ABSTRACT
The replacement of cells is a common strategy during animal development. In the Drosophila pupal abdomen, larval epidermal cells (LECs) are replaced by adult progenitor cells (histoblasts). Previous work showed that interactions between histoblasts and LECs result in apoptotic extrusion of LECs during early pupal development. Extrusion of cells is closely preceded by caspase activation and is executed by contraction of a cortical actomyosin cable. Here, we identify a population of LECs that extrudes independently of the presence of histoblasts during late pupal development. Extrusion of these LECs is not closely preceded by caspase activation, involves a pulsatile medial actomyosin network, and correlates with a developmental time period when mechanical tension and E-cadherin turnover at adherens junctions is particularly high. Our work reveals a developmental switch in the cell extrusion mechanism that correlates with changes in tissue mechanical properties.
INTRODUCTION
Epithelia play important roles in animal development by providing mechanical support and acting as an effective barrier against pathogens. The mechanical robustness and integrity of epithelia depends on adherens junctions that mechanically couple neighboring cells. At the same time, epithelia are also dynamic structures with cells in constant flux. This plasticity of epithelia depends on the ability of cells to remodel their adherens junctions (Takeichi, 2014). Adherens junctions are molecularly composed of E-cadherin cell-cell adhesion molecules that form, with their ectodomains, homophilic interactions between neighboring cells. The intracellular region of E-cadherin is linked via catenins to the underlying cortical actomyosin network. Contraction of the cortical actomyosin network can exert tensile stresses on cell junctions and can influence E-cadherin turnover (Iyer et al., 2019). The interplay between actomyosin network contractility and E-cadherin turnover contributes to the dynamics of junctional remodeling. Junctional remodeling plays important roles during cell division, tissue elongation or the expulsion of cells from tissues through extrusion (Guillot and Lecuit, 2013).
Cell extrusion is an important process for the maintenance of tissue homeostasis and function and its dysregulation can lead to disease (Fadul and Rosenblatt, 2018; Gudipaty and Rosenblatt, 2017; Slattum and Rosenblatt, 2014). Once a tissue has achieved its final size, superfluous cells generated by continuous cell proliferation are extruded from the tissue. Examples of this cell behavior have been demonstrated in the mammalian gut, fins of developing zebrafish, and the midline of the thoracic segment of Drosophila (Eisenhoffer et al., 2012; Marinari et al., 2012). Moreover, topological defects in epithelia can result in cell death and extrusion (Saw et al., 2017). Furthermore, cells that are mis-specified maybe extruded, as they could harm the proper development or function of a tissue. Cells harboring mutations in BMP2/4 signal transduction components are, for example, extruded from Drosophila epithelia (Gibson and Perrimon, 2005; Shen and Dahmann, 2005). Moreover, cells can extrude as part of a developmental program whereby one cell type is replaced by another (Ninov et al., 2007).
The development of the adult epidermis of the Drosophila abdomen is a suitable system in which to study the mechanisms underlying cell type replacement by cell extrusion. The abdominal segments are initially covered during larval stages by large polyploid larval epidermal cells (LECs) (Madhavan and Madhavan, 1980; Ninov et al., 2007; Roseland and Schneiderman, 1979). The LECs are then replaced by diploid adult epidermal progenitor cells, termed histoblasts, during pupal metamorphosis (Fig. 1A-C). During embryonic development, four small groups of histoblasts, called histoblast nests, are laid down in each abdominal segment on the left and the right side of the embryo. Histoblasts are quiescent during larval development, but then start to proliferate during pupal development. Histoblast proliferation coincides with the extrusion of the LECs neighboring the histoblast nests, cell by cell. LECs undergo caspase-dependent apoptosis that triggers the formation of an apical actomyosin cable at the level of adherens junctions in the extruding cell and its neighbors (Kester and Nambu, 2011; Nakajima et al., 2011; Ninov et al., 2007; Teng et al., 2017). Contraction of this cable contributes to the extrusion of single LECs from the basal side of the epithelium, where the cell corpses are then cleared by circulating hemocytes (Ninov et al., 2007; Teng et al., 2017). Once all LECs have been extruded, the cells of the growing left and right histoblast nests make contact at the ventral and dorsal midlines of the pupa to form a continuous epidermal sheet of cells. Proliferation and active migration of histoblasts is coordinated with the extrusion of LECs to maintain the integrity of the tissue. It has been reported that the majority of LECs extrude while in contact with a histoblast nest (Ninov et al., 2007). Moreover, it has been proposed that signals between histoblasts and LECs trigger LEC extrusion, thereby coordinating histoblast proliferation with LEC extrusion (Ninov et al., 2007).
Here, we identify a population of LECs that extrudes from the pupal abdominal epidermis by a mechanism that is independent of the presence of histoblasts. These LECs frequently extrude at the dorsal midline where left and right abdominal segments contact each other. Cell extrusions at the dorsal midline involve pulsatile contractions that correlate with accumulations of medial Myosin II. These extrusions take place during late pupal development, a stage when mechanical tension and E-cadherin turnover at adherens junctions of LECs are elevated.
