Wingless counteracts epithelial folding by increasing mechanical tension at basal cell edges in Drosophila

Animals undergo a series of shape changes during development to reach their final form. The reshaping of developing animals is mainly accounted for by alterations in the shape of epithelial sheets of cells that line the outside of the animal and its internal organs. Changes in the shape of epithelia depend in part on the coordinated deformation of spatially defined subsets of cells. Cell deformations require the modulation of mechanical forces that constrict or expand cell edges and surfaces. For example, modulations of mechanical forces acting at the apical or basal side of cells can result in apical cell constriction or basal cell expansion that drive the folding of epithelial sheets (Martin and Goldstein, 2014; Pearl et al., 2017). However, the issue of how tissue-level mechanical forces are modulated in space and time remains poorly understood.

In recent decades, studies have revealed the importance of signal transduction pathways in instructing the spatiotemporal pattern of gene expression and thereby cell fate specification in embryos and tissues. The Wnt/Wingless signal transduction pathway plays an evolutionarily conserved and widespread role in controlling growth and specifying cell fates in animal development. The Wnt/Wingless ligand primarily acts through binding to plasma membrane receptors of the Frizzled family to ultimately activate or repress the transcription of target genes by modulating the activity of the transcription factor T-cell factor (TCF) (Nusse and Clevers, 2017). However, whether and how signal transduction pathways such as Wnt/Wingless also control the generation of mechanical forces in cells is largely unexplored.

The larval Drosophila wing disc, a single-cell-layered epithelial sheet, has been extensively used in the past to study signal transduction pathways and their influence on gene expression and cell fate specification (Beira and Paro, 2016). Wnt/Wingless plays an important role in the early growth of the wing disc and its regionalization during larval development (Ruiz-Losada et al., 2018). The wing disc gives rise to three regions of the adult fly: the wing blade, the hinge and part of the body wall (notum) (Cohen, 1993). Wnt/Wingless is required during larval development for the specification of the wing blade, the wing blade margin and the distal part of the hinge (Ng et al., 1996; Whitworth and Russell, 2003; Williams et al., 1993). During mid-larval development, the initially flat wing disc epithelium generates three stereotypic folds within the prospective hinge region: the hinge/notum (H/N) fold, separating the prospective hinge and notum regions of the wing disc; the hinge/pouch (H/P) fold, separating the prospective hinge and pouch (blade) regions; and a central hinge/hinge (H/H) fold that is located in between the other two folds (Cohen, 1993) (Fig. 1A). We recently characterized the mechanisms by which the H/H fold forms. A local reduction of extracellular matrix (ECM) density beneath prospective fold cells leads to decreased mechanical tension at basal cell edges (henceforth referred to as basal edge tension) and, subsequently, to a basal widening of these cells that drives epithelial folding (Sui et al., 2018). Basal edge tension results from the contraction of a basal actomyosin network and an elastic straining of the ECM during tissue growth (Sui et al., 2018). How basal edge tension is specifically reduced in the cells that will form the H/H fold is unknown.

Here, we show that the absence of Wingless expression in the central hinge region contributes to low basal edge tension in this region. Forced expression of Wingless in this region results in high basal edge tension and suppresses H/H fold formation. Moreover, in the wing disc pouch, where Wingless is normally expressed, Wingless is necessary and sufficient for high basal edge tension. Our data reveal that the spatial pattern of Wingless expression is converted into a spatial pattern of mechanical tension that is important for proper wing disc morphogenesis.

To test whether Wingless plays a role during H/H fold formation, we analyzed the expression pattern of Wingless protein in wing discs during mid-larval development [68–88 h after egg lay (AEL)] using immunohistochemistry. At 68 h AEL, Wingless protein was mainly detectable in the prospective notum region and in parts of the pouch region of the wing disc (Fig. S1A), as described previously (Ng et al., 1996; Whitworth and Russell, 2003; Williams et al., 1993). As development proceeded, Wingless expression became confined to broad stripes of cells in the prospective notum region and along the dorsoventral compartment boundary in the pouch region (Fig. 1B, Fig. S1B-F). In addition, Wingless expression gradually coalesced into a ring of distal hinge cells surrounding the pouch region. Interestingly, at 84 h AEL the proximal edge of the ring-like expression domain of Wingless neighbored the cells that initiated folding and later gave rise to the H/H fold (Fig. 1B, Fig. S1E). Closer examination showed that the fold cells were positioned just outside of the detectable Wingless protein gradient emanating from this ring of distal hinge cells (Fig. S1G,H). Thus, cells forming the H/H fold are characterized by little, if any, Wingless expression.

