Slow maturation time of Cherry relative to the turnover of Notch

One possible difference between GFP and Cherry is maturation speed, which depends on the kinetics of both protein folding and chromophore maturation (Verkhusha et al., 2001; Shaner et al., 2005; Merzlyak et al., 2007; Subach et al., 2009; Iizuka et al., 2011; Macdonald et al., 2012; Hebisch et al., 2013). Although measured maturation speed varied greatly depending on experimental approaches, it is usually considered to be in the 5–20- and 40–80-min range for GFP and Cherry, respectively, at 37°C (Verkhusha et al., 2001; Shaner et al., 2005; Merzlyak et al., 2007; Subach et al., 2009; Iizuka et al., 2011; Macdonald et al., 2012; Hebisch et al., 2013). Of note, longer maturation times were measured at 20°C (Verkhusha et al., 2001; Shaner et al., 2005; Merzlyak et al., 2007; Subach et al., 2009; Iizuka et al., 2011; Macdonald et al., 2012; Hebisch et al., 2013), close to the temperature at which flies are usually kept (25°C). This difference in maturation time has previously been used to design a fluorescent timer that reports on the stability of cytoplasmic proteins (Khmelinskii et al., 2012). Thus, the very weak Cherry fluorescence of NiCherry and NiGFP4Cherry5 at the plasma membrane relative to GFP of NiGFP or NiGFP4Cherry5 might result from the slower maturation of Cherry. Accordingly, Notch at the plasma membrane would mostly correspond to recently synthesized molecules. To test this interpretation, we studied the fluorescence of NiGFP4Cherry5 in cells that are mutant for the nonvisual β-arrestin gene known as kurtz (krz) in flies. The activity of krz is required to down-regulate Notch: in krz mutant cells, Notch accumulated at the cell surface (Mukherjee et al., 2005). Here, we found that the Cherry fluorescence of NiCherry was clearly detectable at the cortex of krz mutant cells (Fig. 2, A and A′). Thus, slowing down the turnover of Notch allowed the detection of the Cherry fluorescence of NiCherry at the plasma membrane. We therefore suggest that the Cherry fluorescence of NiGFP4Cherry5 was not detectable at the plasma membrane as a result of the rapid turnover of Notch and the slow maturation of Cherry (Fig. 2 B).

Open in a separate window

Figure 2.

Maturation of Cherry-tagged Notch. (A and A′) Cherry fluorescence of NiCherry (red) was detected at the apical plasma membrane in krz mutant cells (marked by the loss of the GFP marker; clone border indicated in white) but not in wild-type cells. This indicated that slowing down the turnover of Notch at the cell surface by removing the activity of krz allowed for the maturation of Cherry. Bar, 5 µm. (B) Diagram illustrating the time-dependent maturation of NiGFP4Cherry5. Initially, dark NiGFP4Cherry5 (white spots) is synthesized. Because GFP is maturing faster than Cherry, bright NiGFP4Cherry5 mostly include GFP-positive (green spot) molecules when turnover is rapid (wild-type cells, left), whereas some GFP/Cherry-positive molecules (green and red spots) could be detected when turnover is slow (krz mutant cells, right). The horizontal purple arrow indicates time.

Detecting Notch in acidic endosomes via the quenching of GFP at low pH

A second difference between GFP and Cherry is pH sensitivity. Indeed, the reported pKa (acid dissociation constant) of EGFP and Cherry are 6.0 and <4.5, respectively (Fig. 3 A; Shaner et al., 2005; Doherty et al., 2010). Indeed, GFP fluorescence reversibly responds within seconds to pH changes (Kneen et al., 1998; Taylor et al., 2011). Clearly, this property has in vivo relevance because early endosomes, late endosomes, and lysosomes have a pH in the 6.1–6.8, 4.8–6.0, and <4.5 range, respectively (Yamashiro and Maxfield, 1987; Huotari and Helenius, 2011), and this difference has been used to design reporters for acidic autophagic compartments (Kimura et al., 2007). The quenching of GFP at low pH therefore suggested that the GFP-negative, Cherry-positive intracellular dots might correspond to acidic intracellular vesicles.

