Introduction

The reiterative tube formation and its ramification in our vasculatures, airways and lungs generate the pulmonary-vascular network to efficiently supply oxygen to the whole body. The airway tubes allow the airflow to alveolar structures, where blood cells inside the fine vascular tubes exchange carbon dioxide with oxygen to deliver it to the body (Herriges and Morrisey 2014; Potente and Makinen 2017; Kishimoto and Morimoto 2021). Surprisingly, a similar pulmonary-blood cell connection is also found in the fruit fly Drosophila. There, reiterative ramification of the epithelial tubes (the tracheal system) allows the airflow inside the body that directly delivers oxygen to the target cells (Manning and Krasnow 1993; Samakovlis et al. 1996a; Hu and Castelli-Gair 1999; Ghabrial et al. 2003), while the blood cells in the open circulatory system transiently attach to the fine airways to boost oxygen transport (Shin et al. 2024).

The lungs and the Drosophila airways both derive from 2D- primordial cell fields (Romero et al. 2025). Lung formation initiates as tube evaginations from the foregut epithelium (Herriges and Morrisey 2014; Kishimoto and Morimoto 2021). Distal ramification of these buds and further differentiation into distal airways and alveoli involve FGFR activation by extrinsic FGFs (Min et al. 1998; Weinstein et al. 1998; Arman et al. 1999; Sekine et al. 1999; Brownfield et al. 2022; Jones et al. 2022; Sountoulidis et al. 2023). Similarly, Drosophila airway formation initiates from cell primordia specified on the ectodermal plane, each invaginating to form a primitive cavity (Perrimon et al. 1991; Manning and Krasnow 1993; Samakovlis et al. 1996a; Hu and Castelli-Gair 1999). Extrinsic FGF/Branchless (Bnl) activates FGFR/Breathless (Btl) to guide branch ramification and network connection (Klambt et al. 1992; Guillemin et al. 1996; Samakovlis et al. 1996a; Samakovlis et al. 1996b; Sutherland et al. 1996) (Figure 1A). Accordingly, highly ramified fine tubes are generated around each branch terminus that serve for oxygen transport.

Figure 1.

Figure 1 figure supplement 1.

trh is the earliest TF gene marking the Drosophila airway primordia and the mature airways (Isaac and Andrew 1996; Wilk et al. 1996). trh has been regarded as the master TF of the Drosophila airways because tubes as well as differentiation markers are not detected in trh mutants at later stages (Isaac and Andrew 1996; Wilk et al. 1996; Brodu and Casanova 2006; Sotillos et al. 2010; Chung et al. 2011; Matsuda et al. 2015b). However, in trh null mutants, expression of rhomboid (rho) (Bier et al. 1990), a protease activating the EGFR ligand Spitz (Schweitzer et al. 1995) initiates in the primordia (Ogura et al. 2018) but its propagation (Matsuda et al. 2015b) fails (Ogura et al. 2018). Consistent with the role of rho in promoting primordial invagination (Brodu and Casanova 2006; Nishimura et al. 2007), trh mutants initiate primordia invagination but neither sustained invagination nor maintained tube structures (Kondo and Hayashi 2019). The initiation of invagination and the early localized rho expression in trh mutants suggest additional regulators, other than trh that are critically responsible for gene expression in airway primordia and for tubulogenesis.

We had previously shown that the proximo-distal axis of the Drosophila airway tubes is generated from the centro-peripheral patterning of the 2D primordia (Matsuda et al. 2015b). One set of genes, the distalizing factors, hh, vvl, rho and btl/FGFR cooperate to realize the distal gene expression program at the expense of the proximal one, whereas another set, the proximal factors including grn, negative regulators of EGFR signaling and hh signaling realize the proximal gene expression program (Figure 1A). The two programs establish distinct proximal and distal cellular domains in the branching network (Matsuda et al. 2015b). Two of the distalizing factors, rho and btl/FGFR promote primordia invagination but are not essential for tubulogenesis (Brodu and Casanova 2006; Matsuda et al. 2015b), indicating that essential regulators of primordial morphogenesis have been elusive.

