Multifunctional role of GPCR signaling in epithelial tube formation

The formation of three-dimensional tubes by invagination from flat epithelial sheets is a fundamental process in forming organs, such as lungs and kidneys (Andrew and Ewald, 2010). The Drosophila embryonic salivary gland (SG) is a premier model system to study the mechanisms underlying epithelial tube morphogenesis (Chung et al., 2014; Girdler and Röper, 2014). The SG begins as a two-dimensional plate of cells on the embryo surface. Neither cell division nor cell death occurs once the SG cells are specified; all the morphogenetic changes arise by changes in cell shape and rearrangement.

A major cell shape change during SG invagination is apical constriction, wherein the apical side of epithelial cells shrinks while keeping the volume nearly constant (Lubarsky and Krasnow, 2003; Martin and Goldstein, 2014). During stage 11, SG cells begin to invaginate at the dorsal/posterior region of the placode through apical constriction (Fig. 1A). This leads to the formation of a narrow invagination pit through which all SG cells eventually internalize. Apical constriction is observed in both invertebrates and vertebrates, driving morphogenetic processes ranging from gastrulation in Drosophila and Xenopus to lens formation in the mouse eye to formation of tubular organs such as the Drosophila SG and chicken lungs (Kim et al., 2013; Myat and Andrew, 2000; Plageman et al., 2011; Sweeton et al., 1991). Defects in apical constriction result in defects in overall tissue shape, suggesting that coordinated apical constriction is required for final tissue architecture (Chung et al., 2017; Guglielmi et al., 2015; Izquierdo et al., 2018).

Contractile actomyosin networks generate the cellular forces driving coordinated cell behaviors during epithelial morphogenesis (Gorfinkiel and Blanchard, 2011; Levayer and Lecuit, 2012; Munjal and Lecuit, 2014). Non-muscle myosin II (hereafter referred to as myosin) uses its motor activity to pull on actin filaments anchored to adherens junctions (AJs) to apply tension on the junction and neighboring cells. Three distinct pools of myosin generate the contractile forces required for SG invagination (Booth et al., 2014; Chung et al., 2017; Röper, 2012). Medioapical myosin forms in the center of the apical side of the SG cell and drives apical constriction (Booth et al., 2014; Chung et al., 2017; Röper, 2012). Junctional myosin, associated with AJs of SG cells, is involved in cell intercalation (Sanchez-Corrales et al., 2018). A supracellular myosin cable surrounding the entire SG placode generates the tissue-level compressing force (Chung et al., 2017; Röper, 2012). From studies in the past several years by us and others, we are beginning to understand how each myosin pool is created and regulated during SG invagination (Booth et al., 2014; Chung et al., 2017; Le and Chung, 2021; Röper, 2012; Sanchez-Corrales et al., 2018; Sidor et al., 2020).

Localized activation of the RhoA (Rho1 in Drosophila) GTPase is key to creating and/or polarizing contractile myosin structures in epithelial cells (Blanchard et al., 2010; Dawes-Hoang et al., 2005; Martin et al., 2009; Mason et al., 2013). Studies in Drosophila have identified the Folded gastrulation (Fog) pathway as a key signaling pathway regulating Rho1 signaling during epithelial morphogenesis (Manning and Rogers, 2014). During Drosophila gastrulation, two G protein-coupled receptors (GPCRs), Smog (ubiquitously expressed) and Mist (mesoderm-specific; also known as Mthl1), respond to Fog to activate heterotrimeric G proteins, Rho1 and Rho kinase (Rok), to regulate myosin contractility (Costa et al., 1994; Dawes-Hoang et al., 2005; Kerridge et al., 2016; Kölsch et al., 2007; Manning et al., 2013; Mason et al., 2013; Parks and Wieschaus, 1991). In our previous study in the SG, we also revealed a tissue-specific regulation of Fog signaling to activate myosin contractility: the Fork head (Fkh) transcription factor regulates SG upregulation of fog, which promotes accumulation of Rok and myosin specifically in the medioapical region of SG cells to drive clustered apical constriction (Chung et al., 2017; Fig. 1A). However, it is not yet known how the Fog signal is sensed and transduced into downstream Rho1 signaling during epithelial tube formation.

Studies in early Drosophila embryos have suggested a role of GPCR signaling as a common module for Rho1 activation in different subcellular regions of the cell. In the mesoderm, Smog and Mist function together to activate myosin in the medioapical region of cells to drive apical constriction (Kerridge et al., 2016). In the ectoderm, where no apical constriction occurs, Smog is required for junctional myosin activation to drive cell intercalation (Kerridge et al., 2016). Recent work has shown that distinct Rho guanine nucleotide exchange factors (RhoGEFs) activate myosin contractility in the apical and junctional domain of epithelial cells under the control of specific G proteins in early Drosophila embryos (Garcia De Las Bayonas et al., 2019). However, whether all three myosin pools in the SG are regulated by independent mechanisms or by a common module during epithelial tube formation remains unclear.

The contractile force of the actomyosin networks causes an increase of hydrostatic pressure in the cytoplasm (Charras et al., 2005). During epithelial morphogenesis, cells maintain the surface integrity by proper organization of actin at the cell cortex. Loss of organized cortical actin networks leads to the formation of blebs, spherical protrusions of the plasma membrane (Charras, 2008). During Drosophila gastrulation, blebs occasionally arise during apical constriction of mesodermal cells; these blebs correlate with apical filamentous actin (F-actin) holes (Costa et al., 1994; Jodoin et al., 2015). Disruption of cortical actin by latrunculin B, which sequesters globular actin (G-actin) and prevents F-actin assembly, enhances blebbing in epithelial cells in early Drosophila embryos (Kanesaki et al., 2013). Mutations in heterotrimeric G proteins also enhance blebbing (Kanesaki et al., 2013), suggesting a role of G proteins in stabilizing cortical actin. The mechanisms of cortical actin regulation during epithelial tube formation are not well understood.