Open in a separate window

Figure 3.

pH-dependent fluorescence of tagged Notch in endosomes. (A) pH-dependent emission profiles of GFP and Cherry (adapted from Doherty et al., 2010). a.u., arbitrary unit. (B and B′) intracellular NiGFP (GFP) remained associated with Cherry-Rab5 endosomes in living pupae. See also Video 1. (C–G) NiCherry (Cherry) did not colocalize with GFP-Rab5 (GFP in C; see also Video 3) nor YFP-Rab11 (YFP in D) but partly overlapped with GFP-Lamp1 (GFP in E) and YFP-Rab7 (YFP in F and G; see also Video 4). Snapshots of a video showing the tethering and fusion of NiCherry-positive endosomes with YFP-Rab7 endosomes (G; t is in minutes and seconds). (H and I) Live imaging of NiGFP4Cherry5 in living pupae in control (H) and Rbcn-3A^RNAi pupae. The silencing of Rbcn-3A led to the detection of GFP (green) and Cherry (red) fluorescence in enlarged Notch-containing endosomes in living notum cells. (J–L) Imaging of NiGFP4Cherry5 of living third instar larvae showing the pouch of wing imaginal discs in control (K) and chloroquine-fed (L) N^55e11 rescued larvae. In control larvae, GFP showed the apical cortex of wing pouch cells and Cherry revealed intracellular dots, particularly along the anterior wing margin, a region of high Delta-Notch signaling. Upon chloroquine treatment, the GFP and Cherry signals of NiGFP4Cherry5 colocalized in dots (L), and the GFP/Cherry signal intensity ratio was significantly increased (J). Error bars represent the standard error of the mean (n = 5). (M and M′) Apical distribution of NiGFP4Cherry5 (GFP [green] and Cherry [red]) was largely unaffected by the loss of lgd activity (M; mutant cells were marked by the loss of a nuclear RFP marker) in living genetically mosaic pupae. In contrast, enlarged high GFP and low Cherry endosomes were specifically detected basally in lgd mutant cells (M′). The number and size of the high Cherry endosomes was not detectably changed upon the loss of lgd activity (M′; this observation was confirmed using GFP as a clone marker; not depicted). The white lines indicate the borders of the mutant clone. Bars: (B–F, H, and M′) 5 µm; (G) 1 µm; (K) 25 µm.

To test this hypothesis, we studied the endosomal localization of NiGFP and NiCherry in living cells. Live imaging of NiGFP showed that the apical GFP-positive dots remained associated with early/sorting endosomes marked by Cherry-Rab5 and Cherry-Rab4 (Fig. 3, B and B′; Video 1; and Video 2). This indicated that the GFP fluorescence of NiGFP revealed a pool of Notch localizing at early/sorting endosomes. In contrast, the Cherry fluorescent dots of NiCherry did not colocalize with GFP-Rab5 endosomes (Fig. 3 C and Video 3; note that GFP-Rab5 largely colocalized with Cherry-Rab5 with the exception of strong Cherry-positive, GFP-negative dots that might correspond to acidic autophagosomes; Kimura et al., 2007). Also, NiCherry did not colocalize with YFP-Rab11 (Fig. 3 D). However, many Cherry-positive NiCherry endosomes were associated and/or overlapped with late endosomes marked by YFP-Rab7 (Fig. 3, F and G; and Video 4) and with late endosomes/lysosomes marked by GFP-Lamp1 (Fig. 3 E). Thus, endosomal GFP-fluorescent Notch was mostly detected at early/sorting endosomes, whereas Cherry-fluorescent Notch was seen, at least in part, at late acidic endosomes.

