The glue produced by Drosophila melanogaster for pupa adhesion is universal

Natural adhesives have a very important significance for the biology of organisms and are a great material for innovation of biologically inspired technical adhesives. The most studied bioadhesives are from marine organisms – mussels and barnacles (Power et al., 2010) – and they are now used in a variety of biomimetic applications such as surgical sealants to repair tissues or synthetic polymer coatings (Lee et al., 2011). Bioadhesives in insects are less studied although they are extremely diverse (Gorb, 2001; Graham, 2008; Li et al., 2008; Betz, 2010). Some insects produce glue to secure their eggs or cocoon (Betz, 2010). A few studies have been carried out on egg glue composition and particularly on the glue adhesive strength in different insect species. For example, in Opodiphthera moths, females secrete a viscous fluid from their accessory reproductive gland that sticks their eggs to each other and to the substratum. This hydrogel-type glue is highly elastic and mainly made of proteins. The shear strength reaches 1–2 MPa on wood, which could potentially be enough for some industrial applications (Li et al., 2008). Furthermore, these eggs and eggs from other species such as Crioceris asparagi have the ability to attach to plants covered with crystalline epicuticular waxes known to be non-adhesive (Voigt and Gorb, 2010). Similar properties have been revealed in the egg glue of codling moths (Al Bitar et al., 2012, 2014). Thus, investigating the glue of various insect species appears to be a great strategy to find novel universal and substrate-specific adhesives. Surprisingly, although Drosophila melanogaster is a model organism that is extensively used in laboratories, little is known about pupal adhesion in this species.

Drosophila melanogaster larvae secrete a glue from their salivary glands right before pupariation (Fraenkel and Brookes, 1953). After expectoration, following larval peristaltic movements, the secreted liquid spreads between the body and the substrate and dries within a few minutes (Beňová-Liszeková et al., 2019). This glue allows the animal to stay firmly attached to an external surface for several days, until the adult fly emerges from the pupal case, while the pupal case remains attached to the substrate. The pupa of other Brachycera fly species is also often attached to a substrate during metamorphosis (Fraenkel and Brookes, 1953).

In the wild, Drosophila pupae have been reported to adhere to a great variety of substrates, from fruits to beer bottles (Vouidibio et al., 1989), and in diverse environments. Particularly, pupae have been observed on dry parts of various fruits, some species have also been found fixed to wood, wet and rotten parts of fruits, deep in the soil and to one another (Grossfield, 1978; Sokolowski, 1985; Vandal et al., 2008; Beltramí et al., 2010; Pino et al., 2014). Pupal attachment might be important for not being taken away by predators, to resist environmental conditions (wind or rain) or to help adult flies emerge from the pupal case after metamorphosis (Da Lage et al., 2019).

In D. melanogaster (Korge, 1975, 1977) and other Drosophila species [D. virilis (Kress, 1982), D. natusa (Ramesh and Kalisch, 1988) and D. gibberosa (Shirk et al., 1988)], the glue is composed of a small number of proteins called salivary gland secreted (Sgs) proteins. These proteins present repeated motifs and glycosylations, which are commonly found in adhesive proteins (Graham, 2008; Betz, 2010). Some of the Drosophila glue proteins are rich in cysteine, like many marine adhesive proteins, and this property could be important to maintaining a secondary structure by forming disulfide bonds (Graham, 2008). Other Drosophila glue proteins are rich in serine and threonine and highly O-glycosylated, suggesting that they may interact with water to hydrate or dehydrate the glue (Farkaš, 2016). The Sgs genes have evolved rapidly between species and within species (Korge, 1975, 1977; Beckendorf and Kafatos, 1976; Da Lage et al., 2019).

Here, we investigated the adhesive properties of the glue of D. melanogaster on different substrates. First, we analyzed the contact region between the pupa and the surface to which it is attached. Then, we designed a pull-off force measurement set-up. We assessed the force required to detach the pupa from different substrates, with different hydrophilic and charge properties or with different roughnesses, and analyzed the way in which rupture occurs.