Here, we report that in the absence of the remaining two distalizing factors hh and vvl, the trachea primodia cells frequently fail invagination and are only detected on the 2D plane. Therefore, extrinsic Hh and intrinsic Vvl represent the missing factors that cooperatively promote primordia invagination. Although trh, vvl or grn alone cannot define the airway progenitors, we propose that combinations of the three TFs can intrinsically define the airway progenitors, considering their subsequent roles on gene expression and morphogenesis. Even in the simple Drosophila airways, combinations of multiple factors induce the organ progenitor field and subsequent tubulogenesis to cope with physiological stress of respiration in a terrestrial environment.

Results and Discussion

The distalizing factors hh and vvl drive airway primordia invagination independent of trh

Among the distalizing factors, hh, vvl, rho and btl/FGFR (Figure 1A), hh vvl double mutants show more extensive distal-to-proximal gene expression conversion than the loss of both rho and bnl/FGF-btl/FGFR signaling (Matsuda et al. 2015b). As EGFR (also known as torpedo/top or faint little ball) and btl/FGFR signaling also orchestrate the distal morphogenetic processes, including primordial cell invagination (Llimargas and Casanova 1999; Brodu and Casanova 2006; Nishimura et al. 2007; Kondo and Hayashi 2013) and subsequent branching (Glazer and Shilo 1991; Klambt et al. 1992; Sutherland et al. 1996; Llimargas and Casanova 1999; Matsuda et al. 2015b), hh and vvl are expected to control the distal morphogenetic program as well. In the absence of hh and vvl, rho expression is detected at early stages sporadically and weakly (Figure 1-figure supplement 1A) and btl expression initiates in the primordia, but both fade away shortly thereafter (Matsuda et al. 2015b). Despite the initial expression of rho and btl/FGFR in hh vvl mutants, hh vvl embryos show severer branching phenotypes than the rho btl mutants (Matsuda et al. 2015b).

We thus investigated potential invagination defects of trh positive cells in hh vvl double mutants and detected aberrant expression of both trh RNA and trh-LacZ persisting on the epidermal plane at stage 12 (Figure 1-figure supplement 1B). However, massive ectodermal apoptosis precluded analysis at later stages. In H99 hh vvl triple mutants, where apoptosis is suppressed by deletion of major pro-apoptotic genes (White et al. 1994), we often detected failure in invagination, where all the Trh+ cells remained in a 2D plane at stages 12-13 (Figure 1A-B). By stage 15, small cavities are detected in most metameres of H99 hh vvl mutants, suggesting that proximal cell types invaginate by a mechanism independent of distal cell types differentiation. Taken together, the two distalizing factors hh and vvl define the central/distal cell identity, controlling both gene expression and cell invagination. It follows that 3 intrinsic TFs, Trh, Vvl and Grn cover the major functional aspects of airway progenitor differentiation.

dpp/BMP specifies the airway progenitors along the D-V axis

We further investigated upstream determinants of the primordial field defined by the expression of these 3 TFs. We first allocated their expression domains along the D-V axis of the embryo. The surface of the embryonic trunk at mid embryogenesis is largely divided into 3 sectors along the D-V axis, amnioserosa, the dorsal and the ventral ectoderm (Figure 2A) (Wharton et al. 1993; von Ohlen and Doe 2000). Both the dorsal and the ventral ectoderm are further subdivided into 3 parts along the D-V axis, medial, intermediate and lateral (Figure 2A) (von Ohlen and Doe 2000). The dorsal intermediate column marked with caupolican (caup)/araucan (ara) spans several cells located dorsally to the trh-LacZ positive spiracular branches (Figure 2-figure supplement 1A), whereas the Grn-GFP expression domain, which also marks the most proximal spiracular branches extends ventrally from the dorso-lateral ectoderm (Figure 2- figure supplement 1B). On the other hand, the ventral limits of the initial trh expression at stage 10 abut Dichaete-LacZ (D-LacZ) expressing cells in the intermediate and medial columns of the ventral ectoderm (Figure 2B-C) (Zhao and Skeath 2002). Thus, the initial trh expression straddles from the ventro-lateral column to halfway to the dorso-lateral column.