The aforementioned data suggested that the GFP signal of NiGFP4Cherry5 might be quenched in late acidic compartments. To test this hypothesis, the pH was increased in late endocytic compartments by disrupting the activity of the V-ATPase vacuolar proton pump using RNAi against the V-ATPase regulator Rabconnectin-3A (Rbcn-3A). Consistent with an earlier study (Yan et al., 2009), the silencing of Rbcn-3A led to the accumulation of NiGFP4Cherry5 in enlarged endosomes. Moreover, strong GFP and Cherry signals were detected in these endosomes in living pupae (Fig. 3, H and I). This indicated that the defective acidification of late endosomes prevented the quenching of GFP. To further test the quenching hypothesis, NiGFP4Cherry5 larvae were fed with chloroquine, an endosome acidification inhibitor that acts as a lysosomotropic weak base (Ohkuma and Poole, 1978; Vaccari and Bilder, 2005). This treatment led to strong GFP fluorescence and increased GFP/Cherry fluorescence ratio in wing disc cells of living third instar larvae that were directly imaged through the cuticle (Fig. 3, J–L; Nienhaus et al., 2012). We therefore conclude that the GFP fluorescence signal of NiGFP4Cherry5 is quenched in acidic late endosomes and that the low GFP, high Cherry signal seen in living cells identified Notch in late endosomes and possibly lysosomes.

Because GFP is cytosolic in the NiGFP4Cherry5 fusion receptor, quenching of GFP by acidic pH should depend on the transfer of Notch into the endosomal lumen of multivesicular bodies. To test this prediction, we studied the fluorescence pattern of NiGFP4Cherry5 in cells mutant for the lethal(2) giant discs (lgd) gene. Earlier studies showed that Notch accumulated at the surface of enlarged endosomes in lgd mutant cells because of its defective intraluminal transfer (Jaekel and Klein, 2006; Troost et al., 2012). Consistent with this, NiGFP4Cherry5 accumulated at enlarged vesicles in lgd mutant cells and remained GFP fluorescent (Fig. 3, M and M′). We conclude that the quenching of GFP required the intraluminal transfer of NiGFP4Cherry5, indicating that GFP is quenched in the lumen of acidic endosomes. Thus, the fluorescence properties of GFP and Cherry could be used to identify distinct pools of Notch in living cells.

Opposite apical–basal gradients of GFP and Cherry fluorescence

Both maturation of Cherry and endosome acidification are continuous processes. Thus, opposite gradients of GFP and Cherry fluorescence are expected to be seen at the level of the population of GFP/Cherry-tagged Notch traveling along the endocytic path from the plasma membrane to lysosomes. In polarized epithelial cells, high GFP/low Cherry Notch should be observed at apical early/sorting endosomes and low GFP/high Cherry Notch should be detected more basally in late endosomes (Fig. 4 A). To test this idea, we measured the ratio of the GFP and Cherry fluorescence intensities of endosomes of epithelial cells in the notum of living pupae. Endosomes were automatically and sequentially detected in the GFP and Cherry channels using a 3D wavelet transform Spot Detector plugin (de Chaumont et al., 2012). For each endosome, the GFP/Cherry ratio was calculated, and the mean ratio value for a given z position was plotted along the apical–basal axis. We found that apical endosomes had a high GFP/Cherry ratio and that this ratio decreased for more basal endosomes (Fig. 4 B). This observation was fully consistent with the known apical–basal gradient of endosome maturation, from early and apical endosomes to late and basal ones (Huotari and Helenius, 2011).

Open in a separate window

Figure 4.

Apical–basal gradient of endosomal GFP/Cherry-tagged Notch. (A) A schematic representation of the distribution of the NiGFP4Cherry5 fluorescence along the apical–basal axis in notum epithelial cells. The GFP fluorescence of NiGFP4Cherry5 is strong at the apical plasma membrane (top, green) and in early/sorting endosomes and weak in late basal endosomes (bottom, red) as a result of quenching. Conversely, the Cherry fluorescence is weak at the plasma membrane and at early apical endosomes as a result of slow maturation time and strong in late basal endosomes. (right) This results in opposite gradients of GFP and Cherry fluorescence along the apical–basal axis. (B) The GFP/Cherry mean intensity ratios measured in endosomes were plotted along the apical–basal axis of notum epithelial cells (apical, z = 0; n = 4 nota; GFP- and Cherry-positive endosomes were combined to calculate mean ratio values). Ratios were calculated from raw pixel intensity values in the GFP and Cherry channels without normalization.