RESULTS
LECs extrude at borders of histoblast nests and at the dorsal midline
The replacement of LECs by histoblasts results in the remodeling of the dorsal and ventral epidermis of the pupal abdomen (Madhavan and Madhavan, 1980; Ninov et al., 2007). Previous work has mainly focused on the extrusion of LECs at the border of histoblast nests that begins around 15 h after puparium formation (APF) (Nakajima et al., 2011; Ninov et al., 2007, 2010; Teng et al., 2017). To test whether LEC extrusion may also occur later at additional sites, we imaged entire hemisegments of the pupal abdomen between 20 h and 32 h APF at 3-min intervals using E-cadherin-GFP as a marker for adherens junctions (Fig. 1D-F′, Movie 1). To follow the extrusion of LECs during development, we segmented images to identify cell outlines and tracked LECs over time. We observed extrusion of LECs at the border of histoblast nests (Fig. 1G-I, Movie 1), as described previously (Koenderink and Paluch, 2018; Nakajima et al., 2011; Ninov et al., 2007, 2010; Teng et al., 2017). In addition, LECs extruded at additional positions within the segments, including in the vicinity of the dorsal midline (Fig. 1G-I, Movie 1), where left and right hemisegments of the abdominal epidermis meet. We defined three subsets of positions at which LECs extruded: (1) border of histoblast nest: LECs that at the time of extrusion were in contact with a histoblast nest (and >50 µm away from the dorsal midline); (2) dorsal midline: LECs that at the time of extrusion were <50 µm away from the dorsal midline and were not in contact with a histoblast nest; and (3) ‘in between’: LECs that extruded at positions other than those defined in (1) and (2).
LECs extrude at the border of histoblast nests and at the dorsal midline. (A-C) Schemes of a dorsal view of a pupa depicting the dorsal histoblast nests. Larval epidermal cells (LECs) are shown in blue and histoblasts in gray. Anterior and posterior histoblasts of abdominal segment 3 are highlighted in yellow and purple, respectively. The dashed line indicates the dorsal midline separating left and right hemisegments. The rectangle in A depicts the approximate region of the pupa that is shown in the subsequent microscopic images. B and C show the cellular organization within a hemisegment. Anterior and posterior histoblasts are initially separated by large polyploid LECs (B). The LECs are then extruded and replaced by histoblasts (C). (D-F′) Images from a time-lapse movie (D-F) showing dorsal views of an abdominal epidermal segment of a pupa at the indicated times after puparium formation (APF). Adherens junctions are visualized by E-cadherin-GFP. D′-F′ show segmentations. In these and all subsequent images dorsal is up and anterior is left, unless otherwise stated. The dashed line indicates the dorsal midline. (G) The image shown in D with colored dots indicating the position of LEC extrusion for the indicated time periods. (H) The segmentation shown in D′ with color code (as in G) indicating the time the LEC will extrude. (I) The segmentation shown in D′ with color code indicating whether the LEC will extrude at the dorsal midline, border of histoblast nest or in between. (J) Cumulative number of LECs extruding at the dorsal midline, at the border of the histoblast nest or in between as a function of time. Mean and s.e.m. are shown (n=4 hemisegments of 3 pupae). Scale bars: 50 µm.
LECs extrude at the border of histoblast nests and at the dorsal midline. (A-C) Schemes of a dorsal view of a pupa depicting the dorsal histoblast nests. Larval epidermal cells (LECs) are shown in blue and histoblasts in gray. Anterior and posterior histoblasts of abdominal segment 3 are highlighted in yellow and purple, respectively. The dashed line indicates the dorsal midline separating left and right hemisegments. The rectangle in A depicts the approximate region of the pupa that is shown in the subsequent microscopic images. B and C show the cellular organization within a hemisegment. Anterior and posterior histoblasts are initially separated by large polyploid LECs (B). The LECs are then extruded and replaced by histoblasts (C). (D-F′) Images from a time-lapse movie (D-F) showing dorsal views of an abdominal epidermal segment of a pupa at the indicated times after puparium formation (APF). Adherens junctions are visualized by E-cadherin-GFP. D′-F′ show segmentations. In these and all subsequent images dorsal is up and anterior is left, unless otherwise stated. The dashed line indicates the dorsal midline. (G) The image shown in D with colored dots indicating the position of LEC extrusion for the indicated time periods. (H) The segmentation shown in D′ with color code (as in G) indicating the time the LEC will extrude. (I) The segmentation shown in D′ with color code indicating whether the LEC will extrude at the dorsal midline, border of histoblast nest or in between. (J) Cumulative number of LECs extruding at the dorsal midline, at the border of the histoblast nest or in between as a function of time. Mean and s.e.m. are shown (n=4 hemisegments of 3 pupae). Scale bars: 50 µm.