Figure 2.

Figure 2 figure supplement 1.

Compared to trh expression, vvl expression in the airway primordia is more restricted (Figure 2D). vvl expression straddles the boundary of the dorsal and the ventral ectoderm demarcated by Drop (Dr)/muscle homeobox (msh) whereas the ventral limit of vvl expression is far from the ventro-intermediate column (Figure 2D) (von Ohlen and Doe 2000). Consistently, the ventro-lateral ectoderm enhancer of Dr/msh (Pfeiffer et al. 2008; Manning et al. 2012) marks the ventral parts of the invaginated airway progenitors (Figure 2E).

Medium level of Dpp/BMP signaling positively regulates the expression of vvl and grn (Matsuda et al. 2015b) but its function of trh regulation is only partially investigated. In the absence of Dpp/BMP, the dorsal 2 sectors, amnioserosa and the dorsal ectoderm take the cell differentiation program of the ventro-lateral ectoderm (von Ohlen and Doe 2000). Consistent with that, the initial trh expression straddles the ventro-lateral ectoderm (Figure 2B-C), trh expression expands to the dorsal midline in dpp null mutants (Figure 2F) (Isaac and Andrew 1996). However, trh expression is extinguished by stage 12 in dpp null mutants (Figure 2G), consistent with the Dpp/BMP’s role on expression of vvl and grn, which in turn maintain trh expression (Matsuda et al. 2015b). In milder inactivation conditions of dpp/BMP hypomorphic mutants (dpp^H46/dpp^hr92 trans-heterozygotes) (Wharton et al. 1993), trh maintenance, if any, occurs only around the dorsal midline, where reduced Dpp/BMP activity levels are presumed to be sporadically present (Figure 2H). In even milder conditions of dpp/BMP inactivation in dpp^hr92 hypomorphs (Wharton et al. 1993), trh maintenance occurs in the more ventral cells as well (Figure 2I). However, we note that concomitant with reduction of Dpp/BMP activities, more cells in the ventral part of the initial trh expression domain fail to maintain trh expression (Figure 2F-I, Figure 2-figure supplement 1C-D). Loss of the ventral trh expression appears to occur also in wild type embryos (Figure 2-figure supplement 1C), possibly reflecting that the most ventral cells are farthest from the Dpp/BMP source.

The distal and the proximal progenitors are specified in CycA mutants (Matsuda et al. 2015b), where the last progenitor mitosis does not occur (Beitel and Krasnow 2000). As the distal cells are double positive for trh and vvl and occupy 90% of the mature airways, we speculate that Trh⁺ cells that are in and near the Vvl⁺ areas are the airway progenitors whereas the remaining Trh⁺ cells would become epidermal. These airway progenitors are specified straddling the canonical dorsal-ventral ectoderm boundary. We conclude that an optimal Dpp/BMP activity specifies the airway progenitors along the D-V axis.

The radial EGFR signaling primes the airway progenitors and realizes airway differentiation along the P-D axis

Airway cells expressing late differentiation markers are reduced in number in H99 EGFR btl/FGFR mutants, where anti-apoptotic functions of btl/FGFR and EGFR are compensated by H99 deficiency (Matsuda et al. 2015b). These missing cells prompted us to investigate the potential roles of the two RTKs on trh expression earlier.

trh expression is comparable in the distal and the proximal cells at stage 12 in H99 btl/FGFR mutants (Figure 3A). In contrast, trh expression in the proximal area becomes reduced or extinguished at stage 12 in H99 EGFR mutants (Figure 3B). Moreover, trh expression in the main airways is very much reduced in H99 EGFR btl/FGFR mutants, leaving only residual expression in a subset of the invaginated cells (Figure 3C). Thus, EGFR signaling is the predominant factor promoting maintenance of trh expression. In its absence, this function can be compensated by Btl/FGFR signaling (Figure 3D).

