The Drosophila AWP1 ortholog Doctor No regulates JAK/STAT signaling for left–right asymmetry in the gut by promoting receptor endocytosis

Many animals show directional left–right (LR) asymmetry in their body structure and function. Several mechanisms, such as cilia-generated flow, contribute to LR axis formation in vertebrates (Yoshiba and Hamada, 2014). However, the mechanisms underlying LR asymmetric development in invertebrates are relatively obscure and remain an elemental question in biology (Vandenberg and Levin, 2013).

Stereotypical LR asymmetry is present in several organs of Drosophila, including the gut, testis, male genitalia, and brain (Coutelis et al., 2014; Okumura et al., 2008). Of these organs, the embryonic gut is the first to show LR asymmetry during development (Hayashi and Murakami, 2001; Hozumi et al., 2006). Intriguingly, the anterior and posterior parts of the embryonic gut are controlled by two distinct gene groups. The LR asymmetry of the posterior part is regulated by Myo1D, which determines cell chirality (Hozumi et al., 2006; Inaki et al., 2016, 2018; Nakamura et al., 2013; Spéder et al., 2006; Utsunomiya et al., 2019). An entirely different mechanism governs the LR asymmetry of the anterior part. The anterior gut of the embryo consists of the foregut (FG) and midgut (MG), which are complex structures with directional and stereotypical LR asymmetry (Fig. 1A,B) (Hayashi and Murakami, 2001). Various genetic pathways, including the JNK and Wnt pathways, play important roles in LR asymmetric development (Hayashi et al., 2005; Kuroda et al., 2012; Maeda et al., 2007; Okumura et al., 2010; Shin et al., 2021; Taniguchi et al., 2007). A genetic screen performed by our group suggested that the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling pathway is also involved in LR asymmetric development. The present study reveals the involvement of JAK/STAT signaling in LR asymmetric development of the anterior gut.

JAK/STAT signaling, which is evolutionarily conserved from Drosophila to humans, is essential for morphogenesis, cell proliferation, cell differentiation, cell death, immunity and other biological events (Arbouzova and Zeidler, 2006; Calò et al., 2003; Lee et al., 2017; Myllymäki and Rämet, 2014; Recasens-Alvarez et al., 2017; Seif et al., 2017). In JAK/STAT signaling in Drosophila, the ligands are encoded by unpaired (upd; upd1), upd2 and upd3 and the receptor is encoded by domeless (dome) (Agaisse et al., 2003; Brown et al., 2001; Ghiglione et al., 2002; Harrison et al., 1998; Hombría et al., 2005). JAK is encoded by hopscotch (hop) in Drosophila. Moreover, similar to the mammalian JAK/STAT system, Hop constitutively associates with the intracellular domain of Dome in Drosophila (Arbouzova and Zeidler, 2006; Binari and Perrimon, 1994). Ligand binding induces conformational changes in Dome that lead to the phosphorylation of JAK, which phosphorylates Dome to create docking sites for STATs, particularly Stat92E in Drosophila (Yan et al., 1996). JAK then tyrosine phosphorylates STATs, which are subsequently dimerized and translocated to the nucleus to bind to enhancers of target genes and activate transcription (Arbouzova and Zeidler, 2006; Hou et al., 1996; Yan et al., 1996) (Fig. 3A).

In mammals, endocytosis regulates JAK/STAT signaling through various mechanisms (German et al., 2011; Lei et al., 2011). Endocytosis and the regulation of JAK/STAT signaling activity are also closely connected in Drosophila (Devergne et al., 2007; Moore et al., 2020; Ren et al., 2015; Vidal et al., 2010). In the classic working model of endocytosis, membrane receptors are internalized into endosomal compartments, where they are degraded and recycled. This reduces the number of receptors available to transduce signaling (Cendrowski et al., 2016; Elkin et al., 2016; Piper et al., 2014). Moreover, receptor activity requires endocytic trafficking in some signaling pathways (Cendrowski et al., 2016; Irannejad et al., 2013). Studies on the relationship between endocytosis and JAK/STAT signaling in Drosophila have provided contradictory results regarding whether endocytosis upregulates or downregulates JAK/STAT activity (Devergne et al., 2007; Moore et al., 2020; Ren et al., 2015; Vidal et al., 2010). To clarify these conflicting results, it would be helpful to identify and examine a factor that specifically regulates the endocytosis of Dome.

In this study, we found that the Drosophila ortholog of AWP1 (associated with PRK1, also known as ZFAND6), which we refer to as Doctor No (Drn), positively regulates JAK/STAT signaling by facilitating endocytic trafficking of the Dome receptor, which is required for normal LR asymmetric development of the embryonic gut. Drosophila Drn was named after Ian Fleming's fictional character, whose heart was located on the right side of his chest (Fleming, 1958). Drn protein contains an A20-type zinc finger at the N terminus and an AN1-type zinc finger at the C terminus; this is similar to the mammalian ortholog, which was first identified in humans and mice (Duan et al., 2000). In vertebrates, AWP1 proteins bind to ubiquitin, regulate NF-κB activity, and stimulate the export of Pex5 from the peroxisome, among other roles (Chang et al., 2011; Fenner et al., 2009; Miyata et al., 2012). During Xenopus development, AWP1 modifies Wnt and FGF signaling to specify neural crest cells (Seo et al., 2013). However, the molecular mechanisms by which AWP1/Drn proteins influence these various cell signaling pathways are not well understood. In this study, we found that AWP1/Drn plays a crucial role in internalizing the Dome receptor and propose a mechanism by which AWP1/Drn positively modulates JAK/STAT signaling.

