Summary
Notch signalling is involved in numerous cellular processes during development and throughout adult life. Although ligands and receptors are largely expressed in the whole organism, activation of Notch receptors only takes place in a subset of cells and/or tissues and is accurately regulated in time and space. Previous studies have demonstrated that endocytosis and recycling of both ligands and/or receptors are essential for this regulation. However, the precise endocytic routes, compartments and regulators involved in the spatiotemporal regulation are largely unknown. In order to identify intracellular trafficking regulators of Notch signalling, we have undertaken a tissue-specific dsRNA genetic screen of candidates potentially involved in endocytosis and recycling within the endolysosomal pathway. dsRNA against 418 genes was induced in the Drosophila melanogaster sensory organ lineage in which Notch signalling regulates binary cell fate acquisition. Gain or loss of Notch signalling phenotypes were observed in adult sensory organs for 113 of them. Furthermore, 26 genes were found to regulate the steady state localisation of Notch, Sanpodo, a Notch co-factor, and/or Delta in the pupal lineage. In particular, we identified 20 genes with previously unknown function in D. melanogaster intracellular trafficking. Among them, we identified CG2747 and we show that it regulates the localisation of clathrin adaptor AP-1 complex, a negative regulator of Notch signalling. Together, our results further demonstrate the essential function of intracellular trafficking in regulating Notch-signalling-dependent binary cell fate acquisition and constitute an additional step toward the elucidation of the routes followed by Notch receptor and ligands during signalling.
Introduction
Notch cell-cell signalling is required in a vast majority of developmental processes and during the adult life of many organisms. It regulates cell fate specification as well as stem cell behaviour and defects can lead to numerous developmental pathologies and cancers underlying its crucial role (reviewed by Gridley, 2003; Miele et al., 2006). The challenging question is to understand the mechanisms allowing one cell to act as a signalling cell and the other one as the receiving cell, when both cells can potentially express both ligands and receptors. Although it can be performed through a spatial and temporal regulation of their expression, DSL (Delta, Serrate, Lag2) ligand and Notch receptor differential expression could not be sufficient to explain the subtle directionality of Notch signalling. In this context, regulation of the availability of both receptors and DSL ligands at the cell surface appears crucial to ensure a proper Notch signalling activation. Therefore ligand and receptor post-translational modifications and trafficking are emerging as crucial regulatory mechanisms.
Several lines of evidence suggest that endocytic trafficking of DSL ligands enhances their signalling activity while receptor trafficking insures their steady state level at the cell surface thereby regulating their availability for ligand binding (reviewed by Bray, 2006; Fürthauer and González-Gaitán, 2009; Kopan and Ilagan, 2009; Le Borgne, 2006; Weinmaster and Fischer, 2011; Yamamoto et al., 2010). Although recycling of DSL ligands is necessary to produce an active DSL ligand, the nature of this maturation is still poorly characterised and two models are actually favoured: endocytosis and pulling forces (Klueg and Muskavitch, 1999; Nichols et al., 2007; Windler and Bilder, 2010) versus endocytosis and recycling (Benhra et al., 2010; Emery et al., 2005; Jafar-Nejad et al., 2005; Le Borgne and Schweisguth, 2003; Rajan et al., 2009; Wang and Struhl, 2004). The cellular context dependence could account for these two non-mutually exclusive models and the Drosophila melanogaster sensory organ lineage, in which Notch unidirectional signalling is the only pathway involved (Heitzler and Simpson, 1991), represents an interesting study model in which the signal sending and receiving cells are easily distinguishable.
Each sensory organ, present on the adult D. melanogaster notum, is derived from a single precursor cell (pI), which undergoes a stereotyped series of four asymmetric cell divisions to generate five different cells, four composing the mechanosensory bristle and a glial cell (Fig. 1A,B). During each division, Notch signalling is involved in cell fate acquisition. For example, Notch is inhibited in the pI daughter cell, which adopts the pIIb cell identity and eventually activates Notch signalling in the adjacent daughter cell becoming the pIIa cell. Although data from different laboratories have emphasised the role of intracellular trafficking in the uni-directionality of Notch signalling between these two daughter cells (Benhra et al., 2011; Benhra et al., 2010; Berdnik et al., 2002; Coumailleau et al., 2009; Couturier et al., 2012; Djiane et al., 2011; Emery et al., 2005; Gallagher and Knoblich, 2006; Hutterer and Knoblich, 2005; Jafar-Nejad et al., 2005; Langevin et al., 2005; Rajan et al., 2009; Roegiers et al., 2005; Tong et al., 2010), little is known and understood about the regulators and membrane compartments involved in this process during the pI mitosis and/or in each of its daughter cells. Nonetheless, some recent data have emphasised the importance of a pI-daughter-cell-specific intracellular trafficking of Delta, Notch and/or a Drosophila Notch co-factor, Sanpodo (Spdo) (O'Connor-Giles and Skeath, 2003). In the signal sending pIIb cell, both basal to apical transcytosis of Delta mediated by Neuralized (see above and Benhra et al., 2010) and its trafficking toward an apical Actin Rich Structure (ARS) driven by WASp and the Arp2/3 complex (Rajan et al., 2009) are required for proper Notch signalling activation. While in the receiving cell, the clathrin adaptor complex AP-1 was genetically shown to be required for the correct localisation of Notch and Spdo (Benhra et al., 2011).
Sensory organ lineage and screen results. (A) Diagram of the adult sensory organ composed of two external cells (shaft and socket) and two internal cells (sheath and neuron). (B) Scheme of the cell precursor pI pupal lineage leading to the specification of the adult sensory organ cells and one apoptotic glial cell after four asymmetric cell divisions. In A and B, blue nuclei indicate cells responding to Notch signalling and red nuclei indicate cells sending Notch signals. (C–F). Examples of Notch-like bristle phenotype screened for, in the dsRNA genetic screen induced in the Drosophila notum. (D′,F′) Scheme of putative pI pupal lineages in case of a loss (D′) or gain (F′) of Notch signalling in all or some of the asymmetric cell divisions. (G) Numbers of candidates with dsRNA-induced adult phenotypes for each screen category (dark grey box: candidates with phenotype; light grey box: known Notch regulators with phenotype). Numbers into brackets indicate candidate genes/known Notch regulators screened in each category.