To test whether the absence of Wingless expression in the central hinge is important for H/H fold formation, we ectopically expressed Wingless throughout the hinge region using the zfh-Gal4 driver line (Whitworth and Russell, 2003). Control wing discs expressing the marker CD8-mCherry under the control of zfh-Gal4 formed a normal H/H fold (Fig. 1C,C′). By contrast, expression of Wingless under the control of zfh-Gal4 resulted in wing discs that lacked an H/H fold; the hinge region was flat (Fig. 1D,D′). H/N and H/P folds formed under these conditions, although the latter fold appeared shallower compared with controls (Fig. 1D,D′). Similarly, the expression of a constitutively active form of β-catenin (Arm^S10), a transducer of Wingless signaling (Pai et al., 1997), under the control of zfh-Gal4, resulted in wing discs that lacked an H/H fold (Fig. S1M). Conversely, lowering Wingless expression by RNA interference using a double-stranded RNA targeting wingless under the control of zfh-Gal4 resulted in a deeper H/H fold (Fig. 1E,E′). Finally, the expression of Wingless or Arm^S10 within a part of the hinge region led to a long-range and cell-autonomous inhibition of H/H fold formation, respectively (Fig. S1I-L). Taken together, these data show that Wingless signal transduction suppresses H/H fold formation. These results further indicate that the Wingless protein gradient emanating from the distal hinge cells contributes to the precise positioning of the H/H fold.

To test whether a local reduction of Wingless signal transduction is sufficient to promote fold formation, we inhibited Wingless signal transduction locally in a stripe of cells along the anteroposterior compartment boundary by expressing a dominant-negative form of Tcf, Tcf^DN (van de Wetering et al., 1997), under the control of the dpp-Gal4 driver line. In wild-type wing discs, the Wingless signal is transduced in the wing disc pouch (Alexandre et al., 2014; Neumann and Cohen, 1997; Zecca et al., 1996); expression of Tcf^DN thus locally reduces Wingless signal transduction activity along the anteroposterior compartment boundary in this region (see schematic in Fig. 1F). Interestingly, wing discs expressing Tcf^DN under the control of dpp-Gal4 formed an ectopic fold along the anteroposterior compartment boundary specifically in the pouch region (Fig. 1G,G′, Fig. S2). Thus, a local reduction of Wingless signal transduction is sufficient to promote folding in the wing disc pouch.

