one or a small group of cells, clonal analysis for studying the tumorigenic potential of particular gene mutations rather than analysis of the whole mutant tissue is a more disease-relevant scenario. In addition, an accepted hallmark in cancer biology is the acquisition of several mutational hits, not just single mutations, for tumor development (Hanahan and Weinberg, 2000, 2011), and our results support this. The differences in growth shown by cno scrib double-mutant clones (i.e. no overgrowth versus overgrowth, even massive overgrowth) might be due to the position of the mutant clone and differences in the environment or signals that could act to counteract or restrain mutant clone expansion.
The human afadin (AFDN, MLLT4 or AF6 – for ALL1 fusion product on chromosome 6) gene was first isolated as a fusion partner of the mixed lineage leukemia gene (MLL, ALL1 or KMT2A) that is frequently detected in acute lymphoblastic leukemias (Prasad et al., 1993). Recent work has established a direct connection between low levels of afadin and tumor growth in different cancer types (Fournier et al., 2011; Xu et al., 2015). Intriguingly, both Cno and afadin bind the activated form of Ras (Ras-GTP), RAS being one of the most commonly mutated oncogenes in human cancers, and we showed some years ago that Cno represses Ras during the specification of Drosophila muscle and heart progenitors (Carmena et al., 2006; Kuriyama et al., 1996). This inhibitory effect of Cno on Ras has been shown to be conserved in mammals (Fournier et al., 2011; Radziwill et al., 2003). Moreover, upregulation of Ras-GTP has recently been associated with afadin knockdown or to its translocation from the cytoplasm to the nucleus in distinct cancer types (Fournier et al., 2011; Manara et al., 2015). This is highly relevant because even though mutations in RAS alone do not normally progress in malignant tumors, they provide a sensitized genetic background for further mutational events, which along with mutated RAS can lead to neoplasias. This is the case with RasV12 mutation, which causes aberrant Ras activation, and scrib tumor suppressor gene mutation, as mentioned above. In humans, SCRIB subcellular localization is altered or its levels are low in several cancer types, including breast cancer in which afadin is also downregulated with concomitant activation of the RAS-MAPK signaling pathway (Feigin et al., 2014; Fournier et al., 2011; Gardiol et al., 2006; Navarro et al., 2005; Zhan et al., 2008). Hence, the synergistic cooperation found in Drosophila between Cno and Scrib might be conserved between the human orthologs afadin and SCRIB.
MATERIALS AND METHODS
Drosophila strains and genetics
All stocks used were from the Bloomington Drosophila Stock Center and the Vienna Drosophila Resource Center, unless otherwise noted (see the supplementary Materials and Methods for a detailed description of strains). The crosses GAL4×UAS were carried out at 29°C. Balancer chromosomes bearing the Tubby dominant marker were used to identify third instar larva mutant chromosomes.
Generation of GFP-tagged endogenous cno gene lines
Intronic MIMIC insertions MI00782 (phase 1) and MI04707 (phase 0), containing two inverted φC31 attP target sites flanking a gene trap cassette, in the cno locus were replaced by pBS-KS-attB1-2-PT-SA-SD-1 or 0 (phase)-EGFP-FlAsH-StrepII-TEV-3xFlag using recombination-mediated cassette exchange (RMCE) (Venken et al., 2011).
MARCM clones
To generate mutant clones in the brain as well as in imaginal discs, w; Dll-Gal4 UAS-CD8::GFP; FRT82B tubGal80 males or hsFLP; Dll-Gal4 UAS-CD8::GFP; FRT82B tubGal80 females were crossed with females or males, respectively, of different genotypes (see the supplementary Materials and Methods for a detailed description of the crosses performed). The clones were identified by the presence of CD8::GFP. hsFLP was induced for 2 h at 37°C in late first/early second instar larvae and clones were analyzed in late third instar larvae.