Here, we reveal multiple roles of Smog GPCR in regulating actomyosin networks during Drosophila SG tube formation. We show that the GPCR Smog transduces Fog signal to regulate myosin contractility in the medioapical region of SG cells to drive apical constriction during SG invagination. Our study supports a model wherein Smog regulates medioapical and junctional/supracellular myosin pools in a Fog-dependent and -independent manner, respectively. We further reveal new roles of Smog in regulating the microtubule networks, key apical and junction proteins, and cortical actin organization during SG invagination, suggesting multifunctional roles of GPCR signaling during epithelial tube formation.

To test the role of Smog in transducing Fog signal during SG invagination, we first examined smog transcripts and Smog protein levels in the developing SG. Consistent with the ubiquitous expression of smog (Kerridge et al., 2016), fluorescent in situ hybridization revealed low levels of smog mRNA expression in the entire embryo, including the SG. Strong smog mRNA signals were observed in SG cells near the invagination pit (Fig. 1B,B′), potentially due to the shrinking apical domain in those cells. We also examined Smog protein localization in the SG using a functional Smog-GFP fusion protein expressed under the control of the sqh promoter (Kerridge et al., 2016). Smog-GFP signals were detected as small punctate structures enriched near AJs in SG cells [Fig. 1C-C″; AJs were marked with the key AJ protein E-Cadherin (E-Cad; Shg)]. Importantly, Smog-GFP signals were dramatically increased in the entire apical domain of Fog-overexpressing SGs (Fig. 1D-D″; quantification in Fig. 1E), suggesting that Smog is recruited to the apical domain of SG cells by overproduced Fog.

To test whether Smog transduces Fog signal in the SG to regulate myosin contractility, we performed a genetic suppression test. We hypothesized that if Smog is an SG receptor for Fog, knocking down smog in the SG should suppress the gain-of-function effect of Fog. We used sqh-GFP, a GFP-tagged version of myosin regulatory light chain (Royou et al., 2004), and Rok-GFP, a GFP-tagged Rok fusion protein (Abreu-Blanco et al., 2014), as readouts of Fog signaling. sqh-GFP signals were significantly increased in the medioapical region of Fog-overexpressing SG cells (compare Fig. 1G′,G‴ with F′,F‴; quantification in Fig. 1H) but not at AJs (Fig. 1I,J), suggesting increased medioapical myosin by Fog overexpression. Fog-overexpressing SG cells showed highly distorted AJs (Fig. 1G″) compared with control SG cells (Fig. 1F″), suggesting increased pulling forces due to increased medioapical myosin in Fog-overexpressing SG cells. Consistent with our finding that Fog signal promotes accumulation of Rok in the medioapical region of SG cells (Chung et al., 2017), quantification of areas occupied by Rok-GFP puncta revealed overaccumulated Rok-GFP signals in the medioapical region of Fog-overexpressing SG cells (compare Fig. 1L′,L‴ with K′,K‴; quantification in Fig. 1N). Knockdown of smog in the SG using RNA interference (RNAi) and the SG-specific fkh-Gal4 driver (Henderson and Andrew, 2000) both suppressed the medioapical accumulation of Rok-GFP signals and reduced the magnitude of AJ distortion (Fig. 1M-M‴; quantification in Fig. 1N,O). These data suggest that Smog transduces Fog signal in the SG to facilitate Rok and myosin accumulation in the medioapical region of SG cells.

To further test the role of Smog in transducing Fog signal to regulate myosin contractility, we performed an in vitro cell contraction assay using Drosophila S2 cells. S2 cells do not express Fog receptors and, therefore, do not respond to Fog in the normal culture condition (Manning et al., 2013; Fig. 2A-B′). As shown in a previous study (Manning et al., 2013), S2 cells transfected with Mist, the mesoderm-specific receptor for Fog, contracted robustly when cultured in Fog-containing media (Fig. 2C-D′). As Smog was not detectable in Drosophila S2 cells (modENCODE Cell Line Expression Data; Flybase), we tested whether S2 cells transfected with Smog also contract upon Fog treatment. Indeed, Smog-transfected cells showed a significantly higher percentage of contraction upon Fog treatment compared with non-transfected S2 cells (Fig. 2A-B′,E-F′; quantification in Fig. 2G). Thus, Smog can transduce Fog signal to regulate cellular contractility both in the SG and in vitro.

To test the role of Smog in apical constriction of SG cells as a Fog receptor, we knocked down smog in the SG (using fkh-Gal4) using RNAi and compared apical areas of SG cells with those in fog mutants. Two independent short hairpin RNAi lines, TRiP.HMC03192 (a stronger line; Fig. 3B,B′; also used for genetic suppression assay in Fig. 1M-M‴) and TRiP.GL01473 (a weaker line; Fig. S1G,G′), were used. smog knockdown using either line reduced Smog-GFP signals in the SG, with a stronger effect with TRiP.HMC03192, confirming the effect of the RNAi lines (Fig. S1A-D). Cell segmentation analysis revealed that SG cells with smaller apical areas showed a less coordinated spatial distribution in smog RNAi SGs compared with control (Fig. 3A-B′; Fig. S1F-G′), similar to the previously reported fog mutant phenotype (Chung et al., 2017; Fig. S1E,E′). However, quantification of percentage and cumulative percentage of cells of different apical areas revealed different behaviors between fog and smog. In fog mutants, the number of constricting cells is not significantly different from wild type (WT), but cells with small apical areas are dispersed across the entire SG placode (Chung et al., 2017). On the other hand, SGs with smog knockdown with either RNAi line showed larger apical areas compared with control (Fig. 3G; Fig. S1H). The difference in apical constriction defects between fog mutants and smog knockdown suggests a potential Fog-independent role of Smog.