Distinct pools of E-Cadherin (Cad) and Spdo are seen by GFP/Cherry tagging

The fluorescence properties of GFP and Cherry seen with NiGFP4Cherry5 are intrinsic to GFP and Cherry and could, therefore, be used more generally to study the subcellular distribution and trafficking of membrane proteins in living cells. To test this, we first studied the distribution of GFP- and Cherry-tagged versions of DE-Cad. To do so, we used functional CadGFP and CadCherry knock-in constructs (Huang et al., 2009). A high GFP/low Cherry signal was seen at adherens junctions, whereas a low GFP/high Cherry signal was observed at endosomes of CadGFP/CadCherry trans-heterozygous pupae (Fig. 5, A–A″). We interpret these data to suggest that recently synthesized GFP-positive/Cherry-negative E-Cad were mostly present at the plasma membrane, whereas GFP-negative/Cherry-positive E-Cad mostly localized into late acidic endosomes.

Open in a separate window

Figure 5.

Fluorescence patterns of GFP/Cherry-tagged E-Cad and Spdo. (A–A″) Fluorescence signals produced by CadGFP (green) and CadCherry (red) in living pupae showing that GFP mostly revealed the junctional pool of E-Cad, whereas the strongest Cherry signal was detected at endosomes. Note that the junctional pool of E-Cad was also weakly Cherry positive. (B) Domain structure of Spdo showing the four transmembrane segments (black) and the LHELL and NPAF trafficking signals (purple; Tong et al., 2010; Upadhyay et al., 2013). SpdoGFP1 was described previously (Couturier et al., 2013). The points of GFP (green), Cherry (red), and GFP-Cherry tandem dimer (green and red) insertion are indicated (in amino acids). The results of the genomic rescue experiments are shown on the right. (C) spdo mutant flies rescued by one copy of the SpdoCherry2GFP3 BAC. (D–D″) Live imaging of SpdoCherryGFP (GFP [green] and Cherry [red]) showing that GFP and Cherry revealed distinct pools. Bars: (A) 10 µm; (D) 2 µm.

We next examined the distribution of the four-pass transmembrane protein Spdo using two BAC-encoded fusion proteins, SpdoCherry2GFP3 and SpdoCherryGFP (Fig. 5, B and C). For both fusion proteins, the GFP fluorescence revealed a pool of Spdo localizing primarily at the plasma membrane and at endosomes, whereas Cherry was mostly detected at endosomes that partly overlapped with those detected by GFP (Fig. 5, D–D″; and Video 5). Again, these data were consistent with GFP marking recently synthesized proteins located at the plasma membrane and early/sorting endosomes and Cherry marking mostly proteins that have been trafficked into late acidic endosomes. Thus, we propose that the GFP/Cherry tagging approach used here for Notch and Spdo should be generally useful to study the trafficking of membrane proteins in living cells and tissues.

Equal segregation of endosomal Notch in dividing SOPs

Having established the properties of NiGFP4Cherry5, it became clear that our earlier analysis of Notch distribution in living cells that was based on the fluorescence of NiGFP did not cover the late endosomal pool of Notch. We therefore reexamined the distribution of Notch in endosomes in the context of asymmetrically dividing SOPs.