Drosophila melanogaster Meigen 1830 flies were cultured in plastic vials on standard medium [1 liter: 62.5 g yeast, 62.5 g cornmeal, 10.0 g agar-agar, 20.0 g glucose monohydrate, 30.0 g molasses, 30.0 g sugar beet syrup, 10.0 ml propionic acid (10%), nipagin (10%)] at room temperature. The stock Canton S (gift from Roger Karess, Institut Jacques Monod, CNRS, Université de Paris) was used for force measurement assays and SEM microscopy. The stock w[*]; P{w[+m*]=Sgs3-GFP}2 (FBti0016953; Bloomington Drosophila stock 5884; Biyasheva et al., 2001) was used for confocal microscopy.

Third instar wandering larvae were washed in PBS to remove traces of food and microorganisms from their surface, put in empty Petri dishes with a paintbrush and kept in a high humidity atmosphere at room temperature in a closed plastic box (15×10×5 cm) containing wet cotton. When larvae stopped moving, they were transferred on the substrate of interest with soft forceps, kept at high humidity as mentioned above and left to pupariate. Five to 24 h after transfer to the substrate of interest, the substrate with pupae attached to it was placed in room humidity for 1 h to allow the glue to dry completely, and then the pupae were used for adhesion assays. Pupae not used for assays were weighed individually using a Mettler Toledo AG203 (DeltaRange^®, Gießen, Germany). Temperature, humidity and atmospheric pressure were monitored daily.

Two types of non-coated microscopic glass slides were used to measure adhesion on glass: Menzel Superfrost microscope glass slide from Thermo Fisher Scientific (AGAB000080) and microscopic glass slide from Roth (0656.1). Atmospheric plasma-treated glass slides were prepared using a PlasmaBeam (Diener electronic GmbH, Ebhausen, Germany) on Roth glass slides for 1 min. To prepare poly-l-lysine–polyethylene glycol-coated (PLL–PEG-coated) glass slides, Roth glass slides were cleaned with a plasma cleaner then coated with 0.1 mg ml⁻¹ non-biofunctionalized PEG side-chains (methoxy-terminated) (SuSos). Poly-l-lysine-coated (PLL-coated) glass slides from Thermo Fisher Scientific (J2800AMNZ) and pieces of Teflon (Polytetrafluorethylene, Kelux, Geldern, Germany) were also tested.

First, a silicone cast was made by pouring polyvinylsiloxane (light body Affinis, Coltène/Whaledent GmbH+Co. KG, Langenau, Germany) over a cleaned Roth glass slide [to create the smooth resin, roughness average (R_a)=80 nm (Salerno et al., 2018)] or over glass slides covered by polished papers with different grain sizes: 1 and 9 μm [R_a=0.54 and 2.47 µm measured over 1400×1050 µm area using a NewView 6k white light interferometer (Zygo, Middlefield, CT, USA); both with FiberMet Abrasive disks, Buehler] (Salerno et al., 2018), and P1000 and P80 [R_a=3.94 and 40.40 µm measured over the same area using a VR 3100 3D profilometer (Keyence, Neu-Isenburg, Germany); polishing papers, Bahaus]. After a couple of minutes, the slide and the polished paper were removed and the edges of the polyvinylsiloxane cast were made higher. Then, Spurr epoxy resin (Spurr, 1969) was poured into the polyvinylsiloxane cast and polymerized at 65°C overnight in a Memmert U 15 oven (Schwabach, Germany). Resin was then allowed to cool down for a couple of hours before unmolding. Each resin was used for several assays.

Water contact angles on different substrate surfaces were measured using a contact angle measurement device OCA20 (DataPhysics Instruments, Filderstadt, Germany). A 2-µl water droplet was deposited on a substrate, then an image of the droplet was taken after 5 s, and contact angles were determined from the fit of the droplet’s shape with a sphere using SCA 202 software (DataPhysics Instruments, Filderstadt, Germany). All the contact angle measurements were done before the pull-off force assays except for PLL–PEG-coated glass.