As smog has a strong maternal contribution (Flybase Developmental RNA-seq), we also knocked down smog both maternally (using matα-Gal4; Häcker and Perrimon, 1998) and zygotically in the SG (using fkh-Gal4) and observed slightly more severe apical constriction defects in SG cells (hereafter referred to as smog M/Z knockdown; Fig. 3C,C′; quantification in Fig. 3G). Maternal knockdown of smog using the stronger RNAi line (TRiP.HMC03192) resulted in severe defects in egg-laying. Therefore, all M/Z knockdown experiments were performed using the weaker line (TRiP.GL01473). Consistent with the ubiquitous expression of smog and its role in epithelial morphogenesis in the early Drosophila embryo (Kerridge et al., 2016), smog M/Z knockdown resulted in a slight disruption of the overall embryonic morphology, including defects in the head region, a wavy embryo surface and an irregular ventral midline (Fig. 3M,N). Unlike the circular SG placode of WT, smog M/Z knockdown also resulted in a SG placode elongated along the dorsal/ventral axis (compare Fig. 3A,A′ with C,C′). Three-dimensional reconstruction of confocal images revealed disrupted epithelial morphology in smog M/Z knockdown, with wide and wavy grooves, enlarged epithelial cells and an elongated SG placode (Fig. 3I,J).

smog zygotic null mutants (Kerridge et al., 2016) had a range of defects similar to those observed with smog knockdown. Relatively normal embryonic morphology with mild apical constriction defects in SG cells was seen in 73.8% of embryos (59/80), like smog zygotic knockdown (classified as ‘smog null [mild]’; Fig. 3D-E′,K; quantification in Fig. 3H); 26.2% of embryos (21/80) showed a disrupted embryonic morphology, with SGs elongated along the dorsal/ventral axis (Fig. 3F,F′,L) and more severe apical constriction defects (Fig. 3H), like smog M/Z knockdown (classified as ‘smog null [severe]’). All smog null mutant embryos at late stages (stages 15-17) showed relatively normal embryonic morphology (Fig. S2A,B), suggesting that severely defective embryos with smog loss do not survive/develop past stage 14. Notably, these embryos have relatively normal internalized SGs (Fig. S2E-G), except for rare cases of crooked SG morphology (Fig. S2C,D). Altogether, these data suggest that Smog is required for coordinated apical constriction during SG invagination and proper epithelial morphogenesis during embryogenesis.

We next examined myosin levels and distribution in the SGs in smog knockdown or smog mutants using sqh-GFP. Medioapical myosin is the only myosin pool defective in the SGs of fog mutants (Chung et al., 2017; Fig. S3A-E). Consistent with the idea that Smog transduces Fog signal in the SG, smog knockdown (either zygotic or M/Z knockdown) and smog loss resulted in a significant reduction in the intensity of medioapical myosin in SG cells near the invagination pit, compared with control (Fig. 4A-D″; quantification in Fig. 4I). Moreover, compared with clear web-like structures of medioapical myosin in control SG cells (Fig. 4A-A″), SG cells in smog knockdown or smog null mutants showed reduced areas of sqh-GFP puncta (Fig. 4L), with dispersed myosin along the entire apical surface in smog M/Z knockdown SG cells (Fig. 4C-C″). Interestingly, the intensity of junctional myosin was also significantly reduced in SG cells with smog knockdown or smog loss, compared with control (Fig. 4A-D″; quantification in Fig. 4J). This phenotype is distinct from what was observed in fog mutants, where junctional myosin was unaffected in SG cells (Chung et al., 2017; Fig. S3D). The ratio of medioapical to junctional myosin intensity was significantly reduced in smog M/Z knockdown but not in smog zygotic knockdown or smog null [mild] SGs (Fig. 4K). We conclude that Smog is required for maintaining normal levels of both medioapical and junctional myosin in the SG in a Fog-dependent and -independent manner, respectively.

We also tested Rok distribution in SG cells in smog RNAi and smog null mutants using Rok-GFP. Whereas Rok-GFP formed large punctate structures in the medioapical region in control SG cells (Fig. 4E-E″), Rok-GFP failed to accumulate and was dispersed along the entire apical domain in SG cells in smog knockdown and smog null mutants (Fig. 4F-H″; quantification in Fig. 4M), similar to the dispersed Rok-GFP distribution observed in fog mutant SG cells (Chung et al., 2017). These data suggest that Smog is required for Rok accumulation in the medioapical domain of SG cells.

Reduced myosin levels and the elongated SG placode in smog null [severe] embryos led us to test the third pool of myosin in the SG, the supracellular myosin cable surrounding the SG placode. Consistent with the idea that this myosin cable is under tension (Chung et al., 2017; Röper, 2012), control SGs showed a relatively smooth tissue boundary (Fig. 4N-N″). smog null [mild] embryos also maintained a smooth SG boundary comparable with WT (Fig. 4O-O″). However, SGs in smog null [severe] embryos showed an irregular boundary and a less clear myosin cable (Fig. 4P-P″).