In order to identify novel regulators of the intracellular trafficking of Notch signalling major components, we have undertaken a tissue-specific double-strand RNA (dsRNA) genetic screen of 418 genes potentially involved in endocytosis and/or recycling within the endolysosomal pathway. To validate our in vivo Notch-specific strategy, 50 previously known Notch signalling regulators were screened, including 24 for which the function has not yet been studied during sensory organ lineage development. We took advantage of the fact that the genetic impairment of Notch signalling directly affects the development of external sensory organs and therefore allows for adult phenotype screening (Hartenstein and Posakony, 1990). Among the 113 Notch regulators identified based on adult phenotype, 61 were screened for, and 26 were found to cause a change in the steady state localisation of Notch, Sanpodo and/or Delta, in the pupal sensory organ lineage. In particular, we identified genes with previously unknown function in intracellular trafficking in Drosophila melanogaster such as CG27247 a regulator of AP-1 localisation, CG7787 putatively involved in the recycling pathway and members of the Tetraspanin family.
Results
Principle and validation of gene silencing-inducible screen
To screen specifically in the sensory organ lineage, we made use of a well-characterised and previously described dsRNA in vivo strategy (Mummery-Widmer et al., 2009). Taking advantage of the GAL4–UAS binary expression system (Brand and Perrimon, 1993), we induced gene silencing of selected genes specifically in the notum where the sensory organs develop. To do so, transgenic females carrying GAL4 under the control of a sensory organ promoter were crossed with males carrying an upstream activating sequence (UAS)–dsRNA transgenic construct. In the F1 progeny, GAL4 specifically activates the UAS and eventually induces gene silencing in the fly notum during sensory organ development. For each cross, two experimenters analysed at least 20 F1 progenies blindly. In order to identify specific regulators of Notch signalling, we scored for bristle phenotype on the notum (Fig. 1D–F). While a loss of bristle and/or double shafts without a socket cell reflects a loss of Notch signalling in the sensory organ lineage (Fig. 1D,D′), an excess of socket cells and/or double shafts with socket cells is correlated with a gain of Notch signalling in the sensory organ lineage (Fig. 1F,F′). As Notch signalling is also involved in the process of pI specification, we could, additionally, score for an excess of sensory organs reflecting a loss of Notch signalling in lateral inhibition (Fig. 1E). A genome-wide dsRNA screen was previously performed to identify regulators of Notch signalling in the sensory organs in which one sensory organ driver-GAL4 was used: pannier (pnr)–GAL4 (Mummery-Widmer et al., 2009). In this previous screen, we noticed that the phenotype observed for 360 (86%) of our genes could not be assessed as expression of the dsRNA induced either lethality or a morphological defect of the notum. This observation led us to modulate the strength of gene silencing by placing the F1 progenies at 18, 25 or 29°C. As the efficiency of the GAL4–UAS system is partially temperature sensitive (Mondal et al., 2007), this allows inducing lower (at 18°C) or higher (at 29°C) dsRNA expression. Additionally, to circumvent any technical bias of GAL4–UAS-induced phenotype and further describe the Notch-like phenotype, we independently used two GAL4 transgenic constructs which both drive expression in the notum during development: apterous (ap)–GAL4 (Calleja et al., 1996) and scabrous (sca)–GAL4 (Mlodzik et al., 1990).
In order to validate our Notch signalling-specific strategy (Knoblich, 2010), we choose to screen 50 known Notch signalling regulators and observe the same phenotype than previously described for 24 of them (see supplementary material Table S1). Although we could not reproduce the Notch loss-of-function-like phenotype of only two known Notch regulators, aristaless (Kojima et al., 2005) and Liquid facets (Wang and Struhl, 2004), our data indicate that our strategy allows specific screening for Notch regulators in the sensory organ lineage as previously described (Mummery-Widmer et al., 2009). Interestingly, we also observed a bristle phenotype for 14 of the 24 known Notch regulators whose function in the sensory organ lineage had not been previously described. Not all the known Notch signalling regulators appear to be involved in the Drosophila sensory organ lineage, which further highlights the cellular context dependence of Notch signalling in vivo as previously reported (Fuwa et al., 2006). For example, dsRNA against Kurtz and Nedd4 did not induce an adult phenotype while they are negative regulators of Notch signalling in the Drosophila wing vein (Mukherjee et al., 2005; Sakata et al., 2004).
Identification of Notch signalling regulators
To identify intracellular trafficking regulators of Notch signalling in Drosophila melanogaster sensory organs, we specifically screened for 368 genes from the endolysosomal pathway (see supplementary material Table S2). We selected these genes among members of intracellular trafficking regulator families mostly identified from yeast genetics and involved in different trafficking aspects such as, coat components [clathrin mediated endocytosis (Maldonado-Báez and Wendland, 2006)], lipid microdomain organisation [non-clathrin mediated endocytosis (Simons and Gerl, 2010)], cytoskeleton [actin, myosin and/or microtubules (Hehnly and Stamnes, 2007)], small GTPases, ubiquitylation/deubiquitylation factors involved in vesicle targeting (Murphy et al., 2009; Wennerberg et al., 2005), Endosomal Sorting Complex Required for Transport (ESCRT) complexes (Henne et al., 2011), membrane recognition and/or fusion regulators [such as SNAP receptors, SNAREs (Malsam et al., 2008), Exocyst (Hsu et al., 2004)] and ATPases (Forgac, 2007). We also based our selection on gene ontology (GO) annotation from FlyBase (using the GO terms: endocytosis, endosomal sorting, secretion) and selected putative orthologue(s) of traffic regulators identified in a Caenorhabditis elegans genetic screen (Balklava et al., 2007) or mammal proteomic screens (Baust et al., 2008; Baust et al., 2006). Noteworthy, the molecular function of 54 of these genes has not yet been defined in D. melanogaster (‘novel unknown function’ category in supplementary material Table S2).