To quantify the relative number of LECs extruding at the border of histoblast nests, at the dorsal midline and in between, we calculated the cumulative number of extrusions for each subset of cells. A similar number of LECs extruded at the border of histoblast nests and at the dorsal midline during the analyzed time window (Fig. 1J). The onset of extrusion at the dorsal midline was around 20-22 h APF, and thus later than the onset of extrusion at the border of histoblast nests (Fig. 1J). We conclude that LECs extrude not only at the border of histoblast nests, but also at other positions within the abdominal epidermis, including at the dorsal midline. We went on to investigate the mechanisms by which cells extrude at the dorsal midline.
LEC extrusion at the dorsal midline depends on caspase-dependent apoptosis
The extrusion of LECs at the border of histoblast nests requires caspase-dependent apoptosis (Kester and Nambu, 2011; Nakajima et al., 2011; Ninov et al., 2007; Roseland and Schneiderman, 1979; Teng et al., 2017). To test whether LEC extrusion at the dorsal midline also depends on apoptosis, we blocked apoptosis by the expression of P35, an apoptosis inhibitor (Hay et al., 1994). We expressed P35 using the Gal4-UAS system under control of the tsh-Gal4 driver, which is active in LECs of the pupal epidermis (Nakajima et al., 2011). We restricted expression of P35 to the pupal stage by using a temperature-sensitive form of Gal80, encoded by gal80ts, a repressor of Gal4. P35 expression was induced at 0 h APF by shifting pupae to the restrictive temperature for gal80ts of 30°C. Compared with controls, P35 expression resulted in a strong delay of LEC extrusion and the number of LECs extruding at the border of histoblast nests and at the dorsal midline were reduced (Fig. 2A-D′,G, Movie 2). In controls, LECs had been almost entirely replaced by histoblasts at 27.5 h APF (Fig. 2E,E′). By contrast, when P35 was expressed, numerous LECs persisted at this time point (Fig. 2F,F′). Adult flies hatching from pupae expressing P35 under the control of tsh-Gal4 showed unpigmented regions in the abdominal epidermis (Fig. S1A,B), indicating that some LECs were not replaced by histoblasts under these circumstances. Taken together, these results show that LEC extrusion at the dorsal midline depends on caspase-dependent apoptosis.
LEC extrusion at the dorsal midline depends on caspase activity. (A-F′) Images from time-lapse movies (A-F) showing dorsal views of an abdominal epidermal segment of a control pupa (A,C,E) or a pupa expressing P35 in LECs (B,D,F) at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. A′-F′ show segmentations with color code indicating cells that extrude at the border of histoblast nests (red) or at the dorsal midline (green). Scale bars: 50 µm. (G) Cumulative number of LEC extrusions as a function of time for control and P35-expressing LECs. Mean and s.e.m. are shown (control: n=3 hemisegments for dorsal midline and 2 hemisegments for border of histoblast nest; P35: n=4 hemisegments for dorsal midline and 2 hemisegments for border of histoblast nest).
LEC extrusion at the dorsal midline depends on caspase activity. (A-F′) Images from time-lapse movies (A-F) showing dorsal views of an abdominal epidermal segment of a control pupa (A,C,E) or a pupa expressing P35 in LECs (B,D,F) at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. A′-F′ show segmentations with color code indicating cells that extrude at the border of histoblast nests (red) or at the dorsal midline (green). Scale bars: 50 µm. (G) Cumulative number of LEC extrusions as a function of time for control and P35-expressing LECs. Mean and s.e.m. are shown (control: n=3 hemisegments for dorsal midline and 2 hemisegments for border of histoblast nest; P35: n=4 hemisegments for dorsal midline and 2 hemisegments for border of histoblast nest).
LECs at the dorsal midline uniformly activate caspases during late pupal development
Caspase activation precedes the onset of apical constriction of LECs extruding at the border of histoblast nests by, on average, approximately 20 min (Teng et al., 2017). To test whether extrusion of LECs located at the dorsal midline is also preceded by caspase activation, we monitored by live imaging the activity of Apoliner, a sensor for caspase activity (Bardet et al., 2008), between 18 h and 29 h APF in the abdominal epidermis. In this sensor, two fluorophores, mRFP and eGFP, are linked by a caspase-specific cleavage site. Caspase activation leads to the cleavage of Apoliner and to the translocation of the eGFP from the cytoplasm into the nucleus. Apoliner cleavage was detected in LECs located at the border of histoblast nests that commenced extrusion at subsequent time points, consistent with previous findings (Fig. 3A-F) (Nakajima et al., 2011; Teng et al., 2017). LECs located at the dorsal midline showed tissue-wide fluctuations of Apoliner cleavage between 18 h and 26 h APF (Fig. 3G-I,K-M, Movie 3). From 26 h APF onwards, nearly all LECs, regardless of their position within the abdominal epithelium, showed cleaved Apoliner (Fig. 3J,M, Movie 3). Thus, caspase is uniformly activated throughout the population of LECs at late pupal development.