Figure 3.

Figure 3 figure supplement 1.

In contrast to a model, where maintenance of trh expression correlates with the transition of primordial cells from a planar 2D into a 3D tubular tissue architecture (Kondo and Hayashi 2019), we show that trh expression is maintained in the cells on the 2D ectodermal plane in the hh vvl double mutants (Figure 1C, Figure 1-figure supplement 1B). Additionally, trh expression is very much reduced in the invaginated cells of H99 EGFR btl/FGFR mutants (Figure 3C), uncoupling trh gene expression from the morphogenetic process of invagination.

To further test if tissue architecture contributes to the maintenance of trh expression, we made use of mutations in wingless (wg)/WNT, which is expressed in longitudinal stripes and represses expression of trh and vvl along the anterior-posterior (A-P) axis (Figure 3E-H) (de Celis et al. 1995; Wilk et al. 1996). In wg mutants, some of the progenitors fail to internalize so that cell expressing either distal or proximal markers are detected at the 2D planar embryo surface (Oda et al. 1994; Matsuda et al. 2015b). Correspondingly, trh RNA and trh-LacZ are detected on the embryo surface in wg mutants at stage 13 and later (Figure 3G).

Similar to wg mutants, the initial trh expression expands along the A-P axis in wg EGFR double mutants (Figure 3I). However, trh RNA expression becomes very weak when most cells still reside on the 2D planes around the stage of primordia invagination (Figure 3J). This significant reduction of trh expression domain is still evident even in H99 wg EGFR mutants where H99 deficiency suppresses apoptosis (Figure 3K). Together, we conclude that tissue architecture is dispensable for trh maintenance.

We suggest that trh maintenance is stimulated concurrently with the robust EGFR activation that occurs and propagates in the 2D airway primordia (Gabay et al. 1997a; Gabay et al. 1997b; Wappner et al. 1997; Matsuda et al. 2015b). EGFR activation also distalizes gene expression (Matsuda et al. 2015b) and initiates morphogenesis (Brodu and Casanova 2006; Nishimura et al. 2007). In this scenario, rather than indirectly through tubulogenesis (Kondo and Hayashi 2019), RTK signaling directly promotes trh maintenance irrespective of tissue geometry. We note that in double mutants of aos and Gap1, which are both negative regulators of EGFR signaling, there is expression of airway distal marker on the embryo surface (Matsuda et al. 2015b). This is accompanied by detection of trh expression on the embryo surface (Figure 3-figure supplement 3A-B).

Thus, subsequent to the Dpp/BMP mediated specification of the airway progenitors along the D-V axis, the radial EGFR signaling sustains trh expression in the progenitors and initiates airway cell differentiation along the P-D axis.

DRaf/MAPKKK and Dsor1/MAPKK are required for EGFR mediated priming of the airway progenitors

Trh activity can be boosted by the PI3kinase-PKB pathway through phosphorylation of Serine 665 (Jin et al. 2001). However, it is not clear how far this pathway is required in vivo for Trh activity and trh autoregulation. We thus investigated if EGFR mediated priming and trh expression maintenance in the airway progenitors require the Ras-MAPK pathway (Figure 4A) (Hou et al. 1995; Mishra et al. 2005). When MAPKK-kinase/Draf or MAPK-kinase/Dsor1 is depleted from the embryos with the ovo-FRT germline clone technique (Hou et al. 1995; Chou and Perrimon 1996), trh expression initiates, as detected with trh-LacZ enhancer trap (Figure 4B-C). However, around the stage of primordial invagination, trh expression is very much reduced, in both non-invaginated or invaginated cells (Figure 4B-C).

Figure 4.