To perform our screen we used 716 dsRNA lines, which represent the 368 candidates and 50 known Notch regulators, as we systematically screened with up to five different dsRNA lines, when available, in order to circumvent any effect due to the dsRNA construct insertion site. Expression of 264/716 (36.8%) dsRNA lines induced a bristle phenotype or lethality with either both or one of the GAL4 lines in our screen conditions. When two or more dsRNA lines induced a bristle phenotype, it was consistently the same gain or loss of Notch signalling phenotype(s), depending on the candidate, as we never observed opposite bristle phenotype between different dsRNA lines against the same candidate.
In order to confirm and validate the results, we reproduced the screen, with all the positive and lethal dsRNA hits and some negative ones as controls, using the same GAL4 lines and up to two additional GAL4 also driving expression in the notum: Eq–GAL4 (Pi et al., 2001) and/or pnr–GAL4 (Calleja et al., 1996). Among the dsRNA lines individually crossed with these several GAL4 lines, we observed that 175/264 (66.3%) dsRNA lines, representing 113 candidates, induced a reproducible bristle phenotype with one or more GAL4 (see supplementary material Table S2). To further validate our results, we had included 52 dsRNA lines, obtained from the National Institute of Genetics Fly Stock Center (NIG-Fly) or the Transgenic RNAi Project (TRIP), which target different part of the candidate RNA sequence than the dsRNA lines from the Vienna Drosophila RNAi Center (VDRC). In doing so, we confirmed the specific Notch-like bristle or lack of phenotype observed with the VDRC dsRNA lines (see supplementary material Table S2).
In the end, we firmly identified 113 Notch regulators in the sensory organ lineage (Table 1), which belong to the different screening categories that we initially defined (Fig. 1G). Specifically, we identified 77 previously unknown regulators of the Notch signalling pathway with a role in the sensory organ lineage. These regulators belong to all our initial screen categories, which cover various aspects of intracellular trafficking. The vast majority of the observed phenotypes resemble those of a loss of Notch signalling. Nevertheless, phenotypes similar to gain-of-Notch signalling were observed for 20 genes from various screen categories including members of coat components (AP-1 and AP-2) or ESCRT complexes (0, I and III). Both AP-1 and AP-2 complexes had previously been identified as regulators of Spdo trafficking and eventually as negative regulators of Notch signalling pathway during binary cell fate decision (Benhra et al., 2011; Berdnik et al., 2002; Tong et al., 2010). Therefore, our genetic screen clearly led to the identification of potential intracellular trafficking regulators directly involved in the regulation of Notch signalling via its major components.
Complete list of positive hits from the genetic dsRNA screen
Steady-state localisation of Notch, Sanpodo, Delta and cell fate identity
A Notch-like adult sensory organ phenotype could be due to a defect in Notch signalling component traffic and/or induced by unrelated defects such as in cell fate determinant segregation, cell polarity, cell cycle control and/or general intracellular trafficking. Out of the 113 candidates that we genetically identified as Notch signalling regulators in the sensory organ lineage, we wanted to identify those involved in the intracellular trafficking of the major components of Notch signalling: Delta, Notch and its co-factor Spdo. This study was made feasible as they present a specific steady-state pattern of subcellular localisation in the sensory organ pupal lineage during pI mitosis and at the pI daughter cell stage (Fig. 2A–F′″) (also see Benhra et al., 2010). In a wild-type lineage, while both apical Delta and Notch are mostly localised at the cortex (Fig. 2A″,A′″, D″,D′″), basolateral Delta is found in vesicles in mitotic pI and pIIb/pIIa cells (Fig. 2B″,C″,E″,F″). Spdo has a more dynamic pattern of localisation: cytoplasmic in the mitotic pI (Fig. 2A′–C′), its localisation becomes asymmetric in the pI daughter cells. While Spdo is enriched along the apicobasal interface of pI daughter cells (Fig. 2D′–F′), Spdo is mostly localised in vesicles in the anterior pIIb cell but at the basolateral plasma membrane in the posterior pIIa cell (Fig. 2E′). Changes in Notch, Spdo and/or Delta localisation could either originate from an aberrant cell-fate identity acquisition in the lineage (two pIIb or pIIa-like cells) or reflect trafficking defect(s) causing a defective Notch signalling pathway. As a proof of principle, we recently demonstrated that the clathrin adaptor complex AP-1, identified in this screen, controls Spdo and Notch trafficking in the sensory organ lineage. In particular, a lack of AP-1 function induces Spdo and Notch subcellular localisation changes and an adult gain of Notch signalling phenotype (Benhra et al., 2011). Similarly, loss of Neur, Sec15 or Arp2/3 functions induce changes in Spdo and/or Delta subcellular localisation correlated with adult loss of Notch signalling phenotypes (Benhra et al., 2010; Jafar-Nejad et al., 2005; Le Borgne and Schweisguth, 2003; Rajan et al., 2009; Roegiers et al., 2005).
The steady-state pattern of localisation of Sanpodo, Delta and Notch. Localisation of Sanpodo (green), Delta (red) and Notch (blue) in wild-type pI dividing cell (A–C′″) and at the pI daughter-cell stage (D–F′″). A–A′″, D–D′″ show apical and B–B′″, E–E′″ basal confocal slices. C–C′″, F–F′″ show orthogonal sections of cells from A–B′″ and D–E′″, respectively. The asymmetric localisation of Spdo in endosomes in the anterior cell, and in endosomes and at the basolateral cortex in the posterior cell reflects the differential cell identity of the pI daughter cells (E′). In all panels, anterior is left; scale bar: 5 µm.
Regulators of Notch, Sanpodo and/or Delta subcellular localisation identified in the screen
Among the 113 Notch regulators we identified, we decided to analyse those that were not previously known to cause subcellular localisation changes and/or that do not have described function in cell polarity or in asymmetric cell division (see supplementary material Table S3). Among the 61 genes that we screened for a dsRNA-induced change in Delta, Notch and/or Spdo steady-state localisation (using one dsRNA line for each), 32 did not cause any visible defect while three genes (gigas, CG31048 and CG8435) caused a lack of pI specification (revealed by an absence of Spdo staining, our sensory organ identity marker), which explains the observed adult bristle loss phenotype (see supplementary material Table S3). Twenty-six genes caused a phenotype of Notch, Spdo and/or Delta mis-localisation at the pI and/or pI daughter cell stage (we used the threshold of at least three, out of 20 analysed, sensory organs presenting the same phenotype on two different nota). Although a wide range of phenotypes was observed, they can be subdivided into three major categories (Table 2; Figs 3–5).