LECs at the dorsal midline uniformly activate caspases during late pupal development. (A-D,G-J) Images from time-lapse movies showing dorsal views of abdominal epidermal segments expressing Apoliner. Images show LECs at the border of a histoblast nest (A-D) and an entire hemisegment (G-J). Times APF are indicated. Green arrowheads mark LECs activating caspase and extruding. The dashed line indicates the dorsal midline. Scale bars: 50 µm. (E,F,K,L) Diagrams illustrating the cells shown in A-D and G-J. Gray circles represent nuclei. (M) The fraction of LECs displaying cleaved Apoliner for cells at the indicated locations and times APF. Mean and s.e.m. are shown (n=3 pupae).
LECs at the dorsal midline uniformly activate caspases during late pupal development. (A-D,G-J) Images from time-lapse movies showing dorsal views of abdominal epidermal segments expressing Apoliner. Images show LECs at the border of a histoblast nest (A-D) and an entire hemisegment (G-J). Times APF are indicated. Green arrowheads mark LECs activating caspase and extruding. The dashed line indicates the dorsal midline. Scale bars: 50 µm. (E,F,K,L) Diagrams illustrating the cells shown in A-D and G-J. Gray circles represent nuclei. (M) The fraction of LECs displaying cleaved Apoliner for cells at the indicated locations and times APF. Mean and s.e.m. are shown (n=3 pupae).
LECs extruding during late development show fluctuations in apical cell area and Myosin
During LEC extrusion at the border of histoblast nests at early developmental stages, both LECs and neighboring cells accumulate cortical non-muscle Myosin II (Myosin) (Teng et al., 2017). Cortical Myosin accumulation depends on caspase activity and Myosin activity is required for the shrinkage of the apical LEC surface and extrusion (Teng et al., 2017). We therefore tested whether LEC extrusion at later stages also involves cortical Myosin accumulation. Myosin is a hexamer containing two copies each of the Myosin heavy chain, Myosin light chain and Myosin regulatory light chain (encoded by sqh) (Koenderink and Paluch, 2018). We used Sqh-GFP and Utr-GFP (the F-actin-binding domain of Utrophin fused to GFP) to visualize Myosin and F-actin, respectively, by time-lapse video microscopy in the pupal abdomen. During early pupal development (22-23 h APF), LECs extruding at the border of histoblast nests or at the dorsal midline accumulated cortical actomyosin (Fig. 4A-H, Fig. S2A-D, Movie 4), consistent with previous results (Teng et al., 2017). By contrast, LECs extruding at later time (30 h APF), located both at the border of histoblast nests and at the dorsal midline, accumulated Myosin and F-actin in punctae within the central region of the cell at the level of adherens junctions (referred to as ‘medial’ Myosin; Fig. 4I-P, Fig. S2E-H, Movie 4). Whereas the apical cross-sectional area of extruding LECs at 23 h APF decreased approximately continuously (Fig. 4Q), the apical cross-sectional area of cells extruding at 30 h APF fluctuated over time showing pulsatile contractions (Fig. 4R). Medial and cortical Myosin intensities fluctuated over time in extruding LECs undergoing pulsatile contractions at 30 h APF (Fig. 4S,U,V, Movie 5). Fluctuations of medial, but not cortical, Myosin intensities were negatively correlated with the apical cross-sectional area of extruding cells (Fig. 4W). Similar fluctuations were not observed in LECs that extruded early during development (Fig. 4T,X,Y, Movie 5). Taken together, these data indicate two distinct mechanisms of LEC extrusion: (1) during early pupal development, LEC extrusion both at the border of the histoblast nest and at the dorsal midline involves cortical Myosin accumulation (Teng et al., 2017; Fig. 4); (2) during late pupal stages, LEC extrusion involves anti-correlated fluctuations of apical cross-sectional cell area and medial Myosin levels.
LECs accumulate Myosin II during extrusion at the dorsal midline. (A-P) Images from time-lapse movies showing dorsal views of abdominal epidermal segments expressing Sqh-GFP. Time to extrusion of the cells indicated by the green arrowheads is shown. Images in A-H and I-P show the extrusion of LECs at the indicated position at 22-23 h and 30 h APF, respectively. Scale bars: 20 µm. (Q,R) Apical cross-sectional area of LECs extruding at the border of histoblast nests or at the dorsal midline normalized to their initial apical cross-sectional area as a function of time to extrusion. Area changes for individual LECs are shown. Note the fluctuation in area for LECs extruding at the dorsal midline. (S,T) Images from higher speed time-lapse movies showing dorsal views of abdominal epidermal segments of pupae at 30 h APF (S) or 22 h APF (T) expressing Sqh-GFP. Time is shown relative to extrusion. Scale bars: 5 µm. (U,V) Apical cross-sectional area and medial (U) or cortical (V) Sqh-GFP pixel intensity of a LEC extruding at the dorsal midline as a function of time relative to extrusion at 30 h APF. (W) Cross correlation between apical cross-sectional area and medial or cortical Sqh-GFP intensity of extruding LECs at 30 h APF as a function of lag time. Mean and s.e.m. are shown. n=30 extruding LECs. (X,Y) Apical cross-sectional area and medial (X) or cortical (Y) Sqh-GFP pixel intensity of a LEC extruding at the dorsal midline as a function of time relative to extrusion at 22 h APF. a.u., arbitrary units.