Ras^V12 boosts Trh activity

Ras (Ras85D or Ras1 in Drosophila) is one of the major signaling mediators of RTKs through Draf/MAPKKK-Dsor1/MAPKK-MAPK (Hou et al. 1995; Mishra et al. 2005) and through Pi3kinase- PKB (Figure 4A) (Orme et al. 2006). Consistent with the major role of Ras85D/Ras1 in RTK signaling, comparable defects in maintenance of trh expression are detected in EGFR btl/FGFR double mutants and upon complete loss of Ras85D/Ras1, when anti-apoptotic functions are compensated by the H99 chromosomal deletion (Figure 3C, 4D). Conversely, trh maintenance in EGFR wg/WNT double mutants is significantly rescued resembling wg/WNT single mutants alone upon overexpression of the gain of function form of Ras85D, Ras^V12 (Fortini et al. 1992) (Figure 3L). Thus, Ras serves as the major signaling transducer to promote trh maintenance downstream of EGFR and Btl/FGFR.

The mechanisms by which, Ras promotes trh expression may involve boosting Trh activity because trh auto-regulates its own expression (Figure 4A) (Wilk et al. 1996; Sotillos et al. 2010; Chung et al. 2011). Boosting of Trh activity may involve PKB mediated phosphorylation of Trh (Jin et al. 2001). Alternatively, inferred from the positive roles of MAPK mediated phosphorylation of Hif1, a trh homologue (Mylonis et al. 2006), MAPK may directly upregulate Trh activity. Consistently, overexpression of the phosphomimetic form of Trh, Trh^S665D or simultaneous double overexpression of Ras^V12 and Trh^WT significantly induced ectopic expression of trh-LacZ, whereas single overexpression of Ras^V12 or Trh^WT did not (Figure 5C). The Ras^V12 induced increase of Trh activity is also detected on the expression of both the distal (btl and Gasp/2A12) and the proximal marker (upd and P0144-LacZ) as well (Figure 5A-F). We conclude that Ras activation downstream of RTKs is sufficient to upregulate Trh activity toward its downstream targets (Figure 4A).

Figure 5.

Multiple redundant regulators cooperate during early Drosophila airway tubulogenesis

Tubes are common structures in multicellular organisms and tubular organs fulfill essential functions for life (Romero et al. 2025). An active field in stem cell biology engages in artificial organ generation in vitro to complement tubular organ dysfunction (Lancaster and Knoblich 2014). In vivo studies of tubulogenesis may aid such in vitro organoid development. In line with the revised view that a simple master regulator does not dictate all the steps of the Drosophila airway differentiation (de Celis et al. 1995; Llimargas and Casanova 1997; Chen et al. 1998; Boube et al. 2000; Matsuda et al. 2015b; Ogura et al. 2018; Kondo and Hayashi 2019), we characterize that the cooperation of both extrinsic and intrinsic, partly-redundant regulators ensures their robust morphogenesis.

The airway progenitor specification and their subsequent differentiation is associated with continuous expression of trh (Isaac and Andrew 1996; Wilk et al. 1996). trh is necessary for expression of all airway differentiation markers and for establishment of the airway tubes (Isaac and Andrew 1996; Wilk et al. 1996; Sotillos et al. 2010; Chung et al. 2011; Matsuda et al. 2015b), assigning it as a master TF of Drosophila airway tubulogenesis. Surprisingly, however, trh mutant cells invaginate to form tubes, but later revert to a planar configuration (Kondo and Hayashi 2019). Resolving the issue of what promotes progenitor invagination other than Trh, we identified extrinsic Hh and intrinsic Vvl as the dominant regulators of invagination, leaving trh expression intact.

The airway progenitors are intrinsically defined with at least 3 TFs, trh, vvl and grn. vvl defines the central/distal progenitors that invaginate first, whereas grn defines the peripheral/proximal progenitors that invaginate later, reflecting the proximo-distal differences of the airways (Matsuda et al. 2015b). Maintenance of trh expression is promoted significantly by auto-regulation (Wilk et al. 1996; Sotillos et al. 2010; Chung et al. 2011). However, regulation of trh expression is far more complex, requiring discrete regulations along the A-P, the D-V and the radial/proximo-distal (P-D) axis (Figure 6).