To measure pupal adhesion, a force transducer (100 g capacity, FORT100, World Precision Instruments, Sarasota, FL, USA) mounted on a motorized 3D micromanipulator (DC3001R, World Precision Instruments) was used (Fig. 1). The substrate was horizontally clamped to a lift table (Carl Roth, Karlsruhe, Germany). A piece of aluminium platelet of 1.0×2.0×0.2 mm attached to the force transducer by a metal wire (0.1 mm) was covered by a piece of double-sided adhesive tape (Tesa, extra strong, 05681-00018). A new piece of tape was glued for each measurement. Only pupae attached on their ventral side and that did not contact other pupae were used for measurements. The tape on the aluminium platelet was brought into contact with the dorsal part of the pupa by adjusting manually the height of the lift table. The force transducer was then moved upwards with a velocity of 200 μm s⁻¹ until detachment of the pupa. The signal from the force transducer was amplified using Transbridge TBM4M and digitized using Lab-Trax-4 data acquisition hardware (World Precision Instruments). Three phases could be distinguished on force curves recorded using LabScribe v3 (iWorks, Dover, NH, USA) (Fig. 2A). During the initial phase, the pupa was not attached to the tape and force with no pupa weight was measured. The pulling phase started when the pupa came into contact with the tape. During this phase, the pupa was stretched until it detached from the substrate (maximal force). Finally, during the resting phase, the force went down to a basal value including pupal weight. Overall, the adhesion strength we measured corresponds to the weakest interface: adhesion of the glue to the substrate, cohesion of the glue, adhesion of the glue to the pupal case, cohesion of the pupal case (when the cuticle broke) or the interface between the tape and the pupal case (when the tape detached from the pupa). The set-up was calibrated before the first assay with a standard weight. Pull-off force corresponding to the force at pupa detachment was calculated as the difference between the maximal pulling force and the mean force at the initial phase (offset). The force corresponding to the adjustment of the lift table until pupa glued to the tape between the initial phase and the pulling phase was not recorded.

A total of 328 pupae were measured using our pull-off force measurement setup. In 59 (18%) cases, the pupal case cracked on the dorsal side between the head and the thorax during the pulling process. For 36 (11%) of them, the pupal case covering the ventral part of the head detached from the rest of the pupal case and remained glued to the substrate at the end of the trial, while the complete pupa body was detached together with the posterior part of the pupal case. For the 23 (7%) other cases, the whole animal together with the broken pupal case was detached from the substrate. These 59 (18%) cases were excluded from further analysis. In cases where the animal remained intact, we never observed failure of the pupal case, such as pupal case material remaining attached to the tape or the glue print. Furthermore, two or more trials were sometimes necessary to detach a pupa, when the connection between the double-sided tape and the pupa failed. If the maximal force on the last trial was higher than the previous ones (Fig. 2E), the last trial was used for analysis unless the pupal case was damaged. If the maximal force was higher in one of the first trials, the measurement was excluded, suggesting that the pupa has been partly detached during the first trials (n=23, 7%; Fig. 2F). Eleven other cases (3%) were excluded from the analysis: imperfections in the substrates (n=3), pupa not attached ventrally (n=6), pupa not attached, with no glue (n=1) and tape glued to the substrate (n=1). In total, 93 cases (28%) were excluded.

We then organized the remaining 235 cases (100%) into four groups. The first group gathers the trials for which the whole pupa detached at once and the tape stayed well fixed to the pupa (n=56, 24%; Fig. 2A). The second group gathers the trials for which the piece of tape started to peel from the pupal case surface during the pulling process and the pupa detached at once. For this group, the force–time curve had a negative derivative at some point during the pulling phase (n=106, 45%; Fig. 2B). The third group gathers the trials for which the whole pupa did not detach at once but the head part stayed attached longer than the rest of the body. In such cases, two peaks were observed on the force–time curve (n=29, 12%; Fig. 2C): the first one corresponds to the main body part detachment and the second one to the head part detachment. The first peak always displayed the maximal force. Finally, the fourth group gathers the trials for which tape started to peel before pupa detachment and for which the whole pupae did not detach at once. For this group, the force–time curve had a negative derivative at some point during the pulling phase and a second peak was observed after the peak corresponding to the maximal force (n=44, 19%; Fig. 2D). We tested whether these four groups having different detachment behavior have an effect on the resulting adhesion force. We performed a two-way ANOVA considering force as the dependent variable. We found that substrates but not groups had a significant impact on pupa adhesion force and that there was no significant interaction between groups and substrates (substrate: F=12.86, P<2×10^–16, group and group×substrate interaction: P<0.05).

After pupal detachment, images of glue prints remaining on glass substrates (100 cases) were taken with a Keyence VR 3100 microscope at ×40 or ×80. Prints areas were measured manually using ImageJ (1.50d, java 1.8.0_212, 64 bit) (Schneider et al., 2012). For one case, the print was altered so that the areas could not be measured. For 11 cases, the red area could not be visualized and was not measured.