LECs accumulate Myosin II during extrusion at the dorsal midline. (A-P) Images from time-lapse movies showing dorsal views of abdominal epidermal segments expressing Sqh-GFP. Time to extrusion of the cells indicated by the green arrowheads is shown. Images in A-H and I-P show the extrusion of LECs at the indicated position at 22-23 h and 30 h APF, respectively. Scale bars: 20 µm. (Q,R) Apical cross-sectional area of LECs extruding at the border of histoblast nests or at the dorsal midline normalized to their initial apical cross-sectional area as a function of time to extrusion. Area changes for individual LECs are shown. Note the fluctuation in area for LECs extruding at the dorsal midline. (S,T) Images from higher speed time-lapse movies showing dorsal views of abdominal epidermal segments of pupae at 30 h APF (S) or 22 h APF (T) expressing Sqh-GFP. Time is shown relative to extrusion. Scale bars: 5 µm. (U,V) Apical cross-sectional area and medial (U) or cortical (V) Sqh-GFP pixel intensity of a LEC extruding at the dorsal midline as a function of time relative to extrusion at 30 h APF. (W) Cross correlation between apical cross-sectional area and medial or cortical Sqh-GFP intensity of extruding LECs at 30 h APF as a function of lag time. Mean and s.e.m. are shown. n=30 extruding LECs. (X,Y) Apical cross-sectional area and medial (X) or cortical (Y) Sqh-GFP pixel intensity of a LEC extruding at the dorsal midline as a function of time relative to extrusion at 22 h APF. a.u., arbitrary units.
Blocking cell cycle progression in histoblasts inhibits extrusion of LECs at the border of histoblast nests, but not at the dorsal midline
Apoptosis and extrusion of LECs at the border of histoblast nests is coupled to the proliferation of histoblasts (Ninov et al., 2007). Inhibition of histoblast proliferation, either genetically or through laser light, results in a delay of cell death in the LECs neighboring the histoblast nests (Nakajima et al., 2011). This coupling of LEC extrusion and histoblast proliferation has been proposed to provide a mechanism to coordinate the spread of histoblasts with the extrusion of LECs within the pupal abdominal epidermis (Nakajima et al., 2011; Ninov et al., 2007). To test whether histoblast proliferation is also required for LEC extrusion at the dorsal midline, we blocked cell proliferation of histoblasts using a temperature-sensitive allele of Cdk1, Cdk1E1-24 (Stern et al., 1993). Cdk1 is a key regulator of cell cycle progression (Hochegger et al., 2008). Control and Cdk1E1-24 homozygous mutant pupae were shifted to the restrictive temperature of 30°C at 0 h APF. Histoblasts and LECs were visualized by E-cadherin-GFP and imaged by time-lapse microscopy. In the control, histoblasts divided, dorsal histoblast nests grew in size, and LECs were extruded at both the dorsal midline and at the border of the histoblast nests (Fig. 5A,A′,C,C′,E,G,J, Movie 6). By contrast, histoblasts no longer divided in Cdk1E1-24 mutants, the total area of dorsal histoblast nests remained approximately constant, and, consistent with a previous report (Nakajima et al., 2011), the number of extrusions of LECs at the border of histoblast nests was highly reduced (Fig. 5B,B′,D,D′,F-J, Movie 6) and unpigmented regions were discernible in the abdomen of adult flies (Fig. S1C,D). Interestingly, the cumulative number of extruded cells at the dorsal midline was similar in controls and in Cdk1E1-24 mutants, although LEC extrusion was delayed (Fig. 5G). The onset of extrusion of LECs at the dorsal midline in Cdk1E1-24 mutants roughly correlated in time with an increase in apical cross-sectional area of the remaining LECs (Fig. 5J), in particular of LECs located at the border of the non-proliferating histoblasts (Fig. 5K). This correlation indicates that the apical enlargement of LECs helps to maintain tissue integrity under conditions where the total area of histoblast nests remains approximately constant, yet LECs extrude. Under control conditions, the apical cross-sectional area of LECs remained approximately constant during the analyzed time period (Fig. 5K). Taken together, these results demonstrate that the extrusion of LECs at the dorsal midline is independent of the proliferation of histoblasts and the growth of the dorsal histoblast nests.