How does Wingless signal transduction suppress H/H fold formation? Cells forming the H/H fold are characterized by a reduced density of βPS-integrin, an anchor for ECM components (Maartens and Brown, 2015), and a wider basal cross-sectional area compared with cells neighboring this fold (Sui et al., 2018). A reduction in βPS-integrin density contributes to a reduction in ECM density, which results in a local decrease of basal edge tension and consequently the basal widening of cells. A computational model indicates that the widening of the basal side of cells can account for H/H fold formation (Sui et al., 2018). We therefore tested whether Wingless expression influences integrin density, basal cell area and basal edge tension. To this end, we reduced Wingless expression by RNA interference in the posterior compartment of the wing disc pouch, leaving the anterior compartment as a control. Wingless protein levels were strongly reduced in the posterior compartment compared with the control anterior compartment (Fig. 2A,A′), indicating that the RNA interference was efficient. βPS-integrin density at the basal cell membrane was reduced in the posterior compartment compared with the control anterior compartment (Fig. 2B,B′,D, Fig. S3). We next measured the apical and basal cross-sectional areas of cells in these wing discs. In the control anterior compartment, the apical cross-sectional area was larger than the basal cross-sectional area, resulting in an overall bending of the wing disc towards its basal side (Fig. 2A″,E,F) (Sui et al., 2018). Moreover, whereas the apical cross-sectional area of posterior and anterior cells was comparable (Fig. 2C-C‴,E), the basal cross-sectional area of posterior cells, in which Wingless expression was reduced, was increased approximately twofold compared with the basal cross-sectional area of control anterior cells (Fig. 2B″,B‴,F). Because posterior cells had a larger basal cross-sectional area compared with apical cross-sectional area, the posterior compartment showed an overall bending towards its apical side (Fig. 2A″, Fig. S4). Furthermore, cells in the posterior compartment displayed a reduced height compared with cells in the control anterior compartment (Fig. 2A″). This result is consistent with previous studies showing that a reduction of Wingless signal transduction activity (Widmann and Dahmann, 2009), or reductions of integrin expression or ECM density (Dominguez-Gimenez et al., 2007), all result in the flattening of the columnar wing disc epithelium. A reduction of Wingless signal transduction activity has a similar effect on cell and tissue shape as reductions of integrin expression or ECM density, raising the possibility that Wingless signal transduction controls cell shape by modulating integrin or ECM.

Finally, we measured the mechanical tension along basal cell edges and along adherens junctions (referred to here as apical cell edges). To achieve this, we ablated single edges at the apical and basal sides of cells using laser light and measured the resulting displacement of the two vertices at the ends of the ablated cell edge. The initial speed of vertex displacement (i.e. the recoil velocity, see Materials and Methods) is a relative measure of mechanical tension on the cell edge before it was ablated (Ma et al., 2009). The final displacements and recoil velocities resulting from the ablation of apical cell edges of anterior and posterior cells were similar (Fig. 2G,I, Fig. S5A,B), indicating that the reduction of Wingless expression does not influence apical edge tension. Moreover, final displacement and recoil velocity were similar upon the ablation of basal edges of wild-type anterior and posterior cells (Fig. 2H,J, Fig. S5C,D), showing that basal edge tension is normally similar for anterior and posterior cells. Interestingly, final displacement was approximately halved and recoil velocity was decreased by approximately 60% in posterior cells in which Wingless was depleted compared with control anterior cells (Fig. 2H,K, Fig. S5E,F). Taken together, we conclude that Wingless expression is required to maintain a small basal cross-sectional area of cells, as well as to maintain high βPS-integrin density and high basal edge tension.

We next tested whether increasing Wingless signal transduction was sufficient to reduce the basal area of cells, to increase βPS-integrin density and to elevate basal edge tension. To this end, we expressed an activated form of Arm, Arm^S10, in a stripe of cells perpendicular to the hinge folds under the control of dpp-Gal4 (Fig. 3A). We quantified βPS-integrin density, basal area, and basal edge tension of cells expressing Arm^S10 and of neighboring control cells, which were not transducing Wingless and forming the H/H fold. Similar to the expression of Arm^S10 throughout the hinge region (see Fig. S1M), expression of Arm^S10 in a stripe of cells suppressed H/H fold formation (Fig. 3B,B′). The H/P fold formed normally (Fig. 3B). βPS-integrin density was increased in Arm^S10-expressing cells, unlike their control neighbors (Fig. 3C′,D′,E). Moreover, basal cross-sectional area was decreased by approximately 80% in Arm^S10-expressing cells, in contrast to adjacent control H/H fold cells (Fig. 3C,D,F). Lastly, final displacement and recoil velocity were increased 2- to 3-fold in Arm^S10-expressing cells, unlike their control neighbors (Fig. 3G,H; Fig. S5G,H). Thus, Wingless signal transduction is sufficient to induce basal cell constriction and high basal edge tension in hinge cells that normally do not transduce the Wingless signal.