We calculated the circularity of the SG placode as a measure of smoothness and tension of the SG boundary. The circularity of the WT placode was significantly higher than the circularity of one cell row inside (inner boundary) or outside of the SG placode (outer boundary), suggesting a higher tension of the myosin cable in WT (Fig. 4Q,R). The circularity of the placode of smog null [mild] embryos was comparable with WT (Fig. 4R). However, the circularity of the SG placode in smog null [severe] embryos was significantly lower than that of WT and was not statistically significant from the inner or outer boundaries, suggesting a lack of tension at SG boundary in smog null [severe] mutants (Fig. 4R). Consistent with our previous study (Chung et al., 2017), fog mutants did not show defects in the myosin cable (Fig. S3F-H), even in the severely twisted embryos with extra grooves and folds. Overall, our data suggest that, unlike fog, the loss of smog affects all myosin pools in the SG, including the myosin at the SG boundary, leading to reduced tension at the SG boundary and defects in generating the compressing force.

As we observed a range of defects in whole embryo morphology in smog mutants, we explored the potential cellular basis. In stage 11 smog null [severe] embryos, we occasionally observed large areas in the epidermis where E-Cad (Fig. 5B,B′; compare with WT in Fig. 5A,A′) and the apical determinant protein Crumbs (Crb) (Fig. 5D,D′; compare with WT in Fig. 5C,C′) signals were absent or significantly reduced. Relatively strong signals of the cell membrane marker Gap43-mCherry (Martin et al., 2010) were observed in these regions (Fig. 5E-E″), suggesting that the cell membrane is still intact. Cells in these regions had enlarged apical areas compared with neighboring cells with strong E-Cad signals (Fig. 5E-E″), suggesting reduced epithelial tension and/or integrity in cells with reduced E-Cad levels. Similarly, in late-stage (stage 14) embryos, Crb levels were significantly reduced in large areas; in the most severe cases, Crb was dispersed in the entire epidermis (Fig. S4B,B′). These data suggest that Smog is required for maintaining epithelial polarity and junctional integrity by regulating E-Cad and Crb levels or distribution.

To better understand the basis of the reduced E-Cad and Crb in smog mutants, we examined smog null mutant SG cells at higher magnification. Even in SGs that did not contain large patches of E-Cad/Crb loss, E-Cad/Crb signals were discontinuous, with small gaps in signal along the cell boundary (Fig. 5F-O). Such gaps were observed in both smog null [mild] and smog null [severe] embryos. Although the number and the length of gaps in E-Cad signals in smog null [mild] were comparable with those of WT, they were significantly higher in smog null [severe] compared with WT (Fig. 5G,H; quantification in Fig. 5L,M). The number and the length of gaps in Crb signals in SG cells were significantly higher in both mild and severe smog null mutant embryos compared with WT (Fig. 5J,K; quantification in Fig. 5N,O). These data suggest a role of Smog in regulating Crb and E-Cad levels and continuity during epithelial morphogenesis, including the SG.

Anisotropic localization of Crb at the SG boundary has been suggested to drive the formation of supracellular myosin cable in SG (Röper, 2012). Consistent with this, in control SGs, Crb levels were higher in inside junctions that do not contribute to the myosin cable, whereas myosin was highly enriched at the junction forming the cable at the SG boundary (Fig. S4C-E‴). However, in smog null [severe] SGs, the overall Crb intensity was quite low, and Crb did not show an anisotropic localization at the SG boundary (Fig. S4F-H‴). These data suggest that smog loss results in the reduction of both myosin and Crb, failing to form the supracellular myosin cable at the SG boundary.

Microtubules are required for forming and maintaining apical myosin structures (Booth et al., 2014; Ko et al., 2019) and for transporting several key apical and junctional proteins in Drosophila tubular organs, including E-Cad and Crb (Le and Chung, 2021; le Droguen et al., 2015). We recently showed that microtubule-dependent intracellular trafficking has an important role in regulating apical constriction and medioapical myosin during SG invagination (Le and Chung, 2021). To examine the microtubule networks in the SG of WT and smog mutant embryos, we stained with antibodies against acetylated α-tubulin (Ace-tub), a marker of stable, long-lived microtubules, and tyrosinated α-tubulin (Tyr-tub), a marker of dynamic, short-lived microtubules (Westermann and Weber, 2003). Compared with WT, both Ace-tub and Tyr-tub signals were reduced in smog mutants (compare Fig. 5Q,R,U,V with P,T), with a dramatic reduction in Ace-tub levels in smog null [severe] embryos (Fig. 5R), suggesting disrupted microtubule networks in smog mutants. The reduced Ace-tub and Tyr-tub signals were more prominent in the SG compared with the surrounding epithelial cells; the ratio of Ace-tub and Tyr-tub intensity in SG cells to the intensity in cells outside the SG was significantly decreased in smog mutants compared with WT (Fig. 5X,Y). A slight but less prominent reduction in Ace-tub and Tyr-tub signals was also observed in the SG in fog mutants (Fig. 5S,W; quantification in Fig. 5X,Y). Overall, this finding suggests that loss of smog could affect apical pools of myosin, Crb and E-Cad through effects on microtubule abundance, distribution and/or polarity.