To identify genes affecting LR asymmetry in the anterior gut, including the FG and anterior MG (AMG), we performed a genetic screen using a large collection of P element insertion lines (Drosophila Genes Search Project; http://kyotofly.kit.jp/stocks/documents/GS_lines.html) (Toba et al., 1998). We scored the LR asymmetry phenotypes of the anterior gut in mutants obtained from our genetic screen. The FG is composed of the pharynx, esophagus (ES) and proventriculus (PV), which is a valve-like structure connecting the FG to the AMG (Fig. 1A,B). We defined normal LR asymmetry in the FG and AMG as follows. (1) When viewed from the ventral side, the wild-type Drosophila ES loops in an inverse C shape and is connected to the PV (100%, n=50); this was defined as normal FG laterality (Fig. 1A). (2) The joint between the PV and AMG is located on the right side of the midline in wild-type embryos (100%, n=50); this was defined as normal laterality of the AMG (Fig. 1A). Using these criteria, our genetic screen identified two mutant lines, GS12294 and GS10567, that affect LR asymmetry in the FG and AMG (Fig. 1C).

These lines carry a P element insertion in the CG45050 locus (Fig. 1D). In this study, we named CG45050 as drn. Drosophila Drn and human AWP1 share 42.1% identity and 62.2% similarity in terms of the whole protein (calculated using EMBOSS pairwise alignment algorithms) (Fig. 1E). In particular, an A20-type zinc finger (amino acids 6-40) and an AN1-type zinc finger (amino acids 137-180) are highly conserved (Fig. 1E). A20-type zinc fingers are found in various proteins with ubiquitin-editing functions; these proteins are often associated with human pathogenesis (Heyninck and Beyaert, 1999; Jacque and Ley, 2009; Kim et al., 2021; Rothe et al., 1995; Song et al., 1996). We noted LR asymmetry defects in the anterior gut in 42.5% of Drosophila embryos homozygous for drn^GS12294 and 13.0% of Drosophila embryos homozygous for drn^GS10567, indicating a disturbance in LR asymmetric development (Fig. 1C,F). In contrast, hindgut and posterior MG laterality were normal in all cases examined (n=40), indicating heterotaxy but not situs inversus in these phenotypes. To genetically characterize drn further, we generated deletion mutant alleles of drn through imprecise P element excision. In drn¹, the deduced initiation codon and 5′ portion of the coding region were deleted from the drn alternative RNA products CG45050-RB, -RC, -RD, -RE, -RF and -RG (Fig. 1D). In drn², the deduced initiation codon and most coding sequences were deleted from all drn alternative RNA products, suggesting that drn² is a loss-of-function mutant of drn (Fig. 1D). Various degrees of LR asymmetry defects were observed in drn¹ or drn² homozygotes and transheterozygotes, demonstrating that mutations in drn are responsible for the LR defect phenotypes (Fig. 1F). These mutant embryos had LR defects in both the FG and AMG or in the AMG alone (orientation of the FG was correct, but the joint between the FG and PV was on the opposite side); however, we did not observe defects in the FG alone (Fig. 1F,G). Thus, LR asymmetry defects in the FG are strictly coupled with those in the AMG, suggesting the primary role of the AMG in LR asymmetric development of the anterior gut, as has been observed in other mutants with defective LR asymmetry (Kuroda et al., 2012; Taniguchi et al., 2007). We also noted relatively mild LR defects in heterozygotes of drn¹ or drn², indicating that drn may behave in a semidominant manner (Fig. 1F).

To confirm whether drn is required for LR asymmetric development of the anterior gut, we performed rescue experiments using the GAL4/UAS system (Brand and Perrimon, 1993). In this system, the expression of wild-type drn encoded by UAS-drn is driven by various tissue-specific GAL4 drivers in a drn¹ homozygote mutant background (Brand and Perrimon, 1993; Elliott and Brand, 2008). The FG and AMG are composed of the epithelium, circular visceral muscle (CVMU) and longitudinal visceral muscle (LVMU) (Fig. 2A). LR defects of the FG and AMG in drn¹ homozygotes were reduced by half when UAS-drn was introduced without any GAL4 driver (negative control), probably as a result of leaky expression of UAS-drn (Figs 1F and 2B). These LR defects were efficiently rescued by UAS-drn misexpression driven by da-GAL4 (ubiquitous), NP1522 (in the somatic muscle and CVMU), hand-GAL4 (in the CVMU), 24B-GAL4 (in the CVMU, LVMU and somatic muscle) and 48Y-GAL4 (in the AMG epithelium, CVMU and LVMU) (Fig. 2B). In contrast, LR defects were not rescued by UAS-drn misexpression driven by NP0221 (in the LVMU), NP5021 (in the epithelium) or elav-GAL4 (in the nervous system) (Fig. 2B). Collectively, our data suggest that drn expression is primarily required in the CVMU, but not in the LVMU or other tissues for normal LR asymmetric development of the FG and AMG. We also found that in wild-type embryos UAS-drn misexpression driven by the GAL4 drivers tested here did not induce marked LR defects in the FG and AMG, suggesting that enhanced drn expression did not affect LR asymmetric development in wild-type embryos (Fig. S1).