LEC extrusion at the dorsal midline is independent of histoblast proliferation. (A-F) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a control pupa (A,C,E) and a Cdk1E1-24 mutant pupa (B,D,F) shifted to 30°C at 0 h APF and analyzed at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. A′-D′ show segmentations with color code indicating cells that extrude at the border of histoblast nests (red) or at the dorsal midline (green). (G) Cumulative number of LEC extrusions as a function of time for control and Cdk1E1-24 mutant pupae. Mean and s.e.m. are shown (control: n=3 hemisegments for dorsal midline and 4 hemisegments for border of histoblast nest; Cdk1E1-24: n=4 hemisegments for dorsal midline and 8 hemisegments for border of histoblast nest). (H,I) Images from a time-lapse movie showing dorsal views of an abdominal epidermal segment of a Cdk1E1-24 mutant pupa shifted to 30°C at 0 h APF and analyzed at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. Histoblasts are labeled in green. (J) Total area of dorsal histoblast nests (black) and cell area of LECs (red) are shown as a function of time for control and Cdk1E1-24 mutant pupae. Mean and s.e.m. are shown. (K) Cell area of LECs located at the dorsal midline or at the border of the histoblast nest are shown as a function of time for control and Cdk1E1-24 mutant pupae. For J and K, n=249 LECs of 2 pupae for control and n=197 LECs of 2 pupae for Cdk1E1-24 mutant; n=2 nests of 2 pupae for control and n=2 nests of 2 pupae for Cdk1E1-24 mutant. Scale bars: 50 µm.
LEC extrusion at the dorsal midline is independent of histoblast proliferation. (A-F) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a control pupa (A,C,E) and a Cdk1E1-24 mutant pupa (B,D,F) shifted to 30°C at 0 h APF and analyzed at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. A′-D′ show segmentations with color code indicating cells that extrude at the border of histoblast nests (red) or at the dorsal midline (green). (G) Cumulative number of LEC extrusions as a function of time for control and Cdk1E1-24 mutant pupae. Mean and s.e.m. are shown (control: n=3 hemisegments for dorsal midline and 4 hemisegments for border of histoblast nest; Cdk1E1-24: n=4 hemisegments for dorsal midline and 8 hemisegments for border of histoblast nest). (H,I) Images from a time-lapse movie showing dorsal views of an abdominal epidermal segment of a Cdk1E1-24 mutant pupa shifted to 30°C at 0 h APF and analyzed at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. Histoblasts are labeled in green. (J) Total area of dorsal histoblast nests (black) and cell area of LECs (red) are shown as a function of time for control and Cdk1E1-24 mutant pupae. Mean and s.e.m. are shown. (K) Cell area of LECs located at the dorsal midline or at the border of the histoblast nest are shown as a function of time for control and Cdk1E1-24 mutant pupae. For J and K, n=249 LECs of 2 pupae for control and n=197 LECs of 2 pupae for Cdk1E1-24 mutant; n=2 nests of 2 pupae for control and n=2 nests of 2 pupae for Cdk1E1-24 mutant. Scale bars: 50 µm.
Genetic ablation of histoblasts inhibits extrusion of LECs at the border of histoblast nests, but not at the dorsal midline
To test further whether or not histoblast nests are required for LEC extrusion at the dorsal midline, we genetically ablated histoblasts by expressing the death-promoting gene reaper (White et al., 1994) in these cells using the esg-Gal4 driver line in combination with gal80ts. Reaper expression was induced for 6 h each at 1 and 2 days before puparium formation by shifting larvae to the restrictive temperature for gal80ts of 30°C. In controls not expressing reaper, histoblast nests grew and LECs were extruded both at the border of histoblast nests and at the dorsal midline (Fig. 6A,A′,C,C′,E, Movie 7). By contrast, expression of reaper resulted in the near-complete absence of histoblasts in the pupal abdomen (Fig. 6B,B′,D,D′, Movie 7) and a concomitant reduction in pigmentation on the adult epidermis (Fig. S1E,F). Strikingly, the cumulative number of LECs extruding at the dorsal midline was even greater than that of controls (Fig. 6E), indicating that the absence of LEC extrusion at the border of histoblast nests is compensated for. These results show that histoblasts are not required for the extrusion of LECs at the dorsal midline.
LEC extrusion at the dorsal midline is independent of histoblasts. (A-D′) Images from time-lapse movies (A-D) showing dorsal views of an abdominal epidermal segment of a control pupa (A,C) or a pupa expressing the death-inducing gene reaper in histoblasts (B,D) at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. A′-D′ show segmentations, with color code (A′,B′) indicating cells that extrude at the border of histoblast nests (red), at the dorsal midline (green) or in between (cyan), or that remain in the tissue (yellow). Dashed line in A,B indicates position of the dorsal midline. Scale bars: 50 µm. (E) Cumulative number of LEC extrusions as a function of time for control and reaper-expressing histoblasts. Mean and s.e.m. are shown (n=3 hemisegments of 3 pupae for control and n=3 hemisegments of 3 pupae for reaper).