We observed numerous blebs in the apical membrane of SG cells in smog null [severe] mutants, which were most prominent near the invagination pit (compare Fig. 6A-A‴ with B-B‴). Such blebs were not observed in SGs in WT, smog null [mild] or fog mutant embryos (Fig. S5A-C″). Rok-GFP was enriched in many of these blebs (Fig. 6B-B‴), consistent with the recruitment of Rok to retracting blebs (Aoki et al., 2016). In smog null [severe] embryos, blebs were also occasionally observed in cells outside the SG placode (Fig. S5D-E″). During Drosophila gastrulation, bleb formation correlates with cortical F-actin holes (Jodoin et al., 2015), and disruption of cortical actin enhances blebbing (Kanesaki et al., 2013). We therefore tested whether Smog affects the cortical actin network during apical constriction in the SG. Indeed, Phalloidin staining revealed that, whereas strong F-actin signals were observed in the apical domain of SG cells near the invagination pit in WT, F-actin was reduced in SGs in smog mutants (Fig. 6C-E).

Apical F-actin formation in Drosophila tubular organs requires the formin family actin nucleator Diaphanous (Dia) (Massarwa et al., 2009; Rousso et al., 2013). As the apical localization and activity of Dia are crucial for restricting F-actin formation to the correct membrane domain, we tested the localization and levels of Dia in the SG in smog mutants. Similar to reduced F-actin (Fig. 6C-E), we observed reduced Dia signals in the apical domain of SG cells in smog mutants, compared with WT (Fig. 6G-I). Quantification of Dia levels revealed that the overall Dia intensity is reduced in the SG placode in smog mutants, with a stronger effect in the smog null [severe] than smog null [mild] embryos (Fig. 6K). SGs in fog mutants appeared to show a slight reduction of F-actin (Fig. 6F), but Dia levels in fog mutants were comparable with WT (Fig. 6J,K). Our data suggest that Smog is required for the cortical actin networks during SG invagination, at least in part via modulating Dia levels.

We next tested whether the reduction in actin levels in smog mutants is due to defects in actin turnover. To test this, we modulated the level of actin regulators in the SG in smog mutants and assayed the blebbing phenotype. Using RNAi, we knocked down chickadee (chic), which encodes Drosophila profilin, a protein that increases F-actin by promoting actin polymerization (Cooley et al., 1992), in the smog mutant background. If there is a shift from F-actin to monomeric G-actin in smog mutants, we expect enhancement of the disorganized actin phenotype and bleb formation with a reduced chic level. Blebs were not observed in smog null [mild] embryos or in SGs knocked down for chic in the otherwise WT background (Fig. 6L-M″). However, when chic was knocked down in the SG in the smog mutant background, we observed enhanced blebbing even in SG cells in smog null [mild] embryos (Fig. 6N-N″). As profilin promotes formin-mediated actin filament assembly (Romero et al., 2004; Zweifel and Courtemanche, 2020), our data suggest that reduced profilin following chic knockdown aggravated defects in cortical actin organization with reduced Dia in smog mutants. Overall, our data suggest that Smog is required for cortical actin organization as well as myosin activation during epithelial morphogenesis.

Fog signaling triggers epithelial cell shape changes driving tissue folding and invagination during development of Drosophila and other insects (Benton et al., 2019; Manning and Rogers, 2014). During Drosophila embryogenesis, fog is upregulated in multiple tissues undergoing apical constriction and tissue invagination, such as ventral furrow (mesoderm), the posterior midgut (endoderm) and the SG (ectodermal derivative) (Nikolaidou and Barrett, 2004). During Drosophila gastrulation, the ubiquitously expressed GPCR Smog and the mesoderm-specific GPCR Mist respond to Fog signaling to regulate myosin contractility during mesoderm invagination (Kerridge et al., 2016; Manning et al., 2013). Here, we provide evidence that Smog functions as an SG receptor for Fog to activate myosin during SG invagination. Knockdown of smog by RNAi suppresses the gain-of-function effect of Fog in the SG (Fig. 1). Also, Smog-transfected S2 cells contract upon Fog signal (Fig. 2), suggesting that Smog responds to Fog to regulate myosin contractility both in vivo and in vitro. Our data show that Fog overexpression leads to Smog recruitment to the medioapical region of SG cells (Fig. 1). Consistent with our data, Fog overexpression induces oligomerization of Smog in early Drosophila embryos (Jha et al., 2018).

Our study also dissects the roles of Smog in myosin activation during SG morphogenesis and reveals Fog-independent roles of Smog. Whereas fog mutants exhibit a decrease in only the medioapical myosin pool in SG cells (Chung et al., 2017; Fig. S3), smog loss and SG-specific knockdown of smog result in a significant reduction of both medioapical and junctional myosin (Fig. 4). We propose that Smog regulates junctional myosin during epithelial morphogenesis in response to an unknown, ubiquitously expressed ligand and regulates medioapical myosin in response to Fog in tissues with high Fog signals, such as the mesoderm (Kerridge et al., 2016; Manning et al., 2013) and the SG (this study). Consistent with this model, during Drosophila gastrulation, Smog activates different myosin pools in tissues with different Fog levels. In the mesoderm, where Fog levels are high, Smog and Mist transduce Fog signal to activate myosin in the medioapical region and drive apical constriction. In the ectoderm, where Fog levels are very low, Smog is required for junctional myosin activation to drive cell intercalation. SG cells undergo both apical constriction and cell intercalation, and our data show that Smog regulates both pools of myosin in the SG in a Fog-dependent and -independent manner. It is possible that in tissues with high Fog signals, an additional tissue-specific GPCR(s) functions with Smog in response to Fog to fine-tune the Fog signaling for proper apical constriction, like Mist does in the mesoderm. It will be interesting to determine whether additional SG-specific GPCRs function with Smog during SG invagination.