One pupa detached from the Menzel non-coated glass slide was mounted on aluminium stubs using double-sided carbon conductive tape (Plano, Wetzlar, Germany). The sample was frozen in liquid nitrogen, sputter-coated with gold–palladium (8 nm thickness) at −140°C, and examined in a cryo-SEM (Hitachi S-4800; Hitachi Ltd, Tokyo, Japan) at −120°C and an accelerating voltage of 3 kV.

To visualize the thickness of the glue at the pupa–substrate interface, w[*]; P{w[+m*]=Sgs3-GFP}2 larvae were let to pupate as described for measurement assays but on circle cover slips (Thermo Fisher Scientific, 32 mm diameter, 0.17±0.01 mm thickness). Untreated (living) samples were directly imaged with a Leica SP5 confocal microscope with a Plan-Apochromat 40×/1.25 NA objective. Serial optical sections were made every 0.29 μm. Maximal Z projections and Z sections were computed using ImageJ (1.52a, java 1.8.0_112, 64 bit) (Schneider et al., 2012). For the picture of the full attached pupa, several images were taken with an Olympus IX83 inverted microscope using a U-Plan FLN 4×/0.13 NA objective and stitched manually.

Statistical differences in pull-off forces between substrates, groups and interactions were tested by two-way ANOVA using the aov function in R version 3.4.4 (https://www.r-project.org/). Statistical differences in pull-off force between substrates were tested by one-way ANOVA with the same R aov function followed by multiple pairwise comparison tests using Tukey test with the TukeyHSD function in R. Pearson correlation between pull-off force and contact areas were tested using the lm function in R, as were the effects of humidity, temperature, pressure and age.

Pupae of D. melanogaster attach naturally to substrates on their ventral side via a layer of glue which forms an oval-shaped patch visible on glass slides of approximately 2 mm length and 0.5 mm width (Fig. 3A,B). Glue near the posterior part of the animal usually forms a bigger plug that can spread on the substrate (Fig. 3B, arrow). We performed SEM of pupae detached from non-coated glass slides and found that only thin traces of glue remain on the former contact area (Fig. 3C–G, arrow in F). The glue layer covering the pupal case appears to be organized in thin layers (Fig. 3F,G). Air bubbles are observed between the layers and the glue does not fill all the asperities of the cuticle surface (Fig. 3I,J).

Using confocal microscopy of attached pupae which produce fluorescent glue, we found that the thickness of the glue varies within a single animal from 0 to about 20 μm due to the irregularities of the pupal case and its barrel shape (Fig. 4). We note also that this fluorescent glue could have slightly different properties than wild-type glue.

We performed pull-off force measurements (Table S1) using pupae naturally attached to 11 different substrates: non-coated glass (from Roth and from Menzel), PLL-coated glass, PLL–PEG-coated glass, oxygen-activated glass, Teflon and Spurr epoxy resin with five different roughnesses. We measured the contact angle of these 11 substrates (Table S2). Contact angle values ranged from 11 deg (highly hydrophilic substrate) to 112 deg (highly hydrophobic substrate). In total, 328 pupae were measured. We excluded 93 cases for which adhesion measures were not reliable, for instance when the pupal case was damaged during the assay or when several trials with the same pupa disrupted its adhesion (see Materials and Methods).

Medians of the pull-off forces on glass substrates (non-coated, PLL-coated, PLL–PEG-coated and oxygen-activated glass) ranged from 184 mN (oxygen-activated glass, s.d.=78) to 229 mN (PLL–PEG-coated glass, s.d.=122), whereas for Teflon, the pull-off force was 42 mN (s.d.=20) (Fig. 5). Only the pull-off force on Teflon was significantly different from the forces obtained on other substrates (one-way ANOVA F=12.92, P<2×10^–16, followed by all pairwise comparisons, Tukey’s test, P<0.001 for all comparisons with Teflon). Similarly, pull-off force medians on resin with different roughnesses ranged from 151 mN (P80, s.d.=82) to 269 mN (1MIC, s.d.=105) and no statistical differences were found (from the same ANOVA and Tukey tests). During our assays, the ranges for humidity, temperature, pressure and age of pupa were the following: 34.4–58.8%, 23.5–27.9°C, 1005–1026 mb, 3.5–23 h after deposition of the wandering L3 larva on the substrate. No effect of humidity, temperature, pressure or age of pupa was found (multiple linear regression with force as the dependent variable and substrate, humidity, pressure and age as independent variables).