We propose a model in which Wingless signal transduction promotes high basal edge tension, which results in the basal constriction of cells (Fig. 4). The expression of Wingless in the pouch and distal hinge region suppresses epithelial folding by counteracting the basal expansion (widening) of cells. The low or absent Wingless signal transduction in the central hinge therefore results in low basal edge tension and basal cell widening, which leads to the formation of the H/H fold. How does Wingless signal transduction promote high basal edge tension? Similar to the reduction of Wingless expression, lowering βPS-integrin activity resulted in a deeper H/H fold (Fig. S6). Moreover, increased H/H fold depth, caused by reduced Wingless expression, was partly suppressed by the overexpression of βPS-integrin (Fig. S6), suggesting that Wingless promotes high basal edge tension and counteracts H/H fold formation in part through integrins. Integrins may stabilize the basal actomyosin network to generate active tension or provide cellular anchors required for the elastic straining of the ECM during tissue growth that generates passive tension (Sui et al., 2018). In addition, Wingless signal transduction may also promote basal edge tension through its ability to drive tissue growth (Neumann and Cohen, 1996). Differential growth rates have recently been shown to contribute to the precise positioning of folds within epithelia (Tozluoglu et al., 2019). Basal edge tension is also high in the proximal hinge region in the absence of discernible Wingless expression (Fig. 1B; Sui et al., 2018). We therefore speculate that an additional signal(s) is required to increase basal edge tension in this region. We note that high levels of Wingless expression specifically block the H/H fold, but not the neighboring H/P fold. H/P fold formation involves pulses of F-actin and increased tension along the lateral edges of cells, a mechanism that is independent of changes in basal edge tension (Sui et al., 2018). Moreover, alterations of Wingless expression did not noticeably alter apical edge tension (Figs 2 and 3). These results emphasize the specificity for Wingless signal transduction in modulating ECM-based basal edge tension in wing discs and further stress the important role of spatiotemporal modulation of ECM density in epithelial morphogenesis (Pastor-Pareja and Xu, 2011; Ramos-Lewis and Page-McCaw, 2019).

The following fly stocks were used: indy-GFP [a GFP protein trap in indy (YC0017); Quiñones-Coello et al., 2007], E-Cad-GFP (Huang et al., 2009), UAS-CD8-mCherry (Umetsu et al., 2014), dpp-Gal4 (a gift from E. Knust, Max Planck Institute of Molecular Cell Biology and Genetics, Germany), tub-Gal80^ts (McGuire et al., 2003), zfh-Gal4 (BDSC line 25676), UAS-wg (Simmonds et al., 2001), UAS-arm^S10 (Pai et al., 1997), UAS-wg^ds-RNA (VDRC line 39676), UAS-Tcf^DN (BDSC line 4784), UAS-p35 (Hay et al., 1994) and hh-Gal4 (Tanimoto et al., 2000).

The genotypes of larvae were as follows: Fig. 1B: E-Cad-GFP; Fig. 1C: UAS-CD8-mCherry/+;; zfh-Gal4/+; Fig. 1D: UAS-CD8-mCherry /+; UAS-wg/+; zfh-Gal4/+; Fig. 1E: UAS-CD8-mCherry /+; UAS-wg^ds-RNA/+; zfh-Gal4/+ (larvae were incubated at 25°C and transferred to 29°C for 24 h before dissection); Fig. 1G: UAS-p35/UAS-Tcf^DN; UAS-CD8-mCherry /dpp-Gal4, tubGal80^ts (larvae were incubated at 18°C and transferred to 29°C for 48 h before dissection); Fig. 2A-C: UAS-CD8-mCherry /+; UAS-wg^ds-RNA/hh-Gal4 (larvae were incubated at 25°C and transferred to 29°C for 24 h before dissection); Fig. 2H,J: Indy-GFP/y; UAS-CD8-mCherry/+; hh-Gal4/+; Fig. 2G-I,K: Indy-GFP/y; UAS-CD8-mCherry/+; UAS-wg^ds-RNA/hh-Gal4 (larvae were incubated at 25°C and transferred to 29°C for 24 h before dissection); Fig. 3B-D: UAS-arm^S10/+; UAS-CD8-mCherry/+; dpp-Gal4/+; Fig. 3G,H: Indy-GFP/UAS-arm^S10; UAS-CD8-mCherry/+; dpp-Gal4/+. Other fly stocks and genotypes are listed in supplementary Materials and Methods.