We previously revealed that the LR asymmetric tilt of the ellipsoidal nuclei in the CVMU cells covering the MG epithelium is the first point of disruption of LR symmetry in the anterior gut (Fig. 2C) (Kuroda et al., 2012; Okumura et al., 2010; Shin et al., 2021; Taniguchi et al., 2007). In wild-type embryos at early stage 15, the major axis of the ellipsoidal nuclei in the CVMU cells was still perpendicular to the midline of the AMG (Fig. 2C). When the MG chambers began dividing at late stage 15, the nuclei in the lower 30% of the presumptive first chamber on the right began tilting diagonally upward and to the right (Fig. 2C). This LR asymmetric nuclear tilt preceded the LR asymmetric morphological changes in the AMG, indicating that it was not a consequence of LR asymmetric morphogenesis (Taniguchi et al., 2007). To analyze whether this process was disrupted in drn mutants, we used 65E04-GAL4 to express nuclear-localizing RedStinger fluorescent proteins specifically in the visceral muscles. The angle (θ) between the major axis of the ellipsoidal nuclei in the lower 30% of the presumptive first chamber and the anterior–posterior axis of the embryo was measured by a blind test at late stage 15 (Fig. 2D). Consistent with our previous results, the angle measured was smaller for nuclei on the right side than for those on the left side in wild-type embryos (P<0.0001, Kolmogorov–Smirnov test) (Fig. 2E,E′) (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). However, in drn¹ homozygotes, no significant difference was noted in the angle between the left and right sides. Moreover, the angle remained closer to perpendicular on both sides even at late stage 15 (P=0.765, Kolmogorov–Smirnov test) (Fig. 2F,F′). In addition to the LR asymmetric tilting of the nuclei, we previously reported that the nuclei assemble in belt-shaped zones along the anterior–posterior axis. These nuclei are scattered in mutants with defects in LR asymmetry of the anterior gut (Shin et al., 2021). In this study, we observed that the distribution of nuclei was more dispersed in drn¹ homozygotes than in wild-type embryos in all cases examined (n=10) (Fig. 2E-F′). Collectively, these results suggest that drn contributes to LR asymmetric development of the AMG by regulating the LR asymmetric rearrangement of nuclei in CVMU cells.

To verify the roles of drn in the CVMU, we assessed the distribution of drn mRNA in embryos by in situ hybridization at various stages of embryogenesis (Fig. S2A-D). Comprehensive analyses of gene expression in Drosophila have previously revealed that drn is highly expressed in early embryonic stages (Thurmond et al., 2019). In the present study, drn mRNA was strongly detected in the preblastoderm to blastoderm (stage 5) stages, suggesting that drn mRNA is maternally provided (Fig. S2A,B). We also found that drn was broadly expressed at stages 11 and 15, including the trunk visceral mesoderm (TVM, the primordium of the CVMU), in which drn is required for LR asymmetric development of the anterior gut (Fig. S2C). In contrast, the negative control (sense probe) exhibited no signal under the same conditions (Fig. S2E-H).

Although previous research has suggested that Drn negatively regulates JAK/STAT signaling in cultured Drosophila cells (Vidal et al., 2010), the role of Drn has not been explored further in vivo. As drn is involved in JAK/STAT signaling, we hypothesized that JAK/STAT signaling plays a role in LR asymmetric morphogenesis of the anterior gut. Various mutants of genes involved in the JAK/STAT signaling pathway, including dome, Stat92E, upd and hop, were scored for LR phenotypes (Fig. 3B). The roles of these gene products are presented schematically in Fig. 3A. These mutants showed various degrees of LR defects in the FG and AMG, indicating that JAK/STAT signaling is indispensable for normal LR asymmetric development of the anterior gut (Fig. 3B). In addition, when JAK/STAT signaling was augmented by misexpression of an activated form of Hop (UAS-hop^Tum-l) (Harrison et al., 1995), specifically in CVMU cells under the control of hand-GAL4 or 24B-GAL4, the embryos showed LR defects in the FG and AMG (Fig. 3B). Therefore, excessive activation of JAK/STAT signaling can also disrupt the LR asymmetry of the FG and AMG. Taken together, these findings indicate that JAK/STAT signaling activity must be maintained at proper levels for LR asymmetric development of the anterior gut.