In conclusion, we found that the pull-off force is independent of substrate surface chemistry (except Teflon) and of roughness in a wide range (80 nm–40.40 µm).

After pull-off force measurements, glue prints were analyzed for the five glass substrate types. For these substrates, three types of prints can be defined: (1) the glue fully remained on the substrate (Fig. 6A), (2) the glue partly detached from the substrate (Fig. 6B) or (3) the glue went off completely with the pupa (Fig. 6C). Most of the time, the glue was partly detached from the substrate (n=95/124). For the five glass substrate types, print type (but not substrate type) had a significant impact on pupa adhesion (two-way ANOVA, print types: F=5.6, P=0.0013). When the glue completely went off with the pupa, the corresponding pull-off force was significantly lower than the force obtained when glue was partly detached (134 mN, s.d.=62 compared with 231 mN, s.d.=102, Tukey test P<0.05; Fig. 6D). On Teflon, the glue always went off with the pupa (n=33/33). Furthermore, from the glue print, three distinct areas can be defined. The pupa–substrate interface area corresponds to the area where the pupa was in contact with the substrate (green outline in Fig. 6E). The glue–substrate interface area (red outline in Fig. 6E) corresponds to the total surface of the substrate covered by the glue. After pupa detachment, a third area where the glue went away with the pupa can be defined (black outline in Fig. 6E). For all cases (as in Fig. 6A,B) where the glue did not go off completely with the pupa, the three areas were measured from the print images of the five glass substrate types. A relatively high variation in glue–substrate interface area was observed (from 1.3 to 4.6 mm²) compared with the pupa–substrate area, which was relatively invariant, around 0.2 to 1.5 mm² (Fig. 6F). No difference was found in areas between substrates (one-way ANOVA for each area, P>0.05; Fig. 6F). No correlation was found between pull-off force and any of the three areas when combining the five substrates or when testing each substrate individually (linear regression for each area, P>0.05; Fig. S1).

Drosophila glue spreads over an external substrate and sticks the pupa's ventral part to the substrate, forming an oval-shaped interface (Fig. 3B). From SEM images, we observed flat holes that seem to correspond to air bubbles that were trapped between different layers of glue. Several layers may have been deposited successively on the surface of the pupa by the back-and-forth movements of the larvae during expectoration (Fraenkel and Brookes, 1953). The presence of bubbles and the fact that the glue does not fill all the asperities of the cuticles suggest that the glue is relatively viscous when it comes out. It is known that the glue is produced by the salivary glands and is secreted by the mouth (Fraenkel and Brookes, 1953). However, on the glue prints we sometimes observed a plug around the posterior tip of the larva (Fig. 3B). Recently, Beňová-Liszeková et al. (2019) reported that a few larvae did not empty their guts entirely before pupariation. One possibility is that this plug might be a posterior secretion from the digestive tube, which mixes with the anteriorly secreted glue.

During our pull-off force assays, the piece of tape used to stick the pupa in order to pull it peeled out from the pupal case in 153 of the 235 analysed cases, meaning that the adhesion strength between the double-sided tape and the pupal case is close to the adhesion strength between the glue and the pupa. In our experiments, the tape thus adheres to the pupal case just enough for pupa glue measurements. Furthermore, in 76 of the 235 cases, the posterior part of the animal first detached from the substrate before the head detached. In these cases, the pull-off force was probably not applied exactly at the center of the pupa but more posteriorly owing to the asymmetrical shape of the pupa. Furthermore, the pupal case is more fragile around the puparial opercular seam, a seam that runs across the front of the puparium and extends along the sides, and that will break when the adult emerges (Tyler, 1994). In addition, when using forceps to extract the animal from its pupal case, the pupal case generally splits into annular strips (Held, 1992). The zone of breakage that we observed on the pupal case in our pull-off assays therefore corresponds to the most fragile region of the pupal case.

In any case, no significant difference was found between cases for which pupae detached at once and cases where the posterior part detached before the head part. This suggests that stronger head adhesion does not affect the maximal pull-off force.