Notch receptors internalized in SOPs before mitosis have been proposed to localize into endosomes marked by the protein Sara and to be directionally trafficked into the pIIa daughter cell at cytokinesis (Coumailleau et al., 2009). This conclusion was based on the live imaging of internalized anti-Notch–Fab fragment complexes in cultured notum explants. However, whether endogenous Notch is directionally trafficked into pIIa has remained unclear. Although no sign of directional trafficking was observed using GFP-tagged Notch (Couturier et al., 2012), the quenching of GFP in acidic endosomes likely prevented the detection of many Notch-containing endosomes. To circumvent this caveat, we studied the distribution of NiGFP4Cherry5 in mitotic SOPs. Because neither GFP nor Cherry markers could be used to identify SOPs without interfering with the fluorescence signals of NiGFP4Cherry5, we used the recently engineered near-infrared fluorescent protein eqFP670 (Shcherbakova and Verkhusha, 2013) to mark SOPs. The eqFP670 sequence was fused to an NLS, and the resulting nlsFP670 gene was expressed under the control of a SOP-specific enhancer from the neur gene. This allowed us to monitor the distribution of GFP- and/or Cherry-positive endosomes in mitotic SOPs at cytokinesis using three-color live imaging (Fig. 6, A–B; the anterior pIIb and posterior pIIa cells were identified based on nlsFP670 expression and on their relative position; Gho and Schweisguth, 1998). First, similar numbers of Cherry-positive dots were found in pIIa and pIIb at t = 0 (early furrow ingression; in minutes) and t = 3 (complete ingression; Fig. 6 C). Second, no significant change in the number of Cherry-positive endosomes was observed between these two time points in both pIIa and pIIb (P > 0.05, paired t test). Also, as reported earlier for NiGFP (Couturier et al., 2012), very few GFP-positive endosomes (less than one per dividing SOP) were detected, and those appeared to distribute in both pIIa and pIIb. Although we cannot exclude that a subset of specialized endosomes containing Notch traffic directionally into pIIa at cytokinesis, our analysis showed that the bulk of Cherry-positive, Notch-containing endosomes are symmetrically segregated at cytokinesis.

Open in a separate window

Figure 6.

Equal segregation of Notch-containing endosomes in dividing SOPs. (A–B) Snapshots of a three-color video (GFP, Cherry, and nlsFP670, blue) showing the distribution of NiGFP4Cherry5 in mitotic SOPs (marked by nlsFP670) at two time points (t = 0, onset of cytokinesis, in A–A; t = 3 min in B–B). Top (apical) and bottom (basal) images show two confocal sections. Cherry-positive endosomes are shown with boxes (note that some endosomes were best detected at focal planes other than the one shown in A–B). Bar, 5 µm. (C) Box plot showing the number of Cherry-positive endosomes in pIIb and pIIa at t = 0 and t = 3 (n = 12; n is the number of SOPs; in this and other box plots, outliers are shown as small circles). No significant difference was observed in the number of Cherry-positive between pIIa and pIIb at t = 0 and t = 3 or between t = 0 and t = 3 in pIIb and pIIa. Box and whisker plots are uniform in their use of the box: the bottom and top of the box are the first and third quartiles, and the band inside the box is the second quartile (the median). The ends of the whiskers represent the lowest datum still within 1.5 interquartile range (IQR) of the lower quartile and the highest datum still within 1.5 IQR of the upper quartile (often called the Tukey box plot).