Figure 6.

Along the D-V axis, our results show that intermediate Dpp/BMP activity promotes maintenance of trh expression whereas high level Dpp/BMP activity is known to be repressive on initiation of trh expression (Isaac and Andrew 1996; Wilk et al. 1996). Therefore, Dpp/BMP has two opposing effects on trh expression.

Along the A-P axis, segmentally repeated expression of Unpaired ligands and the resultant activation of Domeless-JAK-STAT signaling precedes and may set where expression of trh and vvl initiates (Brown et al. 2001; Sotillos et al. 2010). The airway field is repressed in the embryonic head and the tail by a zinc finger TF spalt (Boube et al. 2000) whereas it is segmentally repressed in the trunk by Wg/WNT (de Celis et al. 1995; Wilk et al. 1996). It is not known what initiates the remaining expression of trh and vvl in the absence of Unpaired family ligands-Domeless-JAK-STAT signaling (Brown et al. 2001).

Along the radial/P-D axis, the centro-peripheral spreading of Rho-mediated EGFR activation in the 2D progenitor fields promotes maintenance of trh expression. Co-expression of Trh and Ras^V12 potentiates ectopic Trh activity toward downstream genes. Therefore, we suggest that EGFR mediated boosting of TFs like Trh or Vvl may underly trh maintenance (Wilk et al. 1996) and the sequential spreading of rho expression (Matsuda et al. 2015b), generating a wave of feed-forward loops for progenitor priming and airway differentiation.

Also, we previously showed that grn potentiates trh maintenance in the peripheral/proximal cells whereas vvl circumvents the trh expression dependency on grn in the central/distal cells (Matsuda et al. 2015b), arguing for region-specific requirements for trh maintenance in the primordium.

In our model, it is pre-determined which cells in the trh expressing 2D fields contribute to airways or epidermis before invagination (Figure 6D). Cells not receiving proper amounts of positive inputs from like optimal Dpp/BMP activities or enough EGFR activities or that receive too much negative inputs like Wg/WNT, would cease trh transcription. The timing of transcriptional silencing and degradation of the remaining transcripts and the proteins would determine when the cells lose trh products. Recently, it was discovered that micro-peptides encoded by polished rice/tarsal-less promote trh expression by suppressing the repressor activity of a zinc finger TF Ovo/Shaven-baby (Mizuno et al. 2026). It is intriguing, where this circuit fits into the regulatory modes of trh expression we described above.

Multiple partially redundant systems on initiation and maintenance of gene expression and morphology sustain development of even simple organs like the Drosophila airways. The crucial roles of tubular organs for survival could be the driving force in evolving multiple overlying regulatory schemes, which may secure development of essential organs even when one regulatory scheme becomes dysfunctional.

Materials and methods

Fly genetics and histochemistry of embryos were done as previously described (Matsuda et al. 2015a; Matsuda et al. 2015b). Marker genes were typically introduced as a single copy, unless otherwise noted.

Fly strains used in this study are:

• aos^Δ7 (BDSC 1004)

• arm-Gal4 (BDSC 1560)

• btl-Gal4 (a gift from Dr. S. Hayashi)

• D-lacZ (a gift from Dr. J. Nambu and Dr. S. Russel)

• Draf^11-22 FRT101 (a gift from Dr. N. Perrimon)

• Dsor1^LH110 FRT101 (a gift from Dr. N. Perrimon)

• btl^Δoh10(Ohshiro and Saigo 1997)

• btl^Δoh24-1(Ohshiro and Saigo 1997)

• da-Gal4 (BDSC 8641)

• dpp^H46(BDSC 2061)

• dpp^hr92(BDSC 2069)

• Dr/msh-lacZ (a gift from Dr. A. Nose)