To analyze the connection between drn and JAK/STAT signaling more directly, we assessed the characteristic phenotypes related to JAK/STAT signaling in drn mutants. The expression of a pair-rule gene, even-skipped (eve), was disturbed in embryos homozygous for Stat92E, such as Stat92E^HJ, with approximately 67% of the embryos showing aberrant phenotypes, such as weak stripe 3 (19%), fusion between stripes 2 and 3 (19%), or fusion between stripes 5 and 6 (67%) (n=27), rather than the seven-stripe pattern noted in wild-type embryos (Fig. 3C-D′) (Yan et al., 1996). We next analyzed the expression of eve in embryos homozygous for drn² and found that all embryos had normal eve expression and displayed a normal seven-stripe pattern (n=20) (Fig. 3E). As we found that the mRNA of drn is maternally supplied (Fig. S2A,B), we genetically removed maternal drn from drn² homozygotes (drn^{2 m/z}). The phenotypes of drn^{2 m/z} embryos were similar to those of Stat92E^HJ homozygotes (36%, n=33) (Fig. 3F,F′). These results suggest that drn is required for JAK/STAT signaling, with Stat92E playing an essential role. To verify this idea, we examined trachea morphology, which is controlled by JAK/STAT signaling in later embryonic stages (stages 15-17) (Li et al., 2003). The trachea was detected by CBP546 staining (Dong et al., 2014). In Stat92E^HJ homozygotes, the dorsal trunk of the trachea was disturbed compared with that in wild-type embryos (Fig. 3G,H). Moreover, it was truncated in some drn² homozygotes (35%, n=20) and drn^{2 m/z} embryos (33%, n=21). Defects in drn^{2 m/z} embryos were more severe than those in drn^{2 z} homozygotes, which can be predicted from the maternal contribution of drn (Fig. 3I,J). These results suggested that, similar to Stat92E, drn plays a positive role in JAK/STAT signaling during embryonic development.

To verify this possibility, we assessed whether the ubiquitous misexpression of drn, driven by actin5c-GAL4, could rescue LR defects in embryos homozygous for Stat92E⁰⁶³⁴⁶. The embryos homozygous for Stat92E⁰⁶³⁴⁶ showed LR defects in the anterior gut at a frequency of 25% (Fig. 3K); however, the ubiquitous misexpression of drn reduced the frequency of LR defects to 4% (Fig. 3K). This result can be explained by a previous finding that Stat92E exhibits a maternal effect (Hou et al., 1996; Li et al., 2003; Tsurumi et al., 2011). We speculated that the supply of maternal Stat92E to Stat92E⁰⁶³⁴⁶ mutant embryos is sufficient to support the activity of overexpressed drn, which can consequently rescue LR defects of these embryos. In particular, our studies demonstrate that drn positively contributes to the activation of JAK/STAT signaling in three different developmental contexts. Therefore, we propose that Drn is a general component positively acting on JAK/STAT signaling, although a previous study involving knockdown by RNA interference (RNAi) and a reporter assay revealed that wild-type drn could downregulate JAK/STAT signaling in cultured Drosophila cells (Vidal et al., 2010). The cause of this discrepancy is unclear, but may involve a negative feedback loop that is active in a particular time frame during JAK/STAT signaling, as detected by the reporter assay in cultured cells.

Although the biochemical roles of Drn have not been studied in Drosophila, the mammalian ortholog AWP1 binds to ubiquitin and modulates the functions of ubiquitylated proteins in mammals (Chang et al., 2011; Duan et al., 2000; Fenner et al., 2009). This process is often related to endocytic trafficking and the lysosomal breakdown of membrane receptors (Piper et al., 2014). Thus, we hypothesized that Drn is involved in endocytic trafficking that regulates the JAK/STAT signaling pathway. As TVM and CVMU cells are located deep inside the embryo, it is difficult to obtain clear microscopic images, making them unsuitable for analyzing the subcellular localization of Drn. However, in situ hybridization analysis revealed that drn is also expressed in the epidermis of the embryo (Fig. S2). Hence, we analyzed the potential colocalization of Drn with various endocytic compartments in the epidermis.

To detect Drn protein, we generated a polyclonal antibody (anti-Drn antibody) against full-length Drn and assessed its specificity using the UAS-GAL4 system and RNAi against drn in order to deplete Drn in the stripe along the anterior–posterior boundary of the wing disc (the region expressing ptc). We found that anti-Drn antibody staining was largely absent from the stripe, confirming the specificity of the antibody (Fig. S3). Using this anti-Drn antibody, we examined the potential colocalization of Drn with various endosomal compartments in the epidermis of wild-type embryos. Drn was detected as punctae in the cytosol and was occasionally found colocalized with endosomal markers, such as Hrs (early endosomes), Rab5 (early endosomes), Rab7 (late endosomes), LAMP1 (lysosomes) and Rab11 (recycling endosomes) (white arrowheads in Fig. 4). Thus, although the distribution of Drn did not concentrate with any particular endosomal markers, Drn appeared to localize with endocytic compartment markers, such as Hrs, Rab5, Rab7 and LAMP1, at a low frequency; this is consistent with our hypothesis that Drn plays some role in endocytic trafficking.