Examples of accumulation at pI daughter cell contacts. Localisation of Sanpodo (green), Delta (red) and Notch (blue) in pI daughter cells of ChmpdsRNA (A–B′″), CG10341dsRNA (C–D′″) and Cullin-3dsRNA (Cul-3; E–F′″). A–A′″, C–C′″ show apical, and E–E′″ basal confocal slices. B–B′″, D–D′″ and F–F′″ show orthogonal sections of cells from A–A′″, C–C′″ and E–E′″, respectively. In ChmpdsRNA, Spdo accumulates subapically (B′, arrowhead) and at the apical interface (A′, arrow), between the two daughter cells, with Delta and Notch (A″,A′″, arrows). In CG10341dsRNA, Spdo, Delta and Notch accumulate at the apical interface between the two daughter cells (C′–C′″ and D′–D′″, arrows). In Cullin-3dsRNA, Spdo and Delta specifically accumulate at the lateral membrane between the pI daughter cells (E′,E″ and F′,F″, arrows). In all panels, anterior is left; scale bar: 5 µm.
Examples of apical and/or basal accumulation. Localisation of Sanpodo (green), Delta (red) and Notch (blue) in pI daughter cells of Exo84dsRNA (A–A′″), Rab35dsRNA (B–C′″) and Tsp68CdsRNA (D–D′″). B–B′″ show apical and A–A′″, C–C′″, D–D′″ basal confocal slices. In Exo84dsRNA, Spdo and Delta are found at the basolateral membrane or in close proximity to the plasma membrane of both daughter cells (A′–A″, arrows), a lineage somewhat reminiscent of two pIIa-like cells. In Rab35dsRNA, Spdo, Delta and Notch accumulate at the apical side of the anterior daughter cell (B′–B′″, arrowheads). Spdo is also found in basal vesicles in both daughter cells (C′, arrows), a lineage reminiscent of two pIIb-like cells. In Tsp68CdsRNA, Delta can be found at the basolateral cortex of epidermal cells (D″, arrowheads). In all panels, anterior is left; scale bar: 5 µm.
Complete list of genes that affect Notch, Sanpodo and/or Delta subcellular localisation at the pI daughter cell stage after dsRNA induction
(1) Accumulation at pI daughter cell contact (Fig. 3). An excess of Spdo is seen subapically between pI daughter cells of CG2747dsRNA, Vacuolar protein sorting 28 (Vps28dsRNA) and Chmp1dsRNA (Fig. 3B′, arrowhead; Fig. 6B′, arrowhead; and data not shown). We also observed an accumulation of Spdo, Notch and Delta at the apical interface between the pI daughter cells in CG2747dsRNA, Signal transducing adaptor molecule (StamdsRNA), Vps28dsRNA, Chmp1dsRNA, Vps2dsRNA and CG10341dsRNA (Fig. 3A′–A′″,C′–C′″,D′–D′″, arrows; Fig. 6A′–A′″, arrows; and data not shown). Finally, we detected an accumulation of Spdo and Delta at the lateral membrane between pI daughter cells of CG7787dsRNA and Cullin-3 (Guftagu/Cul-3dsRNA) (Fig. 3E′,E″,F′,F″, arrows; and data not shown). Strikingly, accumulation of Spdo subapically and/or, with Notch and Delta, at the apical interface between pI daughter cells correlates with a Notch gain-of-function adult phenotype, while accumulation of Spdo and Delta at the lateral membrane between pI daughter cells is associated with Notch loss-of-function adult phenotype (Table 2).
CG2747 regulates clathrin adaptor AP-1 intracellular localisation. (A–B′″). Localisation of Sanpodo (green), Delta (red) and Notch (blue) in pI daughter cells of CG2747dsRNA. A–A′″ show apical confocal slice and B–B′″ show orthogonal section of cells from A–A′″. Spdo accumulates subapically (B′, arrowhead) and at the interface between the two daughter cells with Delta and Notch (A′–A′″, arrows). (C) Excess of double shafts with sockets cells (arrows) observed on a CG2747dsRNA notum. This adult phenotype is reminiscent of gain of Notch signalling. (D–F″). Partial loss of HEATR5B staining (p200, green) in the median part of the notum where UAS–CG2747dsRNA is induced by ap–GAL4 (D,D′ left side, E–E″) compared with the posterior part of the notum where ap–GAL4 does not drive UAS expression (D,D′ right side, asterisk, F–F″). E–F′ are higher magnifications of D–D′, taken at the level of adherens junctions (DE-CAD, red in E,E′ and F,F′). (G) Loss of AP-1γ staining in the median part of the notum where UAS–CG2747dsRNA is induced by ap–GAL4 (right side) compared with the anterior head in which ap–GAL4 does not drive UAS expression (left side, star). (H–H″) HEATR5B staining is not affected in AP-47SHE11 mutant cells (red, H and H″, inside dotted lines). Mutant cells are identified by the absence of nls–GFP (H, blue, inside dotted line). H–H″ are confocal slices taken at the level of adherens junctions (DE-CAD, green in H and H′). In panels A–B′″, D–H″, anterior is left. Scale bar: 5 µm (A–B′″,G), 200 µm (D,D′) and 15 µm’>;(E–F″,H–H″).