• Df(2L)Exel6017=Df(wg) (BDSC 7503)

• Df(3L)Exel6109=Df(vvl) (BDSC 7588)

• Gap1^B2 (a gift from Dr. N. Perrimon)

• GMR19D03 (BDSC 48846)

• grn-GFP (BDSC 58483)

• H99=Df(3L)H99 (BDSC 1576)

• hh^13C (Hosono et al. 2003)

• hh^AC(BDSC 1749)

• hs-FLP (BDSC)

• iroquois-lacZ (a gift from Dr. S. Campuzano)

• ovoD1 FRT101; hs-FLP (BDSC 1813)

• ovoD1 FRT82B (BDSC 2149)

• P0144-lacZ (a gift from Dr. W. Janning, Flyview)

• Ras85D^Δc40b (a gift from Dr. N. Perrimon and Dr. C. A. Berg)

• top^f2 (BDSC 2768)

• top^f24 (a gift from Dr. K. Moses)

• trh-lacZ=1-eve-1 (a gift from Dr. N. Perrimon)

• UAS-GFP.nls (BDSC 4776)

• UAS-trh^WT/CyO (a gift from Dr. A. S. Manoukian)

• UAS-trh^S665D (a gift from Dr. A. S. Manoukian)

• UAS-Ras^V12/CyO (a gift from Dr. G. M. Rubin)

• UAS-Ras^V12/TM3 (a gift from Dr. N. Perrimon)

• vvl^utH599 (a gift from Dr. A. Salzberg)

• wg^CX4 (BDSC 2980)

• Antibodies used for immunohistochemistry are;

• Rabbit anti-GFP (invitrogen)

• Rabbit anti-LacZ (Capel)

• Rabbit anti-Trh (this study)

• Double fluorescent labeling with RNA probes and antibodies was carried out as described (Goto and Hayashi 1997).

• DNA clones used for in situ RNA detection are;

• caup (Drosophila Genomics Resource Center, DGRC)

• trh (Drosophila Genomics Resource Center, DGRC)

• vvl (a gift from Dr. J. Casanova)

Confocal images were taken by Bio-Rad (Hercules, CA) MRC1024, Olympus (Japan) Fluoview 1000 or Zeiss (Germany) LSM800. Images were processed by ImageJ and figures were prepared with Adobe Photoshop and Illustrator.

Data availability

All the data are available.

Acknowledgements

We thank the members of the fly community who isolated, characterized or distributed fly strains, antibodies or DNA clones. Especially, we thank Drs. C. A. Berg, S. Campuzano, J. Casanova, S. Hayashi, W. Janning, A. S. Manoukian, K. Moses, J. Nambu, A. Nose, N. Perrimon, G.M. Rubin, S. Russel, A. Salzberg, DGRC and BDSC for directly sharing fly strains and DNA clones. We thank Flybase for the Drosophila genomic resources. We thank the Stockholm University Imaging Facility and MBW fly services. We thank V. Tsarouhas for microscope help. Special thanks to Y. Emori and F. Ui-Tei for help in maintaining fly strains after the retirement of K. Saigo.

This work was funded by the Ministry of Education, Culture, Sport, Science and Technology of Japan to K.S. and the Swedish Research Council, the Swedish Cancer Society and German Research Foundation to C.S.

Additional information

Author contribution

RM, conceived the project, designed the experiments, performed experiments, interpreted data, drafted the manuscript with inputs from CH, writing, figure preparation

CH, performed experiments, figure preparation, writing

KS, funding acquisition, provided experiment and analysis tools

CS, funding acquisition, provided experiment and analysis tools, writing

Funding

Cancerfonden (Swedish Cancer Society)

• Christos Samkovlis

Vetenskapsrådet (VR)

• Christos Samkovlis

Deutsche Forschungsgemeinschaft (DFG)

• Christos Samkovlis

Ministry of Education, Culture, Sports, Science and Technology (MEXT)

• Kaoru Saigo