The downstream components that transduce signals from the Smog GPCR to regulate myosin in distinct regions of SG cells await discovery. Good candidates include two RhoGEFs, RhoGEF2 and Dp114RhoGEF, which have been shown to activate Rho1 signaling at the medioapical and junctional domain of epithelial cells, respectively (Garcia De Las Bayonas et al., 2019). This process requires upstream activation by heterotrimeric G proteins Cta (Gα_12/13) and Gβ13F/Gγ1 at the medioapical and junctional domains, respectively (Garcia De Las Bayonas et al., 2019). Thus, distinct ligand-receptor binding with Smog may activate distinct G proteins and/or RhoGEFs in the medioapical and junctional domains.

We also discover that Smog regulates cortical F-actin organization during SG invagination (Fig. 6). smog loss significantly reduces levels of F-actin and the formin protein Dia (Fig. 6). Rho1 regulates actin polymerization via Dia (Goode and Eck, 2007), and apical F-actin formation in the Drosophila SG requires apical localization and activity of Dia (Massarwa et al., 2009; Rousso et al., 2013). Our data suggest that Smog activates Rho1 signaling to regulate both cortical F-actin (via Dia) and myosin (via Rok) (Fig. 7). fog mutants also show slightly reduced F-actin and Dia levels, to a similar level to smog null [mild] embryos (Fig. 7), and these embryos do not show membrane blebbing (Fig. 7; Fig. S5). It is possible that Smog regulates cortical actin in both Fog-dependent and independent manners and perhaps subtle changes in Dia and F-actin levels in fog mutants are not enough to cause membrane blebbing. Alternatively, Smog's role in regulating cortical actin is independent of Fog; reduced F-actin levels in fog mutants may be a secondary effect due to a defective medioapical myosin pool. During Drosophila gastrulation, fog mutants show normal GFP-Moesin signals in the ventral furrow, suggesting normal cortical actin organization upon fog loss (Kanesaki et al., 2013). Cta is a key component downstream of Fog that recruits RhoGEF2 to the apical membrane to regulate myosin contractility (Barrett et al., 1997; Dawes-Hoang et al., 2005). Interestingly, mutations in Gαi (G-iα65A) and Gβ13F/Gγ1, but not Cta, disrupt cortical actin organization in early Drosophila embryos (Fox and Peifer, 2007; Kanesaki et al., 2013). Therefore, Smog may signal through Gαi and/or Gβ13F/Gγ1 to organize cortical actin, independent of Fog and Cta.

The link between heteromeric G proteins and Rho1 signaling occurs through activation of specific RhoGEFs (Vázquez-Victorio et al., 2016). Actin polymerization resulting from Fog-independent Smog activation may imply the recruitment of distinct proteins modulating the activation of Rho1-mediated pathways. For example, in the absence of Fog, a specific RhoGEF(s) downstream of Gαi or Gβ13F/Gγ1 may link Smog and Rho1 signaling to regulate cortical actin organization; upon high Fog signals, a different Rho1 modulator(s) downstream of Cta may be recruited to facilitate medioapical myosin formation. Further work is needed to identify the link between Smog and Rho1 activation in the presence or absence of Fog during SG invagination. It will be also interesting to determine whether Fog-independent roles of Smog – regulating cortical actin and junctional myosin – involve common or distinct Rho1 modulators.

smog null mutant embryos show a range of defects, allowing us to dissect the roles of Smog in epithelial morphogenesis. Our data reveal that Smog is required for epithelial integrity by affecting levels and localization of Crb and E-Cad (Fig. 5). As the decrease in the level of Crb and E-Cad in smog mutants correlates with significantly reduced myosin levels, disrupted epithelial integrity may be due to disrupted actomyosin networks. Myosin II is crucial for cells to concentrate E-cad at cell-cell contacts (Shewan et al., 2005). In early Drosophila embryos, the absence of contractile myosin leads to the disassembly of junctions (Weng and Wieschaus, 2016). A possible scenario is that Smog is required to recruit key Rho1 regulators, such as RhoGEF2, Dp114RhoGEF or other SG-upregulated RhoGEFs, to AJs during SG invagination to promote junctional myosin assembly. As Dia helps coordinate adhesion and contractility of actomyosin (Homem and Peifer, 2008) and formin-mediated actin polymerization at AJs stabilizes E-Cad and maintains epithelial integrity (Rao and Zaidel-Bar, 2016), reduced Dia levels in smog mutants may also contribute to loss of epithelial integrity.

Reduced and discontinuous signals of apical and junctional components in smog mutant embryos can also be due, in part, to compromised intracellular trafficking. Disorganized microtubule networks in SGs in smog mutants (Fig. 5) support this idea. Our recent study showed that microtubule- and Rab11-dependent intracellular trafficking regulates apical myosin pools and apical constriction during SG invagination, via apical enrichment of Fog and the continuous distribution of Crb and E-Cad along junctions (Le and Chung, 2021). The mechanism of how Smog affects microtubule organization remains to be revealed. fog mutants also show slightly disrupted microtubule networks, albeit less prominently than smog null [severe] (Fig. 5). Whether this is a direct effect of loss of fog or an indirect result of the disrupted medioapical pool of myosin remains to be answered. Overall, our findings suggest multifaceted roles of Smog during epithelial tube formation in regulating distinct myosin pools and cortical actin in different subcellular domains in a ligand-dependent manner.