(2) Vesicle excess (Fig. 4). A lineage reminiscent of two pIIb-like cells, in which Spdo is present in intracellular basal compartments in both daughter cells and absent from the cortex, is observed in specifically Rac1-associated protein 1 (Sra-1dsRNA), Origin recognition complex subunit 6 (Orc-6dsRNA), peanut (pnutdsRNA), Septin 5 (Sep5dsRNA), Septin 2 (Sep2dsRNA), Rab35dsRNA, Cul-3dsRNA, Vha16-1dsRNA and Vha16-2dsRNA (Fig. 4B′, arrows, Fig. 5C′, arrows, and data not shown). This mislocalisation is consistently associated with Notch loss-of-function phenotypes (Table 2). In a second group of phenotypes, we observed, in both pI daughter cells, enlarged basal vesicles that are positive for Spdo, Delta and Notch in CG7787dsRNA, StamdsRNA, Vps2dsRNA or Vacuolar protein sorting 4 (Vps4dsRNA) (Fig. 4D′–D′″, arrows; and data not shown). Accumulation of Spdo, Delta and Notch in enlarged intracellular compartments is predominantly associated with Notch gain-of-function phenotypes (Table 2). Finally, an excess of Spdo- and Delta-positive basal compartments is observed in the anterior cell and towards the anterior in the posterior cell in Receptor mediated endocytosis 8 (Rme-8dsRNA), while only in the anterior cell in l(2)dtldsRNA (Fig. 4F′–F′″, arrows; and data not shown) and correlates with Notch loss-of-function phenotype (Table 2).
Examples of excess vesicles in pI daughter cells. Localisation of Sanpodo (green), Delta (red) and Notch (blue) in pI daughter cells of Vha16-2dsRNA (A–B′″), Vps4dsRNA (C–D′″) and l(2)dtldsRNA (E–F′″). A–A′″, C–C′″, E–E′″ show apical and B–B′″, D–D′″ and F–F′″ basal confocal slices. In Vha16-2dsRNA, Spdo is found in basal vesicles in both daughter cells (B′, arrows), a lineage reminiscent of two pIIb-like cells. In Vps4dsRNA, Spdo, Delta and Notch colocalise in enlarged basal vesicles in both daughter cells (D′–D′″, arrows). In l(2)dtldsRNA, an excess of Spdo-, Delta- and Notch-positive basal compartments is observed in the anterior cell (F′–F′″, arrows). In all panels, anterior is left; scale bar: 5 µm.
(3) Apical and/or basolateral accumulation (Fig. 5). Spdo and Delta are found at the basolateral membrane or in close vicinity to the plasma membrane of both daughter cells in StamdsRNA, Syntaxin 7 (Syx7dsRNA), Exo84dsRNA, Sec6dsRNA and Sec5dsRNA (Fig. 5A′,A″, arrows; and data not shown). This mislocalisation is somewhat reminiscent of a two pIIa-like cells lineage and consistently associated with Notch gain-of-function phenotypes (Table 2). We also observed an accumulation of Spdo, and to a certain extent Delta and Notch, at the apical side of the anterior daughter cells of Sep2dsRNA and Rab35dsRNA (Fig. 5B′–B′″, arrowheads; and data not shown), associated with a Notch loss-of-function phenotype (Table 2).
Apart from these sensory organ lineage-specific phenotypes, we observed an accumulation of Delta at the basolateral cortex of surrounding epidermal cells in O-fucosyltranferase 1 [O-fut1dsRNA; a known Notch trafficking regulator (Sasamura et al., 2007)], Tetraspanin 47F (Tsp47FdsRNA) and Tetraspanin 68C (Tsp68CdsRNA; Fig. 5D″, arrowheads; and data not shown). In all cases, this phenotype is correlated with a lateral inhibition defect (excess of pI on the pupal notum based on Spdo staining; Table 1), which suggests a loss of Notch signalling during pI specification. This accumulation of Delta at the basolateral cortex could result from either a decrease in Delta internalisation or an increase in Delta exocytosis to the basolateral membrane. To test if basolateral Delta endocytosis could be affected, we performed a 15 min pulse-chase labelling experiment (Benhra et al., 2010) to monitor Delta internalisation in living pupae epidermal cells but did not observe any endocytosis failure (data not shown). These results raise the possibility that O-fut1, Tsp68C and Tsp47F could regulate, directly or not, Delta basolateral exocytosis.
All together, our genetic and cellular results clearly validate the essential function of intracellular trafficking in regulating Notch-signalling-dependent binary cell fate acquisition. Indeed, we identified 26 genes for which a Notch signalling adult phenotype is associated with a change in intracellular localisation of major Notch signalling components after the first asymmetric cell division.
CG2747 regulates clathrin adaptor AP-1 intracellular localisation
Among the genes isolated in the genetic and cellular screen, CG2747dsRNA phenocopies the loss of AP-1 function (Benhra et al., 2011). Indeed, we observed a subapical accumulation of Spdo (Fig. 6B′, arrowhead) and an accumulation of Spdo, Delta and Notch at the apical interface between pI daughter cells (Fig. 6A–B′″, arrows). To further validate the specificity of the phenotype induced by dsRNA, we used two independent dsRNA lines obtained from the NIG-Fly and TRIP Centers (lines 2747R-3 and BL29322, respectively), which target different regions of the transcript (regions 1081–1612 and 407–906 of the transcript CG2747-RD, respectively), and observed the same phenotype (data not shown). We also observed an excess of socket cells and/or double shafts with socket cells with both dsRNA lines (Fig. 6C, arrows; and data not shown), which is reminiscent of Notch signalling gain-of-function phenotype observed in AP-1 loss of function (Benhra et al., 2011).
Although CG2747 function has not yet been studied in D. melanogaster, it belongs to the highly conserved HEAT repeat-containing protein 5 (HEATR5) family (Fernández and Payne, 2006), previously shown to physically interact with the ear-domain of murine γ-adaptin subunit of the AP-1 complex (Lui et al., 2003). For instance CG2747 shares 49% of sequence identities with the Homo sapiens HEATR5B and when we used the p200 antibody against the mammal HEATR5B (Hirst et al., 2005), we observed a staining in intracellular vesicles and at the junction of wild-type notum epithelial cells (Fig. 6F,F″). In the area of a notum where CG2747dsRNA is induced, p200 staining is greatly reduced (Fig. 6E,E″ compared to Fig. 6F,F″), suggesting that the mammal antibody recognises the Drosophila protein. Furthermore, it was demonstrated in Saccharomyces cerevisiae, that Laap1, sole member of the HEATR5 family, is necessary for proper AP-1 intracellular localisation (Fernández and Payne, 2006). To investigate whether CG2747 shares the same function in D. melanogaster, we analysed the localisation of AP-1 on the pupal notum. AP-1γ staining is greatly reduced in the region of the notum where UAS–CG2747dsRNA expression is induced in comparison with the anterior head in which the ap–GAL4 driver is not active (Fig. 6G). In the converse experiment, we observed that p200 staining is not affected in AP-47 (gene encoding the mu subunit of AP-1 complex) homozygote mutant cells (Fig. 6H–H″). Thus CG2747 localisation is independent of AP-1 activity. We concluded that CG2747dsRNA reduces the level of CG2747 at a sufficient degree to prevent AP-1γ membrane localisation and phenocopies the AP-1 cellular and adult phenotypes. All these results strongly support the model in which CG2747 is necessary for AP-1 proper intracellular localisation and function in Drosophila.