LEC extrusion at the dorsal midline is independent of histoblasts. (A-D′) Images from time-lapse movies (A-D) showing dorsal views of an abdominal epidermal segment of a control pupa (A,C) or a pupa expressing the death-inducing gene reaper in histoblasts (B,D) at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. A′-D′ show segmentations, with color code (A′,B′) indicating cells that extrude at the border of histoblast nests (red), at the dorsal midline (green) or in between (cyan), or that remain in the tissue (yellow). Dashed line in A,B indicates position of the dorsal midline. Scale bars: 50 µm. (E) Cumulative number of LEC extrusions as a function of time for control and reaper-expressing histoblasts. Mean and s.e.m. are shown (n=3 hemisegments of 3 pupae for control and n=3 hemisegments of 3 pupae for reaper).
LEC extrusion at the dorsal midline requires neighboring LECs
Histoblasts are not required for the extrusion of LECs at the dorsal midline. LEC extrusion at the dorsal midline could therefore be a cell-autonomous process. Alternatively, LEC extrusion at the dorsal midline could require an interaction between neighboring LECs. To distinguish between these possibilities, we sought to isolate a small group of LECs located in the vicinity of the dorsal midline from their neighbors and then to test whether cells of this isolated group would extrude. We mechanically isolated a small group of LECs in the vicinity of the dorsal midline by ablating all direct neighboring cells with laser light. In controls (unablated), LECs at a comparable position shrank their apical cross-sectional area and extruded (Fig. 7A,B,E, Movie 8). By contrast, LECs that were isolated from their neighbors maintained their apical cross-sectional area (Fig. 7C-E, Movie 8). The cumulative fraction of cells that extruded inside the ablated region was strongly reduced compared with outside the ablated region (Fig. 7F-H). In the control situation, extrusion of LECs at the dorsal midline was associated with movement of the remaining LECs towards the dorsal midline. Consistent with this notion, we observed that outside of the ablated area LECs moved towards the dorsal midline (Fig. 7F,G,I). By contrast, LECs inside the ablated area did not move towards the dorsal midline (Fig. 7F,G,I). Taken together, these results show that LEC extrusion at the dorsal midline requires neighboring LECs.
LEC extrusion at the dorsal midline requires neighboring LECs. (A-D) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a control pupa (A,B) or a pupa in which LECs were ablated at 29 h APF (C,D). Cells were ablated along the lines indicated in yellow; dashed yellow line indicates the future position of ablation. Cell outlines are visualized by Indy-GFP. Magenta dashed lines indicate position of the dorsal midline. Arrowheads indicate extruded cells. Arrow indicates extruding cells. (E) Cell area of LECs as a function of time APF for control pupae and pupae in which cells were ablated. Only cells inside the ablated area (or control equivalent) were analyzed. Yellow line indicates time of ablation at 29 h APF. Mean and s.e.m. are shown (n=36 cells of 1 control pupa and n=40 cells of 3 ablated pupae). (F,G) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a pupa before (F) or after (G) ablation at 27.5 h APF. Cells were ablated along the lines indicated in yellow; yellow dashed line indicates the future position of ablation. Cell outlines are visualized by Indy-GFP. Magenta dashed lines indicate position of the dorsal midline. Color code indicates the corresponding positions of selected LECs inside (red) or outside (cyan) the ablated region at 26.5 h and 30.5 h APF. Cells outside the ablated region move towards the dorsal midline in contrast to cells inside the ablated region. (H) Cumulative fraction of extruding LECs inside and outside the ablated region as a function of time. Yellow line indicates time of ablation at 27.5 h APF. Mean and s.e.m. are shown. (I) Distance of the cell center of LECs from the dorsal midline as a function of time for cells inside the ablated region (red) and cells outside the ablated region (blue). Yellow line indicates time of ablation at 27.5 h APF. Mean and s.e.m. are shown. For H and I, n=31 cells inside ablated region and n=70 cells outside ablated region of 3 pupae. Scale bars: 50 µm.
LEC extrusion at the dorsal midline requires neighboring LECs. (A-D) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a control pupa (A,B) or a pupa in which LECs were ablated at 29 h APF (C,D). Cells were ablated along the lines indicated in yellow; dashed yellow line indicates the future position of ablation. Cell outlines are visualized by Indy-GFP. Magenta dashed lines indicate position of the dorsal midline. Arrowheads indicate extruded cells. Arrow indicates extruding cells. (E) Cell area of LECs as a function of time APF for control pupae and pupae in which cells were ablated. Only cells inside the ablated area (or control equivalent) were analyzed. Yellow line indicates time of ablation at 29 h APF. Mean and s.e.m. are shown (n=36 cells of 1 control pupa and n=40 cells of 3 ablated pupae). (F,G) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a pupa before (F) or after (G) ablation at 27.5 h APF. Cells were ablated along the lines indicated in yellow; yellow dashed line indicates the future position of ablation. Cell outlines are visualized by Indy-GFP. Magenta dashed lines indicate position of the dorsal midline. Color code indicates the corresponding positions of selected LECs inside (red) or outside (cyan) the ablated region at 26.5 h and 30.5 h APF. Cells outside the ablated region move towards the dorsal midline in contrast to cells inside the ablated region. (H) Cumulative fraction of extruding LECs inside and outside the ablated region as a function of time. Yellow line indicates time of ablation at 27.5 h APF. Mean and s.e.m. are shown. (I) Distance of the cell center of LECs from the dorsal midline as a function of time for cells inside the ablated region (red) and cells outside the ablated region (blue). Yellow line indicates time of ablation at 27.5 h APF. Mean and s.e.m. are shown. For H and I, n=31 cells inside ablated region and n=70 cells outside ablated region of 3 pupae. Scale bars: 50 µm.