Histology and immunofluorescence
Larval brains and discs were dissected and fixed in 3.7% formaldehyde in PBS for 40 min. Stainings were carried out using standard protocols. The following primary antibodies were used: mouse anti-Dlg1 1:100 (Developmental Studies Hybridoma Bank, DSHB), rabbit anti-Cnn 1:400 (a gift from T. C. Kaufman, Biology Indiana University, Bloomington, IN, USA), rabbit anti-PH3 1:400 (Millipore, 06-570), guinea pig anti-Dpn 1:200 (this work), rabbit anti-PntP1 1:500 (a gift from J. Skeath, Washington University School of Medicine, St Louis, MO, USA), rabbit anti-aPKCζ 1:100 (C-2 Santa Cruz, sc-216), rabbit anti-Baz 1:1000 (a gift from A. Wodarz, University of Cologne, CECAD Research Center, Cologne, Germany), rabbit anti-Pins 1:200 (a gift from J. Knoblich, Institute of Molecular Biotechnology, Vienna, Austria), guinea pig anti-Numb 1:400 [a gift from Y. N. Jan (Rhyu et al., 1994)], goat anti-Numb 1:200 (Santa Cruz, sc-23579), rabbit anti-Mira 1:1000 (Ikeshima-Kataoka et al., 1997), rabbit anti-Elav 1:400 (DSHB), rabbit anti-Ase 1:80 (this work), rabbit anti-L(2)gl 1:100 (Santa Cruz, sc-98260), mouse anti-diP-MAPK 1:500 (Sigma, M8159), mouse anti-GFP 1:100 (Roche, 11814460001) and rabbit anti-P-SAPK/JNK 1:200 (Cell Signaling, 4668). Secondary antibodies coupled to Alexa Fluor 488, 546 or 633 (Invitrogen) were used for immunofluorescence.
Generation of Dpn and Ase antibodies
An Ase peptide (CLSDESMIDAIDWWEAHAPKSNGACTNLSV) was used to immunize two New Zealand rabbits (Abyntek). Dpn was cloned in an expression vector after being PCR amplified from genomic DNA (a gift from J. Skeath). Dpn protein was expressed in E. coli and purified to immunize two guinea pigs (Abyntek).
Image acquisition, processing and data analysis
Fluorescence images were recorded using a Leica DM-SL upright microscope with spectral confocal acquisition software. All images were taken with an HCX Plan Apochromat 63×/1.32 or 40×/1.25 oil CS objective (Zeiss). Images were analyzed using ImageJ (Schindelin et al., 2015) and FIJI (Schindelin et al., 2012) and assembled from raw data using Adobe Photoshop. Statistical analyses were carried out with SPSS software (IBM) using Student's t-test. Graphical representation of t-tests employed simple bars or dispersion graphics.
Western blots
Six eye-antennal imaginal discs or ten brains from third instar larvae were dissected in PBS and homogenized for 30 min in lysis buffer [50 mM Tris pH 8.0, 150 mM NaCl, 0.1% SDS, 1 M EDTA, 1% Triton X-100, 1% sodium deoxycholate, protease inhibitors, 1 mM NaF, 100 mM Na3VO4, 2 mM PMSF and Complete Protease Inhibitor Cocktail (Roche) freshly added prior to use] on ice. Protein extracts were centrifuged at 5000 g at 4°C for 10 min. 5×Laemmli buffer with 0.1 M DTT was added to each sample, heated at 95°C for 10 min and centrifuged at maximum speed (8000 g) for 2 min before being resolved by SDS-PAGE. From the whole-protein extract, material corresponding to one disc or five brains was loaded in each experiment. Spectra Multicolor High Range Protein Ladder (Fermentas) was used as a molecular weight marker. PVDF filters were immunoblotted with 1:1000 mouse anti-diP-MAPK/ERK1/2 (Sigma, M8159), 1:20,000 rabbit anti-MAPK (Sigma, M5670), 1:500 rabbit anti-pAkt (Ser473; Cell Signaling, 9271), 1:200 rabbit anti-pan Akt (Cell Signaling, 4691) or 1:5000 mouse anti-α-Tubulin (Sigma, T6199). Each experiment was repeated at least three times. Images were analyzed and quantified as above.
Acknowledgements
We thank G. Halder and H. Richardson, A. Wodarz, Y. N. Jan, J. Skeath, T. C. Kaufman, J. Knoblich, the Bloomington Drosophila Stock Center at the University of Indiana, the Vienna Drosophila Resource Center and the Developmental Studies Hybridoma Bank at the University of Iowa for kindly providing fly strains and reagents. We are grateful to S. Speicher for technical assistance.
Author contributions
Conceptualization: A.C.; Validation: N.R.-Q.; Formal analysis: N.R.-Q., M.F., A.d.T.-J.; Investigation: N.R.-Q., M.F., A.d.T.-J.; Writing - original draft: A.C.; Writing - review & editing: A.C.; Visualization: N.R.-Q., A.C.; Supervision: A.C.; Project administration: A.C.; Funding acquisition: A.C.
Funding
N.R.-Q. was supported by a VALi+d contract from the Generalitat Valenciana; our laboratory was supported by Spanish Government (Ministerio de Economía y Competitividad, MINECO) grants BFU2012-33020, BFU2015-64251-P and by FEDER (European Regional Development Fund). The Instituto de Neurociencias in Alicante is a Center of Excellence Severo Ochoa.
References
Competing interests
The authors declare no competing or financial interests.