Wing discs were dissected, fixed and stained according to standard protocols (Klein, 2008). Primary antibodies used were rat anti-DE-cadherin [DCAD2, Developmental Studies Hybridoma Bank (DSHB), 1:50], mouse anti-βPS-integrin (DSHB, 1:200), mouse anti-Wingless (DSHB, 1:200) and rabbit anti-cleaved caspase 3 (D175, Cell Signaling, 1:200). Secondary antibodies (Invitrogen), all diluted 1:200, were anti-rat Cy5 (A10525), anti-mouse Alexa 633 (A21052) and anti-mouse Alexa 488 (A11001). Alexa Fluor 488 phalloidin (Invitrogen, A12379, 1:200) and rhodamine phalloidin (Invitrogen, R415, 1:200) were used to detect F-actin. For imaging fixed samples, wing discs were mounted using double-sided tape (Tesa 05338, Beiersdorf, Hamburg) as a spacer between the microscope slide and the coverslip to avoid flattening of the tissue. Images were acquired using Leica SP5 MP and Zeiss LSM 880 confocal scanning microscopes. Image stacks from apical to basal were taken with sections 1 μm apart. For Fig. 2B and Fig. 3B-D, wing discs were mounted with the basal side facing towards the objective. Image stacks were taken from basal to apical with sections 1 μm apart.

The apical cell mesh was obtained by projecting 3-5 slices of apical z-stacks showing E-Cadherin staining using the maximum intensity projection tool in Fiji (Schindelin et al., 2012). Basal βPS-integrin intensity images and the basal cell mesh, as identified by F-actin staining, were obtained by projecting 3-5 basal z slices of the image stacks with maximum intensity projection. Basal βPS-integrin intensity images were then overlaid with the basal cell mesh. The basal cell area and the average basal βPS-integrin intensity per cell were measured by using the freehand selection tool of Fiji.

The curvature of wing discs was measured by Kappa, a curvature analysis program available as a plugin within the Fiji software (Schindelin et al., 2012). The anterior and posterior basal surfaces of cross-sections of wing discs were traced by 12 equidistant points with a B-spline curve. The average curvature value was determined.

Wingless intensity was measured from maximum-intensity z-projections. A rectangular region of interest was defined and the average intensity was measured using the plot function of Fiji (Schindelin et al., 2012). The background pixel intensity was subtracted from the plot function.

For laser ablation experiments, wing discs were mounted in culture medium as described previously (Sui et al., 2018). Cell edges were visualized using Indy-GFP. For cutting basal cell edges, the basal side was facing the objective. For cutting apical cell edges, the apical side was facing the objective. Images were acquired, and laser ablations were performed, on a Multiphoton Laser Scanning Microscope Zeiss LSM 710 NLO using a C-Apochromat 40×/1.2W objective. The ablation was performed with approximately 60-70 mW of average power (50%) at 800 nm at single cell edges. The images were recorded every 0.2 s except for the first image after laser ablation, which was recorded 0.4 s after ablation. The vertex displacement after laser ablation was analyzed with Fiji. The two vertices of the ablated cell edges were manually tracked in the recorded images and the vertex distance increase over time was measured. The recoil velocity was obtained by measuring the vertex distance increase between the time point before ablation and the first image acquired 0.4 s after ablation, and dividing by 0.4 s. The recoil velocity is taken as a measure of relative mechanical tension on the cell edge before ablation (Ma et al., 2009).

A two-tailed, unpaired Student's t-test was used for statistical analysis.

We thank E. Knust, the Bloomington Drosophila Stock Center and the Vienna Drosophila Resource Center for fly stocks. We are grateful to E. Knust for helpful comments on the manuscript. We thank the light microscopy facilities of the Biotechnology Center at the TU Dresden and of the Max Planck Institute of Molecular Cell Biology and Genetics for technical assistance.

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.184713.reviewer-comments.pdf