As Drn positively contributes to JAK/STAT signaling by regulating endocytic trafficking, we speculated that Drn regulates the JAK/STAT signaling receptor Dome during its endocytic trafficking (Brown et al., 2001). We analyzed the subcellular distribution of Dome using full-length Dome protein that had a C-terminal GFP tag (Dome-GFP) and maintained wild-type Dome functions (Ghiglione et al., 2002). UAS-dome-GFP was ubiquitously driven under the control of da-GAL4 in wild-type and drn homozygous embryos (Fig. 5A-B″).

In wild-type embryos, Dome-GFP was detected in epidermal cells as punctae localized to the vicinity of the plasma membrane and cytosolic vesicles (Fig. 5A,A′; Fig. S4). Cytoplasmic vesicles containing Dome-GFP were occasionally labeled using markers of intracellular components, such as Rab5, Rab11 and LAMP1; however, Dome-GFP did not appear to associate with any particular intracellular component (Fig. S4). Previously, Dome was found to accumulate in ubiquitylated cargoes when endocytosis was impaired (Tognon et al., 2014). Hence, we assessed whether these vesicles containing Dome-GFP were stained with an anti-ubiquitin antibody that could recognize mono- and polyubiquitin. However, we did not detect colocalization between Dome-GFP and ubiquitin in the wild-type cells (Fig. 5L-L″). In the epidermal cells of drn² homozygotes, Dome-GFP was markedly higher than that in wild-type embryos in all cases, as observed in images of Dome-GFP obtained at the same gain of signal amplification (Fig. 5B). In images obtained using reduced gain of signal detection, Dome-GFP was observed as larger clumps located near the plasma membrane (Fig. 5B′,B″). To analyze the nature of these clumps, we co-stained Dome-GFP with markers of representative intracellular compartments, including Pdi (endoplasmic reticulum), GM130 (cis-Golgi), peanut agglutinin (PNA; trans-Golgi), Sec5 (exocytic vesicles), Hrs (early endosomes), Rab5 (early endosomes), Rab7 (late endosomes), Rab11 (recycling endosomes) and LAMP1 (lysosomes), in the epidermal cells of drn² homozygotes. None of these markers colocalized with Dome-GFP (Fig. 5C-K″). However, large clumps of Dome-GFP often colocalized with ubiquitin (Fig. 5M-M″). Our quantitative analyses revealed that 3.6% of these Dome-GFP clumps colocalized with ubiquitin and 5.3% of ubiquitin-positive vesicles colocalized with Dome-GFP. We speculate that Dome-GFP localizes with ubiquitylated cargos only temporarily because colocalization was observed only in a subset of these vesicles. Nevertheless, as ubiquitylation regulates endosomal trafficking and sorting, these results suggest that Dome-GFP accumulates in atypical endosomal compartments in drn mutants. Such a defect in endocytosis may cause JAK/STAT signaling to deteriorate, supporting previous suggestions that endocytosis is essential for Dome activation (Devergne et al., 2007; Moore et al., 2020). Additionally, defective endocytosis may prevent the degradation of Dome-GFP in lysosomes of drn mutants. Unlike Dome-GFP, the Wnt and Notch signaling receptors Frizzled 2 (Fz2) and Notch, respectively, were unchanged in the epidermis of wild-type and drn² homozygotes (Fig. S5). Hence, defective endocytosis in drn² homozygotes is specific to Dome-GFP.

Considering the potential binding of Drn to ubiquitin, we determined whether Dome-GFP and Drn could colocalize in wild-type cells. Although the staining patterns of Drn and Dome-GFP did not broadly resemble each other, we found that Drn and Dome-GFP often colocalized with each other (white arrowheads in Fig. 5N-N″). Our quantitative analyses revealed that 3.5% of Dome-GFP colocalized with Drn and 5.8% of Drn colocalized with Dome-GFP. These results suggest that Drn interacts with the ubiquitin moiety on Dome in some endocytic compartments to facilitate proper Dome trafficking. However, such an interaction may be transient because vesicles demonstrating colocalization did not account for a majority. Nevertheless, based on these results, we speculated that ubiquitylated Dome can be recognized by Drn, which can specifically promote the endocytic transportation and degradation of Dome. Dome was accumulated in the atypical endocytic compartment with ubiquitylated cargoes and JAK/STAT signaling was attenuated in drn mutants, indicating that such an endocytic process is crucial for activating JAK/STAT signaling (Fig. 6).