The fly stocks used in this study are listed in Table S1. All the crosses were performed at 25°C.

Drosophila embryos were collected on grape juice agar plates supplemented with yeast paste at 25°C. Embryos were dechorionated with 50% bleach. For most samples, embryos were fixed in 1:1 heptane: formaldehyde for 40 min and devitellinized with 80% ethanol. For Rok-GFP, sqh-GFP and Phalloidin staining, embryos were hand-devitellinized. Embryos were then stained with primary and secondary antibodies in PBSTB (1× PBS with 0.1% Triton X-100 and 0.2% bovine serum albumin) and PBTr (1× PBS with 0.1% Triton X-100) for ethanol devitalization and hand-devitalization, respectively. Antibodies used in our experiments are listed in Table S2. Embryos were mounted in Aqua-Poly/Mount (Polysciences) and imaged using a Leica TCS SP8 confocal microscope using 63×, NA 1.4 and 40×, NA 1.3 objectives. Images were acquired as z-stacks (each 0.3 µm apart) that span the apical and junctional domains of cells in the SG placode.

Embryos were fixed using formaldehyde/heptane for 30 min followed by devitellinization with methanol. SP6 polymerase-synthesized digoxigenin (DIG)-labeled antisense probe was prepared using the following primers: smog-F, 5′-ACAGAGCCCACCTGTGTAGG-3′; smog-R, 5′-TCGCTGATCGAAAATGATCTC-3′. Fluorescent in situ hybridization was performed using standard methods as described in Knirr et al. (1999). Briefly, the embryos were pre-hybridized for 1 h at 56°C post-fixation. Following this step, the embryos were hybridized with the probe overnight at 56°C. The next day, embryos were stained for DIG using the anti-DIG primary antibody and biotin-conjugated secondary antibody. The embryos were then incubated with AB solution (PK-6100, Vectastain Elite ABC kit). Tyramide signal amplification reaction was performed using Cy3 fluorescent dye (diluted 1:50 using amplification diluent) (Perkin Elmer Life Sciences, NEL753001kt). The embryos were co-stained with CrebA and Crb antibodies for visualizing SG nuclei and apical cell boundaries, respectively.

For Smog-GFP quantification, maximum intensity projections of two z-sections that span the apical and the junctional region of SG cells were used. Intensity means of Smog-GFP signals of SG cells in the entire SG placode were measured using Imaris software (Bitplane). The mean intensity of Smog-GFP was normalized by the median deviation. The integrated density of Smog-GFP was calculated by multiplying the apical area of the cell with the mean intensity of each cell. Total Smog-GFP signals within the whole placode were calculated as the sum of the integrated density of all cells. For background correction, mean gray values of Smog-GFP in ten cells outside of the SG placode were measured using ImageJ (National Institutes of Health). The average value of mean gray values of Smog-GFP in these ten cells was used to subtract the background of the cells inside the placode from the same embryo. Five SGs were used for quantification for each genotype except for control in Fig. 1E, where four SGs were used. Statistical significance was determined using Mann–Whitney U-test in the GraphPad Prism software.

For total Dia intensity quantification, maximum intensity projection of two z-sections of the most apical region of SG cells was used. The images were then exported to ImageJ and converted to the RGB stack image type. Mean gray values of Dia signals in the whole placode were measured. For background subtraction, the average mean gray value of Dia signals for ten cells outside the SG placode was subtracted from the mean gray value of Dia intensity of the whole SG placode. These values were scaled by the mean normalization. The total Dia intensity in the whole placode was calculated by multiplying the normalized intensity with the entire area of the placode. Statistical significance was determined using Mann–Whitney U-test.

SGs that were invaginated within the range of 4.8-9.9 µm depth were used for quantification. Two or three z-sections of the apical domain were used to generate a maximum intensity projection (Leica LasX software). SG cells were marked using CrebA and segmented along E-Cad signals using the Imaris program. Apical areas of segmented cells were calculated using Imaris, and cells were color-coded based on their apical domain size. Frequency distribution was performed using GraphPad Prism. Apical areas from eight SGs were quantified in control, smog knockdown (Z), and smog null [mild] embryos. Six SGs were used for quantification for smog knockdown (M/Z) and smog null [severe] embryos. Statistical significance was determined using Mann–Whitney U-test (percentage of cells) and Kolmogorov–Smirnov test (cumulative percentage of cells).

Maximum intensity projection of two z-sections spanning apical and junctional regions of SG cells with sqh-GFP and E-Cad signals was used. The images were then exported to ImageJ and converted to RGB stack image type. Twenty cells near the invagination pit in each of five SGs were used for quantification. To calculate the medioapical and junctional myosin intensity, regions were drawn manually along the inner and outer boundary of the E-Cad signal of each cell, and mean gray values were measured for medioapical and junctional myosin. For background subtraction, the average mean gray value of sqh-GFP signals for ten cells outside the SG placode was subtracted from the mean gray value of medioapical and junctional myosin for each SG cell. SuperPlots (Lord et al., 2020) were used to address the variability of datasets in each SG. Each data point in the graph represents one SG consisting of twenty cells used for quantification. Statistical significance was determined using Welch's t-test. A normal distribution was tested using Jarque-Bera test (Microsoft Excel) or a bell-shape of data distribution by making a histogram.