Discussion
Our dsRNA genetic screen, performed in D. melanogaster notum using the GAL4–UAS system, allowed us to specifically identify 113 Notch signalling regulators among 418 candidates chosen for previously described or suggested function in intracellular trafficking (Table 1). Importantly, up to 76% of the regulators we identified were not found in a similar genetic screen performed at a genome-wide level (Mummery-Widmer et al., 2009). Our study clearly shows that using multiple GAL4 drivers and three different temperatures increases the efficiency of such dsRNA genetic screen as it allows identifying optimal dsRNA technical conditions for each construct. In particular, using different GAL4 drivers limit the false positive or negative results observed when the driver itself induces morphological defects and/or when the expression of the dsRNA induces lethality or notum morphological defects.
Recovering almost 30% of positive hits in a screen is unusual. We interpret this high number first as a reflection of the tight interconnection between membrane traffic and both Delta and Notch signalling to ultimately ensure the proper spatiotemporal control of the pathway. Second, this high number is also explained by the fact that many regulators are acting as protein complexes (APs, ESCRTs, Exocysts, ATPases, septins etc.). This property could be used to confirm the specificity of the dsRNA effect, as inactivation of gene products belonging to the same protein complex is expected to give similar phenotype. If this prediction is fully fulfilled for AP-1, AP-2, septins and the ubiquitin ligase complex SCF, it is only partially fulfilled for the ESCRT and Exocyst complexes (see below).
Although these results are novel and further validate the concept of a regulation of Notch signalling by intracellular trafficking (for review, see Fortini and Bilder, 2009; Fürthauer, González-Gaitán, 2009; Kopan and Ilagan, 2009; Le Bras et al., 2011; Musse et al., 2012; Yamamoto et al., 2010), we were aiming at identifying novel regulators of the subcellular localisation of three major Notch signalling components: Notch, Spdo and Delta. Indeed, the observed adult phenotypes could result from various and multiple defects in Notch signalling during pI asymmetric cell division and/or at the pI daughter cell stage. Furthermore, our identified genes could regulate various molecular events such as cell fate determinants segregation, Delta, Spdo and/or Notch proper subcellular localisation via endocytosis and/or recycling. Therefore, we decided to take advantage of the fact that Spdo, Delta and Notch localisation are dynamic during pI division and at the pI daughter cell stage in the pupal notum. We were able to observe localisation changes for 26 out of the 61 genes that we studied to further understand the adult phenotypes induced by dsRNA (Table 2). Three major categories of localisation changes were identified at the pI daughter cell stage: accumulation at pI daughter cell contact, vesicle excess (and in some cases, enlarged vesicles), and apical and/or basolateral accumulation. And, in each of these categories, we observed pupal lineages which correlate either with an adult Notch gain- or loss-of-function-like phenotype (as illustrated with Spdo localisation in supplementary material Fig. S1).
For 10 genes, changes in Notch component localisation can be a direct consequence of their inactivation or reflect a change in pI daughter cell fate acquisition. Indeed, we observed a pupal lineage somewhat reminiscent of two pIIa-like (Syx7, Exo84, Sec6, Sec5) or two pIIb-like (Sra-1, Orc6, pnut, Sep5, Vha16-1, Vha16-2) daughter cells (supplementary material Fig. S1B,F respectively). And these lineages correlate with the observed adult phenotypes i.e. Notch gain of function or loss of function, respectively. Although these results do not elucidate the function of these genes on Notch signalling regulation, they indicate that Notch signalling can be similarly regulated in various cellular contexts. For example, it was previously shown that the Vacuolar ATPase functions to control the acidification of endosomes required for Notch activation after binding by its ligand in the eye imaginal disc and ovaries (Vaccari et al., 2010; Yan et al., 2009).
Surprisingly, our results suggest that part of the same complex might regulate different aspects of Notch signalling in the sensory organ lineage. Indeed, we observed a gain-of-Notch-signalling-like adult phenotype for three Exocyst components (Exo84, Sec6 and Sec5) but also a loss-of-Notch-signalling-like adult one for Exo84. These results can either reflect a bias in the RNAi silencing which might not be as effective for each Exocyst subunits and/or suggest that the Exocyst might have several functions during pI asymmetric cell division with opposite role in the regulation of Notch signalling. Further studies will be necessary to validate these results and to define if Exo84 could function with Sec15, another Exocyst component, which positively regulates Spdo and Delta post-endocytic trafficking in pI daughter cells (Jafar-Nejad et al., 2005).
For the first time, we identified members of the septin family (pnut and Sep5) and one of their regulator [Orc-6 (Chesnokov et al., 2003)] as putative regulators of Notch signalling. Septin complexes can act as scaffolds and/or diffusion barriers in various cellular events such as cytoskeleton organisation, cytokinesis, membrane organisation and vesicle targeting which could potentially regulate the pI asymmetric cell division (for a review, see Cao et al., 2009). It is necessary to further decipher the pupal phenotype to define if these septins directly regulate Notch signalling traffic and/or pI cytokinesis (N.B. Founounou and R.L.B. unpublished). While the pupal lineage is somewhat reminiscent of two pIIb-like daughter cells, we also observed an accumulation of Spdo, Notch and Delta at the apical side of the anterior pI daughter cell of Sep2dsRNA as well as of Rab35dsRNA (supplementary material Fig. S1G). Their common phenotype is not surprising knowing that human Rab35 was proposed to play an essential control on the terminal step of cytokinesis in part by controlling SEPT2 subcellular distribution during cell division (Kouranti et al., 2006). Although it is not yet possible to decipher if this apical accumulation is a cause or a consequence of the Notch pupal and adult loss-of-function phenotype, this data identify two putative regulators of apical localisation and confirm that Notch signalling major components are differentially trafficked between pI daughter cells.