In this study, we demonstrate that drn, which encodes a Drosophila ortholog of AWP1, plays crucial roles in the LR asymmetric development of the FG and AMG by positively regulating JAK/STAT signaling. In this process, wild-type drn is required for the normal endocytic trafficking of Dome, which is the Drosophila JAK/STAT signaling receptor. Whether the internalization as well as endocytic trafficking of Dome is required to activate JAK/STAT signaling in Drosophila has been a topic of debate (Devergne et al., 2007; Moore et al., 2020; Ren et al., 2015; Vidal et al., 2010). Analyses of mutations affecting Dome endocytosis have revealed that Dome must be internalized into early endosomes to activate JAK/STAT signaling (Devergne et al., 2007). In contrast, the RNAi-mediated knockdown of genes required for Dome endocytosis was found to enhance JAK/STAT signaling activity, demonstrating that Dome endocytosis negatively regulates JAK/STAT signaling (Vidal et al., 2010). This discrepancy can be explained by the multiple ramifications and parallelism of the endocytic pathway, as described in recent studies (Cendrowski et al., 2016). In particular, blocking a particular step along the pathway can divert endocytic trafficking into various positive or negative regulatory cascades of cell signaling pathways. The relative contributions of endocytic regulators can also change the course of endocytic trafficking, as is well documented in Notch signaling (Shimizu et al., 2014). Moreover, a change in the balance of regulators may change how endocytosis regulates cell signaling pathways according to the context. Hence, to understand comprehensively how endocytosis contributes to the activation of Dome, it is necessary to examine the point at which endocytosis is disrupted and the altered course of endocytic trafficking when normal endocytosis fails.

A study using RNAi of cultured Drosophila cells identified drn as a negative regulator of JAK/STAT signaling; this finding appears to be opposite to our model (Vidal et al., 2010). However, as other genes required for Dome endocytosis were also identified as negative regulators of JAK/STAT signaling in that analysis, it is likely that this discordance regarding the role of Drn in JAK/STAT signaling results from a discrepancy in the contribution of Dome endocytosis to JAK/STAT signaling, as discussed above (Cendrowski et al., 2016). Regarding this discrepancy, it has been proposed that long-term, loss-of-function analyses, including analyses of mutants in which Dome endocytosis is disrupted, may reflect unexpected cell fate changes induced by altered endocytic pathways and their influence on JAK/STAT signaling (Vidal et al., 2010). However, in this study, we examined three developmental contexts in which JAK/STAT signaling is reduced in drn mutants. One of these involves the regulation of eve expression, which occurs before cell fate specification and during a short period in early embryogenesis; thus, it does not fit well with a mechanism involving unexpected cell fate changes. Therefore, we propose a model in which Drn is required in some steps of the endocytic trafficking of Dome, which plays a role in the subsequent activation of JAK/STAT signaling (Fig. 6). Our model is consistent with that proposed by Devergne et al., in which the endocytic trafficking of Dome activates JAK/STAT signaling and also provides a mechanism to regulate it quantitatively (Devergne et al., 2007). However, as where and how Drn can control the endocytic trafficking of Dome remain unclear, the coincidence between previous and current results should be interpreted with caution.

Drn is the Drosophila ortholog of AWP1, which binds to ubiquitin and modulates the functions of ubiquitylated proteins in mammals and Xenopus species (Chang et al., 2011; Duan et al., 2000; Fenner et al., 2009; Miyata et al., 2012; Seo et al., 2013). It is known that ubiquitylation of the receptors involved in JAK/STAT signaling is important for regulating signaling activities in mammals (Gesbert et al., 2005; Martinez-Moczygemba et al., 2007; Wölfler et al., 2009). The sorting of ubiquitylated membrane proteins into intraluminal vesicles relies on protein complexes in the ESCRT family (Babst et al., 2002a,b; Katzmann et al., 2001). ESCRT-0, ESCRT-I, and ESCRT-II include multiple ubiquitin-binding proteins and interpret ubiquitin as a signal to sort membrane proteins (Clague et al., 2012). In Drosophila mutants of the ESCRT-0 complex components Hrs and Stam, ubiquitylated membrane proteins, such as Notch and Dome, aggregate at the cell cortex and in intracellular compartments (Jékely and Rørth, 2003; Tognon et al., 2014). As such intracellular compartments, including aggregated Dome, were stained with an anti-ubiquitin antibody, it has been proposed that Drosophila Dome is ubiquitylated and its endosomal sorting is controlled by ubiquitylation; however, the ubiquitylation of Dome has not been confirmed biochemically (Tognon et al., 2014). We found that a loss of Drn caused Dome to accumulate in large clumps that frequently colocalized with ubiquitin but were not labeled by markers of typical intracellular compartments (Fig. 5C-K″). In addition, Drn occasionally colocalized with markers of various endocytic compartments, demonstrating that Drn is an endocytic protein (Fig. 4). Hence, we speculate that Drn plays a role in the ubiquitin-dependent internalization or sorting of Dome through its potential ubiquitin-binding activity (Fig. 6). This idea is consistent with our observation that Drn often colocalizes with Dome in some intracellular vesicles in wild-type Drosophila (Fig. 5N-N″). Thus, in our model, Dome may be misrouted to atypical endocytic compartments, where it fails to be phosphorylated in the absence of Drn. Conversely, differences in Dome trafficking routes between wild-type and drn mutant embryos should help identify the endocytic compartment where Dome is activated by phosphorylation. However, it is difficult to delineate incorrect Dome trafficking routes in the drn mutant because Dome did not specifically colocalize with typical markers of endocytic compartments under this condition. Our model also predicts that such misrouting can consequently prevent the degradation of Dome in lysosomes and leave it to accumulate, as observed in drn mutants (Fig. 6).