A single z-section of the confocal image with the strongest Rok-GFP signals in the medioapical region was used. Rok-GFP signals were converted to black and white using the threshold tool in Adobe Photoshop. Junctional Rok-GFP signals were removed manually based on E-Cad signals. Areas of Rok-GFP puncta were determined using the Analyze Particles tool in ImageJ. Rok-GFP puncta with area ≥0.02 µm² were measured. Fifteen cells in the dorsal posterior region near the invagination pit of the SG placode were used for quantification. The same strategy was used for sqh-GFP signals to quantify the area of medioapical myosin. SuperPlots show mean values of data points of five SGs. Statistical significance was determined using Mann–Whitney U-test.

Using a single z-section with the highest E-Cad signals and ImageJ software, the shortest (L₀) and the actual distance (L) between vertices were measured in SG cells. The ratio of these distances (L/L₀) was used as the waviness of junctions. We used 10-15 cells near the invagination pit in each of five SGs for quantification. Statistical significance was determined using Mann–Whitney U-test.

Using confocal images and ImageJ, cell boundaries were manually drawn along E-Cad signals at the SG placode boundary and one cell row outside and inside the placode. For most samples, strong myosin cable signals at the dorsal, anterior and posterior boundaries of the placode and CrebA signals in SG cells were used to determine the SG boundary. In smog null [severe], where myosin signals were significantly reduced in all cells, CrebA signals were used to determine the boundary. The ventral midline was used as the ventral boundary of the SG in all cases. The perimeter and the area of the SG corresponding to these boundaries were measured using ImageJ, and circularity was calculated using the formula, C=4π area/perimeter². For a perfect circle, the value of circularity should be 1. Five SGs were used for quantification, and statistical significance was determined using Welch's t-test.

The number and the length of gaps for Crb and E-Cad signals were measured using confocal images and ImageJ. A single z-section with the highest Crb or E-Cad signals was chosen for quantification. Gaps with a length ≥0.2 µm were used for quantification. All junctions from ten cells in five SGs were quantified. If there were multiple gaps on a junction, we added the length of all the gaps on that junction to calculate the ratio of the length of gaps to junctional length. Statistical significance was determined using Mann–Whitney U-test.

Three z-sections spanning the most apical region of SG cells were used for quantification. The SG boundary was drawn manually based on the E-Cad and CrebA signals. The mean gray value of Ace-tub signals in the whole SG placode and ten cells outside the SG of the same embryo was measured using ImageJ. The ratio of the mean gray value of Tyr-tub signals inside the placode to the average mean gray value of cells outside the placode was calculated and plotted. The same method was used for Tyr-tub. Five SGs for each genotype were used for quantification. Mann–Whitney U-test was used to calculate the P-values.

cDNA clones for smog (RE70685) and mist (RE13854) were obtained from Drosophila Genomics Resource Center (DGRC). Isolated cDNAs were cloned in-frame with the V5 tag into the pMT-V5-HisB vector (Addgene plasmid #17589) using the conventional restriction digestion and ligation method. Primers with restriction sites used to isolate the open reading frame are as follows (restriction sites underlined): smog CDS-EcoRI-5′, 5′-CCGGAATTCATGGAACTGTGCATAGCAAC-3′; smog CDS-NotI-3′, 5′-ATTTGCGGCCGCATTGGTCGTGATTGTATCTTTGG-3′; mist CDS-EcoRI-5′, 5′-CCGGAATTCATGGACAGGAGTCGGAGTAGC-3′; mist CDS-NotI-3′, 5′-ATTTGCGGCCGCAGCAAATGGTCTCCATTTTG-3′.

S2 and S2-Mt-Fog-myc cells (Manning et al., 2013) were obtained from DGRC. Cells were tested for contamination regularly. Shields and Sang M3 insect media (HiMedia) was used with 10% fetal bovine serum (Corning) and 1:1000 Antibiotic-Antimycotic (Gibco). Cells were cultivated at 25°C either in a 25 cm² flask in 5 ml media or a 75 cm² flask in 15 ml media. To prepare for the Fog media, S2-Mt-Fog-myc cells were grown undisturbed for 4-8 days to attain ∼100% confluency. Then, 50 µl of 100 mM CuSO₄ was added to nearly 100% confluent cells in a 75 cm² flask to induce the metallothionein promoter. The Fog-containing media was collected by centrifugation at 4°C (2500 g) and further concentrated using 3000 MW concentrators. The presence of Fog was confirmed by western blot using the Myc-antibody (Invitrogen; Table S2). Non-transfected S2 cells were used as a control.

The cell contractility assay was performed as described in Manning et al. (2013). Smog- or Mist-transfected S2 cells were induced with CuSO₄. Cells were then transferred to Concanavalin A-coated coverslips and fixed using 10% paraformaldehyde in PBTr for 15 min at room temperature. After being blocked in 5% normal goat serum in PBTr for 20 min at room temperature, cells were stained with the V5 antibody (Invitrogen; Table S2) to differentiate between transfected and non-transfected cells. Cell contractility of transfected cells was monitored using a phase-contrast filter in a Leica DM2500 microscope using the 40×, 0.8 NA objective. Three independent experiments were performed. Two-way ANOVA was performed to calculate statistical significance.

We thank the members of the Chung laboratory for their comments and suggestions. We thank A. Martin, T. Lecuit and the Bloomington Drosophila Stock Center for fly stocks, and D. J. Andrew and the Developmental Studies Hybridoma Bank for antibodies. We thank Flybase for the gene information. We are grateful to D. J. Andrew, C. D. Hanlon, J. Kim and J. Matthew for their helpful comments on the manuscript.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200519