Pull-off forces required to detach pupae from substrates other than Teflon ranged from 151 mN (P80, s.d.=20) to 269 mN (1MIC, s.d.=105). Using the pupa–substrate interface (green outline in Fig. 6E) as a measure of contact area between the glue and the substrate, we found a median contact area of 1.10 mm², which leads to an adhesive strength ranging from 137 to 244 kPa. Few similar studies have been carried out on insect egg adhesive strength. Two of them reported higher values [1–2 MPa for Opodiphthera eggs on wood (Li et al., 2008), 4.3–12.2 MPa for whitefly eggs on rose leaves (Voigt et al., 2019)], one study found equivalent values (38.8–271.3 kPa for beetle C. asparagi on plant surface; Voigt and Gorb, 2010) and another lower adhesive strength (13.9–97.8 kPa for the moth C. pomonella eggs on fruits; Al Bitar et al., 2014). To our knowledge, our study is the first one to report adhesion strength measurements for Diptera pupa.

No statistical differences in pull-off forces were found between the different substrates except for Teflon (median pull-off force is 42 mN, s.d.=20), suggesting that the glue can stick to hydrophobic as well as hydrophilic substrates. We hypothesize that proteins of the glue are polarizable, which allows the glue to stick strongly to polarizable substrates that are differently charged and not to Teflon, which is not polarizable. Indeed, Sgs proteins are highly charged. They are mostly composed of positively charged amino acids and they are highly glycosylated with negatively charged sugars (Beckendorf and Kafatos, 1976; Korge, 1977).

No differences were found on resin substrates with different roughnesses. We expected that pull-off force would increase with roughness as contact area on rough substrate is bigger. Furthermore, we did not find a significant correlation between force and contact area for any of the substrates used. Both observations can be explained by a possible critical crack initiation stress between the pupal case and the glue. Indeed, on the glue prints on glass substrate types, we observed that the glue from the pupa–substrate contact area is partly detached, suggesting that rupture occurs between the glue and the pupal case. A similar phenomenon may occur on resin substrates, whereas on Teflon the glue completely goes off from the substrate, suggesting that the rupture occurs between the glue and the Teflon. Thus, in all the tested substrates except Teflon, the bond between the glue and the different substrates appears to be stronger than the bond between the glue and the pupal case. It is thus possible that our adhesion tests measure the adhesive force of the bond between the glue and the pupal case, which would explain why we observed the same adhesion strength for all substrates except Teflon. As an alternative explanation for the same adhesiveness measured on all tested substrates except Teflon, it is possible that the animal modulates the amount of glue it produces, or the way it spreads the glue over the substrate and its body. It would be interesting to perform adhesion measurement on isolated glue. However, it is challenging to collect the glue as it is secreted in very small volumes and it polymerizes within 3–5 min after expectoration (Beňová-Liszeková et al., 2019).

Pull-off force values were rather variable between trials for a given substrate (for example, from 19 to 451 mN on Menzel non-coated glass). Such a wide range of values could be due to individual variability. Pupal weight could not be deduced accurately by subtracting the final force from the initial force because pupal weight was within the noise of force values. However, individuals from the same population as the pupae used for trials were weighed instead. We found that pupal mass averaged 1.4±0.3 mg (mean±s.e.m.) (n=24). The corresponding weight variation (13.7±2.9 µN) is negligible compared to the force variation observed within substrates. Surprisingly, considering this average weight and an average force of 217 mN (mean of all trials kept for analysis except trials on Teflon), the glue holds approximately 15,500 times the weight of a pupa.

For the first time, we report here measurements of Drosophila pupal adhesion strength. We present a pull-off force test to measure pupa adhesion, which could be used in the future to explore pupa adhesion in various strains and species of fly. With the powerful genetic tools available in D. melanogaster, we now plan to assess the adhesive function of the glue genes. The use of D. melanogaster as a model organism for the study of bioadhesives is very promising as it makes it possible and easy to manipulate the composition of the glue.

We acknowledge T. Roeder and C. Sandberg (Kiel University, Germany) for help with fly culture, S. De Beco (Université de Paris, Paris) for help with PLL–PEG coating and L. Corté (Mines-ParisTech, ESPCI, Paris) for discussions. We also thank the ImagoSeine facility, member of the France BioImaging infrastructure supported by the French National Research Agency (ANR-10-INSB-04, «Investments for the future») and N. Moisan for technical assistance with confocal microscopy. One stock obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) was used in this study.