Transgenes

The attB-P[acman]-ApR Notch clone (Couturier et al., 2012) and the attB-P[acman]-Cmr CH322-119I20 (spdo) BACs (Couturier et al., 2013) were used to generate transgenic flies. These BACs were further modified using recombineering-mediated gap repair (Venken et al., 2006) using the following primers: For NiCherry, the Cherry sequence was introduced at position 5 using homology arms as described previously for NiGFP (Couturier et al., 2012). NiGFP4Cherry5 was obtained from NiCherry using notchgfp4 forward, 5′-ACGGGGGGAGTGTCGGGCGTGCCGGGTGTAggtatggtgagcaagggcga-3′; and notchgfp4 reverse, 5′-AGCTGCTTGTGCCGCTGAATTCGTTGGCGGaacaccaacacccttgtaca-3′. SpdoGFP2 and SpdoGFP3 were obtained as described previously for SpdoGFP (Couturier et al., 2013; noted here SpdoGFP1) using spdoigfp2 forward, 5′-CAGTTTGCCAGCTTGCCACCGATACTTAATggtgttggcatggtgagcaagggcg-3′; spdoigfp2 reverse, 5′-GTTGTTAATATTCCTGGGCTTGGGCGGCAAgccaacacccttgtacagctcgtcc-3′; spdoigfp3 forward, 5′-CAAAGCCAACAAGTTGGAGCGGTAGTAGGAgttggcatggtgagcaagggcgagg-3′; and spdoigfp3 reverse, 5′-ATTCTGGCGAAGTGGCAAGGTTGGAGGGACgccaacacccttgtacagctcgtcc-3′. SpdoCherry2GFP3 was obtained from SpdoGFP3 using spdocherryi2 forward, 5′-CAGTTTGCCAGCTTGCCACCGATACTTAATggtgttggcatggtgagcaagggcgaggagg-3′; and spdocherryi2 reverse, 5′-GTTGTTAATATTCCTGGGCTTGGGCGGCAAgccaacacccttgtacagctcgtccatgccg-3′. (Uppercase letters correspond to the homology arms for recombination with the BAC DNA; lowercase letters correspond to the sequence priming the PCR amplification reaction to produce the recombination donor DNA.) For SpdoCherryGFP, the sequence of the Cherry-sfGFP tandem dimer (provided by M. Knop, University of Heidelberg, Heidelberg, Germany; Khmelinskii et al., 2012) was codon optimized for Drosophila and synthesized in vitro and used as a template for PCR amplification using the following primers to perform recombineering: spdoCh-sfGfpi forward, 5′-CAGCAGCAGCAGCAGCAGCAACACCAACAGggggttggcatggtgagcaagggcg-3′; and spdoCh-sfGfpi1 reverse, 5′-CTGCTGAGCCAGGTGTTGCTGGTGTATCTGgccaacaccggatcccttgtac-3′. For neur-nlsFP670, the sequence of the eqFP670 gene (Shcherbakova and Verkhusha, 2013) was codon optimized and fused in frame downstream of an NLS from mouse CBP80, and the resulting gene was synthesized in vitro to generate the nlsFP670 gene. The latter was inserted into the Stinger-attB-pneur-GFP plasmid (provided by S. Aerts, KU Leuven, Leuven, Belgium; Aerts et al., 2010) to replace GFP.

All constructs were verified by sequencing the recombined regions before phiC31-mediated integration. The resulting transgenes were integrated at the following attP sites: M[p3×P3-RFP, attP]51D (NiCherry), PB[y⁺ attP-9A]VK19 at position 68D2 (NiGFP4Cherry5), P{CaryP}attP2 at position 68A4 (SpdoGFP2, SpdoGFP3, and SpdoCherry2GFP3), PB[y⁺ attP-9A]VK20 at position 99F8 (SpdoCherryGFP), and PB[y⁺ attP-3B]VK02 at position 28E7 (neur-nlsFP670). BAC injection was performed by BestGene, Inc.

Genomic rescue assays were performed as described previously (Couturier et al., 2012, 2013): Autosomal NiCherry and NiGFP4Cherry5 transgenes were combined with the X-linked N^55e11 mutation, and rescue was studied in males carrying one copy of the transgene. The SpdoGFP2, SpdoGFP3, SpdoCherry2GFP3 transgenes were recombined with the spdo^ZZ27 deletion, and rescue was tested in spdo^ZZ27/spdo^G104 flies.

Immunostaining, microscopy, and image analysis

Dissection of staged pupae and antibody staining were performed using standard procedures. In brief, nota were dissected from staged pupae using Vannas Micro-Scissors, fixed in paraformaldehyde (4% in PBS), and incubated in PBT (PBS with 0.1% Triton X-100). The following antibodies were used: goat anti-GFP (1:500; Abcam), rabbit anti-DsRed (1:200; Takara Bio Inc.), and mouse anti-NICD (1:100; C17 9C6; Developmental Studies Hybridoma Bank). Secondary antibodies were obtained from Jackson ImmunoResearch Laboratories, Inc. After washes in PBT, nota were mounted in 4% N-propyl-gallate and 80% glycerol.