Our analyses revealed that drn mutations induce marked accumulation of Dome but not of Notch or Fz2 (Fig. S5). The intracellular distribution of Notch and Fz2 in the drn mutant appeared to be similar to that in the wild type. Thus, Drn is not a general component of endosomal protein sorting but is specific to Dome; however, how such specificity is achieved remains unclear.

In this study, we found that JAK/STAT signaling activity must be maintained at proper levels for normal LR asymmetric development of the FG and AMG. We previously observed a similar phenomenon in Wnt or JNK signaling activity (Kuroda et al., 2012; Taniguchi et al., 2007). However, we failed to detect any LR asymmetry in the activity or the distribution of molecules involved in the JAK/STAT, Wnt and JNK signaling pathways (Fig. S6) (Kuroda et al., 2012; Taniguchi et al., 2007). As these three signaling pathways are required for LR asymmetric rearrangement of nuclei in the visceral muscles of the MG, they may play permissive roles in rearranging these nuclei upon a common cue, which is yet unknown, of LR polarity.

AWP1/Drn is highly conserved from Drosophila to humans (Fig. 1E). As receptors in the mammalian JAK/STAT pathway are also ubiquitylated, we suggest that the role of AWP1/Drn in JAK/STAT signaling and LR asymmetric development is evolutionarily conserved in various organisms (Gesbert et al., 2005; Martinez-Moczygemba et al., 2007; Wölfler et al., 2009).

We used Canton-S as the wild-type Drosophila strain. We generated the drn¹ and drn² mutants in this study. drn^GS12294 and drn^GS10567 are previously reported GS lines (Toba et al., 1998). We also generated UAS-drn and UAS-myc-drn in this study. UAS-hop^Tuml (Harrison et al., 1995) and UAS-dome-GFP (Ghiglione et al., 2002) have been previously described. UAS-drnRNAi (Vienna Drosophila Resource Center, #103508) was used for RNAi against drn. The following GAL4 lines were used in this study: da-GAL4 (Wodarz et al., 1995), 48Y-GAL4 (Martin-Bermudo et al., 1997), 24B-GAL4 (Brand and Perrimon, 1993), hand-GAL4 (Popichenko et al., 2007), elav-GAL4 (Yao and White, 1994), NP1522 (Hayashi et al., 2002), NP5021 (Hayashi et al., 2002), NP0221 (Hayashi et al., 2002) and 65E04 (Jenett et al., 2012). The lines used to generate homozygotes of drn² lacking its maternal contribution were P{ry^+t7.2=neoFRT}82B ry⁵⁰⁶ (Bloomington Drosophila Stock Center, #2035) and w*;P{ry^+t7.2=neoFRT}82B P{w^+mC=ovoD1-18}3R/st¹βTub85D^Dss¹e^s/TM3, Sb¹ (Drosophila Genomics Resource Center, #106675).

All fly stocks were maintained on standard Drosophila medium at 25°C, unless stated otherwise. Mutant alleles of the second and third chromosomes were balanced with appropriate blue balancers, such as CyO, P{en1}wg^en11, TM6B, AbdA-lacZ and TM3, ftz-lacZ.

We generated the drn-deletion mutants drn¹ and drn² by the imprecise excision of P elements from GS12689 and GS10487, respectively (Toba et al., 1998). Imprecise excision was performed using a standard procedure described previously (Hummel and Klämbt, 2008). drn¹ and drn² mutations contain deletions from 9,353,341 to 9,355,167 and from 9,353,293 to 9,358,171, respectively (FlyBase2015_03, Dmel Release 6.06).

We obtained drn² homozygous embryos lacking the drn maternal contribution (drn^{2 m/z}) using standard crosses, as described previously (Prudencio and Guilgur, 2015). In brief, FRT82B was introduced into the drn² chromosome by recombination. Flies carrying FRT82B drn² were selected by Geneticin (Gibco) and genomic PCR using the primers 5′-TCACGCATTCAGAGCTTCGTGTGCCC-3′ and 5′-ATGTTGCTGCGTTTGCTCTGCGTATTCCAC-3′. FRT82B drn²/TM3b, Sb females were crossed with hsFLP/Y; FRT82B ovoD/TM3, Sb males to obtain hsFLP/+; FRT82B ovoD/FRT82B drn² females through heat-shock treatment. These females were then crossed with FRT82B drn²/TM3b, Sb males to obtain FRT82B drn² homozygous embryos without the drn maternal contribution.