Inactivation of 14 genes led to localisation changes indicating multiple subcellular sites (plasma membrane, vesicular compartments, interface between the two daughter cells), which appear to be essential to the fine regulation of Notch signalling in the Drosophila sensory organ lineage. Indeed, subapical accumulation of Spdo, localisation of Notch, Spdo and Delta at the apical interface between pI daughter cells, and/or an excess of enlarged endosomes in both daughter cells are associated with adult Notch gain-of-function-like phenotypes (supplementary material Fig. S1C,D,E, respectively). However, accumulation of Spdo and Delta at the lateral membrane between pI daughter cells correlates with Notch signalling loss-of-function phenotype (supplementary material Fig. S1H,I).
In a control situation, Notch accumulates transiently at the apical interface between pIIb and pIIa (Benhra et al., 2011; Couturier et al., 2012). Accumulation of Spdo together with Notch at this apical pIIb/pIIa interface has previously been observed in AP-1 loss-of-function mutants and correlated with a gain of Notch signalling (Benhra et al., 2011). Because, impairment of Delta trafficking towards the pIIb/pIIa interface in Arp2/3 mutants leads to a loss of Notch signalling (Rajan et al., 2009), it was proposed that Delta–Notch interaction resulting in Notch activation is taking place at the apical pIIb/pIIa interface that could function as a signalling platform (Benhra et al., 2011). Nonetheless, this proposal awaits experimental demonstration. This proposal was recently challenged by F. Schweisguth and colleagues, who generated a functional Notch construct tagged with GFP and expressed at physiological level (NiGFP) (Couturier et al., 2012). Notch activation is reported to occur 15–45 min after cytokinesis and productive signalling is proposed to take place at the pIIa/pIIb interface of the cytokinetic furrow, where NiGFP accumulates in numb mutant background or when Dynamin-dependent endocytosis is blocked.
Adult and subcellular AP-1 loss-of-function-type phenotypes are observed when two genes with previously unknown function in Drosophila, CG10341 and CG2747, were inactivated. Both of them had been identified as putative membrane trafficking regulators in a C. elegans screen (Balklava et al., 2007). CG10341 belongs to the Nuclear Assembly Factor 1 (NAF1) family involved in ribosome biogenesis, which suggests an indirect role, if so, in intracellular trafficking. On the contrary, CG2747 belongs to the HEATR5 family and we were able to show that CG2747 is required for the clathrin adaptor AP-1 complex subcellular localisation, similarly to its putative orthologue of the S. cerevisiae Laa1p (Fernández and Payne, 2006). We observed that the human HEATR5B/p200 antibody staining is affected in CG2747dsRNA, which supports an evolutionary conservation of the HEATR5 function among metazoans. Nonetheless, the function of human HEATR5B/p200 remains unknown as no phenotype could be observed in p200-depleted cells maybe due to a poor silencing efficiency and/or a redundancy with the other human HEATR5 member, HEATR5A (Lui et al., 2003). Therefore, we identified a new regulator of Notch signalling that acts as a major regulator of the clathrin adaptor complex, AP-1.
We also observed an accumulation of Notch, Spdo and Delta at the interface and/or in endosomes in both daughter cells, correlated with adult Notch gain-of-function phenotypes, when we inactivated members of the ESCRT complex (Stam, Vps28, Chmp1, Vps2 and Vps4). Accumulation at the interface and/or endosomes can result from a blockade in endosome maturation when ESCRT function is impaired and it has already been described that accumulation of Notch in endocytic compartments can result in an ectopic ligand-independent activation of Notch (Herz et al., 2009; Moberg et al., 2005; Thompson et al., 2005; Vaccari and Bilder, 2005; Vaccari et al., 2009). Additionally, ESCRT complexes are involved in various cellular mechanisms: cargo engagement and/or deubiquitylation, maturation of multi vesicular bodies, vesicle budding and/or cytokinesis (for a review, see Henne et al., 2011). In our screen, impairment of different ESCRT pathway components led to opposite adult phenotypes i.e. loss- or gain-of-Notch-signalling-like ones depending on the complex and/or its subunit(s) depleted by dsRNA (see Table 1). However, we did not observe any Spdo, Delta or Notch localisation changes associated with these adult loss-of-function phenotypes. Further studies are, therefore, necessary to elucidate our genetics results by identifying which subcellular mechanisms and which cargo(es) are regulated by the different identified ESCRT subunits to control Notch signalling.
Accumulation of Spdo and Delta at the lateral membrane between pI daughter cells is correlated with adult Notch loss-of-function phenotypes in Cul-3dsRNA and CG7787dsRNA (supplementary material Fig. S1H,I respectively). This lateral membrane was previously named apical actin-rich structure ‘stalk’ and identified as the lateral part of branched actin network present at the interface and through which endocytosed Delta traffic back (Rajan et al., 2009). Therefore, CG7787 and Cul-3 might regulate Spdo and/or Delta basolateral endocytosis and/or recycling required for Notch signalling as this accumulation phenotype is correlated with a loss of Notch signalling. While Cul-3 is a subunit of E3 ubiquitin ligase, CG7787 putatively encodes a guanyl-nucleotide exchange factor. CG7787 belongs to the MSS4/DSS4 family proposed to function as chaperone for misfolded Rab proteins (Nuoffer et al., 1997) and in particular, Rabs associated with the Exocyst pathway (Itzen et al., 2006). Therefore, CG7787 might be involved in the same recycling pathway as Sec15 (see above) and positively regulates Notch signalling. Other data support the general idea that a recycling pathway positively regulates Notch signalling in the sensory organ lineage. Indeed, we observed an excess of Spdo-, Notch- and Delta-positive vesicles either in the anterior or towards the anterior cell in Rme-8dsRNA and l(2)dtldsRNA, which is associated with an adult Notch loss-of-function phenotype. Although l(2)dtl function in intracellular trafficking is still unknown, Rme-8 was shown to regulate a recycling pathway (Shi et al., 2009). All those results confirm that Spdo, Notch and Delta transiently traffic through the lateral membrane and/or endosomes to ensure a proper Notch signalling.