Wing discs were imaged through the cuticle of living third instar larvae as previously described in Nienhaus et al. (2012). In brief, living larvae were placed in a drop of water deposited on a slide between two coverslips and then immobilized by a coverslip put on top on this montage. Imaging was performed using a microscope (DMRXA; Leica) equipped with a spinning disk (CSU-X1; Yokogawa Electric Corporation), a back-illuminated camera (QuantEM 512C; Photometrics), 491/561-nm lasers, and a 40× (HCX Plan Apochromat CS, NA 1.4) objective. Acquisition was performed using the MetaMorph software (Molecular Devices). The GFP/Cherry ratio shown in Fig. 3 J was calculated over a field of cells covering the wing pouch (n = 5 discs).

Live imaging of pupae was performed as described previously (Couturier et al., 2012). In brief, staged pupae were transferred onto a microscope slide with a double-sided tape between two spacers. The pupal case was removed using forceps, and a coverslip coated with Voltalef oil was deposited onto the spacers (Couturier and Schweisguth, 2014). Imaging was performed using a confocal microscope (LSM 780; Carl Zeiss) with a 63× (Plan Apochromat, NA 1.4 differential interference contrast M27) objective and 488 (GFP)-, 514 (YFP)-, 561 (RFP and Cherry)-, and 633 (eqFP670)-nm lasers. The GFP and Cherry signals were always acquired simultaneously.

GFP- and Cherry-positive endosomes were independently identified and defined using the 3D wavelet transform Spot Detector plugin under Icy (de Chaumont et al., 2012). Parameters were set at 30–80 (threshold scale 2) and 25–100 (minimum size, in pixels). Results were exported into Excel (Microsoft) for analysis. The absolute values of the GFP/Cherry ratios varied depending on laser intensity settings. Because the intensity of the He/Ne 488-nm laser was set manually, all ratio data shown in this study were acquired within the same confocal session.

Statistical significance was tested using paired t tests when data distribution followed a normal law (Shapiro test). In all other cases, a nonparametric sign test under R was used.

Western blots

Wing imaginal discs were dissected from 30 third instar larvae, and protein extracts were prepared in a 50 mM Tris, pH 8, 150 mM NaCl, and 1% NP-40 buffer. Protein concentration was determined using a spectrophotometer (NanoDrop1000; Thermo Fisher Scientific), and 20 µg protein was loaded per lane on 4–20% precast Mini-PROTEAN TGX gels for SDS-PAGE. Proteins were transferred onto 0.2-µm nitrocellulose membranes (Bio-Rad Laboratories). The following primary antibodies were used: rabbit anti-GFP (1:4,000; Molecular Probes), rabbit anti-DsRed (1:4,000), and mouse anti-NICD (C17 9C6; 1:2,000). HRP-coupled anti–mouse/rabbit antibodies (1:10,000; GE Healthcare) were detected using a chemiluminescent substrate (SuperSignal West Femto; Thermo Fisher Scientific).

We thank S. Aerts, R. Bodmer, R. DeLotto, S. De Renzis, S. Eaton, Y. Hong, T. Klein, J. Knoblich, M. Knop, H. Krämer, J. Skeath, the Bloomington Drosophila Stock Center, Drosophila Genetic Research Center, Drosophila Genomics Resource Center, and FlyBase for flies, antibodies, DNA, software, and other resources. We thank N. Vodovar who generated the NiCherry flies and F. De Chaumont and N. Chenouard for help with Icy. We thank H. Krämer, R. Le Borgne, and all laboratory members for advice and discussion.

This work was funded by grants from the Fondation pour la Recherche Médicale (DEQ20100318284) and Agence Nationale pour la Recherche (Labex Revive).

The authors declare no competing financial interests.