To construct UAS-drn, a cDNA fragment composed of an entire open reading frame of a drn transcript (CG45050-RC) was PCR amplified using an upper strand primer containing an EcoRI site (5′-CCGGAATTCAGCAGGAAGCAGACGAAACT-3′) and a lower-strand primer containing HindIII and BglII sites (5′-CCCCAAGCTTAGATCTTCCTTGTTATAGCGCAGCAT-3′). The cDNA clone RE70963 was used as a template (Stapleton et al., 2002). The PCR product was digested with EcoRI and HindIII, subcloned into the EcoRI and HindIII sites of pBluescript, and sequenced (Agilent Technologies). The cloned fragment was subcloned into the EcoRI and BglII sites of the pUAST vector (Brand and Perrimon, 1993).

To construct UAS-myc-drn, a DNA fragment composed of an entire open reading frame of a drn transcript (CG45050-RC) was PCR amplified using RE70963 cDNA as a template, an upper strand primer containing EcoRI and BglII sites and the myc-tag coding sequence (underlined) (5′-CCGGAATTCCAAAATGGAGCAGAAGCTGATCTCGGAGGAGGATCTGAGATCTATGGAACGTGAATCTAACCC-3′), and a lower strand primer containing a XhoI site (5′-CCGCTCGAGTCAAATCTTTTGAATCTTCT-3′). CG45050-RC has the same open reading frame as CG45050-RB, -RD, -RE, -RF and -RG (Fig. 1D). The PCR product was digested with EcoRI and XhoI and subcloned into pUAST. The DNA sequence of the coding region was then confirmed. UAS-drn and UAS-myc-drn constructs were introduced into the Drosophila genome using P element-mediated transformation (Spradling and Rubin, 1982).

A fragment of drn cDNA (RE70963) containing an entire open reading frame of CG45050-RC was amplified by PCR, sequenced, and subcloned into the BamHI and EcoRI sites of the pGEX-2T vector (GE Healthcare Life Sciences). A GST-Drn fusion protein was produced in Origami B (DE3) cells (Novagen) and purified using a Glutathione Sepharose 4B column. The purified GST-Drn fusion protein was used to immunize rats, and polyclonal antiserum was purified using a standard protocol.

Embryos were immunostained as described previously using the following primary antibodies: mouse anti-Fas3 [1:100, Developmental Studies Hybridoma Bank (DSHB), 7G10], chicken anti-β-galactosidase (1:500, Abcam, ab9361), mouse anti-Pdi (1:200, Stressgen, 1D3), anti-lectin-PNA (1:500, Vector Laboratories), mouse anti-Sec5 (1:200, 22A2; Murthy et al., 2003), rabbit anti-GM130 (1:50, Abcam, ab30637), guinea pig anti-Rab5 (1:3000, gift from Akira Nakamura, Kumamoto University, Japan), rabbit anti-Rab7 (1:5000; Tanaka and Nakamura, 2008), rabbit anti-Rab11 (1:5000; Tanaka and Nakamura, 2008), guinea pig anti-Hrs (1:1000; Lloyd et al., 2002), mouse anti-multiubiquitin antibody (1:200, MBL, FK2), rabbit anti-LAMP1 antibody (1:1000, Abcam, ab30687), mouse anti-extra domain of Notch (1:500, DSHB, C458.2H), mouse anti-Frizzled 2 (1:20, DSHB, 1A3G4), rabbit anti-GFP (1:500, MBL, 598), rabbit anti-RFP (1:500, MBL, PM005), rat anti-GFP (1:500, Nacalai Tesque, 04404-26), mouse anti-Eve (1:20, DSHB, 3C10) and rat anti-Drn (1:500). The chitin-binding probe CBP546 was prepared from a bacterial expression construct using the protocol provided by Yinhua Zhang (New England Biolabs) (Dong et al., 2014). CBP546 (1:50) was added along with secondary antibodies for other primary antibodies to visualize the trachea. Images were generated using an LSM880 (Carl Zeiss) microscope and processed using Adobe Photoshop. For in situ hybridization, standard protocols were used, as described previously (Jiang et al., 1991). Images were obtained using an Axiosop2 Plus microscope (Carl Zeiss).

The same threshold was set for each pair of images (two channels representing respective markers) using ImageJ. Using the ‘Analyze Particles’ function, particles larger than five-pixel units were defined as the region of interest (Roi). Roi sets from two channels of an image were compared. Each Roi overlapping between two channels was defined as a particle demonstrating colocalization and was used for calculating the colocalization rate between two markers. The number of Rois showing colocalization against the total number of Rois was calculated as the percentage of colocalization.

We thank Dr Norbert Perrimon for providing the UAS-hop^Tuml fly line and Dr Stephane Noselli for providing the UAS-dome-GFP fly lines. Fly stocks were obtained from the Drosophila Genetic Resource Center at the Kyoto Institute of Technology and the Bloomington Drosophila Stock Center at Indiana University. We thank Dr Hugo Bellen and Dr Akira Nakamura for providing antibodies. Some antibodies used in this study were obtained from the Developmental Studies Hybridoma Bank (DSHB) at the University of Iowa.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201224.reviewer-comments.pdf