Finally, we observed a basolateral accumulation of Delta in epithelial cells of O-fut1dsRNA, Tsp47FdsRNA and Tsp68CdsRNA nota and that Delta endocytosis is not affected which suggest the existence of basolateral Delta exocytosis. In support of this hypothesis, it was already demonstrated that Delta can be fucosylated by mammalian O-fut1 (Panin et al., 2002) but this data remained to be demonstrated in vivo and in Drosophila. When confirmed by further studies in classical genetic mutants, these results will eventually highlight a function for Delta basolateral exocytosis and also a new role of the tetraspanins family on Notch signalling.
In conclusion, our screen led us to identify intracellular trafficking regulators of major Notch signalling actors. Although it is still debatable whether the subcellular localisation changes observed are a cause or a consequence of the Notch signalling phenotype, the screen we performed led to the identification of 11 previously unknown regulators of Notch signalling (CG2747, Tsp47F, Orc-6, pnut, Sep2, Sep5, Rab35, CG7787, CG10341, l(2)dtl and Rme-8). Without any doubt, further analyses of our identified genes will bring a better understanding of their trafficking function in regulating Notch-signalling-dependent binary cell fate acquisition, as well as of their putative molecular interactions.
Materials and Methods
Drosophila stocks and genetics
Unless otherwise stated, fly stocks were obtained from the Bloomington Drosophila Stock Center. Driver–GAL4 stocks used in this study were: ap–GAL4 (Calleja et al., 1996), sca–GAL4 (Mlodzik et al., 1990), Eq–GAL4 (Pi et al., 2001) and pnr–GAL4 (Calleja et al., 1996). All dsRNA transgenic lines were supplied by the Vienna Drosophila RNAi Center [VDRC, (Dietzl et al., 2007)]; except lines (as indicated in supplementary material Tables S1–S3), which were obtained from the National Institute of Genetics Fly Stock Center (NIG-FLY) or the Transgenic RNAi Project (TRIP) via the Bloomington Drosophila Stock Center. For RNAi-induced phenotype study, crosses between UAS-hairpin RNAi males and driver-GAL4 females were raised at 18°C, 25°C or shifted at 29°C during L2–L3 larval stages. For each cross in which the genotypes were blinded for objectivity purpose, two experimenters examined at least 20 flies sensory organ distribution and/or morphological phenotypes. w1118 males were crossed with driver-GAL4 females for control experiments. To obtain AP-47SHE11 mitotic clones, we used the FLP-FRT technique and the stocks (1) y w hs-FLP; FRT82B, Ubi-GFP(S65T)nls and (2) FRT82B, AP-47SHE11/TM6 Tb Sb, as previously described (Benhra et al., 2011). Heat shocks were performed at L2 and L3 during 30 min.
Immunocytochemistry
Pupae were aged for 17 h to 20 h after puparium formation, dissected in 1×PBS, fixed in 4% paraformaldehyde and stained as previously described (Le Borgne and Schweisguth, 2003). Primary antibodies used were mouse anti-Notch Extra Cellular Domain (NECD; DSHB, 1∶100), rabbit anti-Spdo [a kind gift from J. Skeath; 1∶1000 (O'Connor-Giles and Skeath, 2003)], guinea pig anti-Delta Extra Cellular Domain (GP582, a kind gift from M. Muskavitch; 1∶2000), rabbit anti-HEATR5B [p200; a kind gift from M. Robinson; 1∶20 (Hirst et al., 2005)], rat anti-DE-CAD (DCAD2; DSHB 1∶250) and mouse anti-AP-1γ [1∶100 (Benhra et al., 2011)]. Cy2-, Cy3- and Cy5-coupled secondary antibodies (1∶500) were from Jackson Laboratories. Delta 15 minutes internalisation assays were performed with mouse anti-Delta DSHB (1∶100), as previously described (Benhra et al., 2011).
Images were acquired with a Leica SPE confocal microscope, which was noise-suppressed using the smooth function of ImageJ. In all figures, Notch (Cy5-) images were colour balanced using ImageJ. Defect in lateral inhibition was acknowledged when more than four to five sensory organs were systematically detected with a 63× 1.4 NA lens, zoom 3 on a notum.
Acknowledgments
We thank M. Muskavitch, M. Robinson, J. Skeath, the Bloomington Stock Center, the Vienna Drosophila RNAi Center, the TRiP at Harvard Medical School (NIH/NIGMS R01-GM084947) and the National Institute of Genetics Fly Stock Center for providing antibodies or fly stocks, as well as the Microscopy Rennes Imaging Center. The monoclonal antibody generated by S. Artavanis Tsakonas (NECD) was obtained from the Developmental Studies Hybridoma Bank, generated under the auspices of the National Institute of Child Health and Human Development, and maintained by the University of Iowa Department of Biological Sciences. We thank members of the Le Borgne laboratory for helpful discussions. We thank A. Pacquelet and G. Michaux for critical reading of the manuscript. Special thanks to Amy Winehouse for her music that accompanied us, while we screened around 100,000 flies.
Footnotes
↵‡ Present addresses: GReD Laboratory, CNRS UMR 6293, INSERM U1103, Clermont Université, 63177 Aubière, France; Université d'Auvergne, Faculté de Médecine, 63000 Clermont-Ferrand, France
↵§ Present address: Department of Developmental and Regenerative Biology, Mount Sinai School of Medicine, New York, USA
Funding
This work was supported by the Action Thématique Incitative Prioritaire programme CNRS to R.L.B.; Région Bretagne (Programme Accueil de COMpétences en Bretagne ‘Notasid’ [grant number 2168 to R.L.B.]; Fondation ARC pour la Recherche sur le Cancer [grant number 4905 to R.L.B.]; Fondation pour la Recherche Médicale to R.L.B., Rennes Métropole [grant ‘Aide d'Installation Scientifique’ to S.L.B.]; and La Ligue contre le Cancer 35 to R.L.B.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.110171/-/DC1
- Accepted June 12, 2012.
- © 2012. Published by The Company of Biologists Ltd