E-cadherin turnover at LEC junctions increases during development
Extrusion of LECs at the border of histoblast nests is associated with a reduction of E-cadherin, indicating that a weakening of adhesion between apoptotic LECs and their neighboring cells contributes to extrusion (Teng et al., 2017). To test whether E-cadherin (Shotgun in Drosophila) levels are altered during pupal development, we analyzed E-cadherin-GFP levels at adherens junctions of LECs at 20 h APF and 27 h APF. E-cadherin-GFP levels on LEC adherens junctions were reduced by approximately 26% at 27 h APF compared with 20 h APF (Fig. 8A-C). We next tested whether the molecular dynamics of E-cadherin also change during development. To this end, we performed fluorescence recovery after photobleaching (FRAP) experiments using E-cadherin-GFP at 20 h APF and 27 h APF. The mobile fraction of E-cadherin-GFP at adherens junctions of LECs was increased by approximately 12% at 27 h APF compared with 20 h APF (Fig. 8D-Q, Movie 9). Moreover, the half-time of fluorescence recovery was decreased by approximately 24% at 27 h APF compared with 20 h APF (Fig. 8R). Taken together, these experiments show that E-cadherin levels at adherens junctions of LECs decrease and E-cadherin turnover increases during pupal development.
E-cadherin turnover increases during development. (A,B) Images showing LECs of pupae expressing E-cadherin-GFP at the indicated times APF. Colors indicate pixel intensities of E-cadherin-GFP. Scale bars: 50 µm. (C) E-cadherin-GFP pixel intensities at adherens junctions of LECs of pupae at the indicated times APF. Pixel intensities are normalized to the mean value at 20 h APF. Mean and s.e.m. are shown (n=5 pupae at 20 h APF, n=5 pupae at 27 h APF, *P<0.05). (D-O) Images of a FRAP experiment from time-lapse movies showing dorsal views of an abdominal epidermal segment of a pupa before and after photobleaching of E-cadherin-GFP of LECs for the indicated times APF. Colors indicate pixel intensities of E-cadherin-GFP. Scale bars: 5 µm. (P) Normalized E-cadherin-GFP fluorescence intensity of a FRAP experiment as a function of time relative to photobleaching of a single adherens junction of a LEC at the indicated time APF. Mean and s.e.m. are shown (n=33 photobleached adherens junctions for 20 h APF; n=37 photobleached adherens junctions for 27 h APF). Mean and s.e.m. of the data fitting curves are shown in black. (Q,R) Mobile fraction of E-cadherin-GFP (Q) and half time of E-cadherin-GFP recovery (R) at adherens junctions of LECs of pupae at the indicated times APF. Mean and s.e.m. are shown (n=33 pupae at 20 h APF, n=37 pupae at 27 h APF, **P<0.01; ***P<0.001).
E-cadherin turnover increases during development. (A,B) Images showing LECs of pupae expressing E-cadherin-GFP at the indicated times APF. Colors indicate pixel intensities of E-cadherin-GFP. Scale bars: 50 µm. (C) E-cadherin-GFP pixel intensities at adherens junctions of LECs of pupae at the indicated times APF. Pixel intensities are normalized to the mean value at 20 h APF. Mean and s.e.m. are shown (n=5 pupae at 20 h APF, n=5 pupae at 27 h APF, *P<0.05). (D-O) Images of a FRAP experiment from time-lapse movies showing dorsal views of an abdominal epidermal segment of a pupa before and after photobleaching of E-cadherin-GFP of LECs for the indicated times APF. Colors indicate pixel intensities of E-cadherin-GFP. Scale bars: 5 µm. (P) Normalized E-cadherin-GFP fluorescence intensity of a FRAP experiment as a function of time relative to photobleaching of a single adherens junction of a LEC at the indicated time APF. Mean and s.e.m. are shown (n=33 photobleached adherens junctions for 20 h APF; n=37 photobleached adherens junctions for 27 h APF). Mean and s.e.m. of the data fitting curves are shown in black. (Q,R) Mobile fraction of E-cadherin-GFP (Q) and half time of E-cadherin-GFP recovery (R) at adherens junctions of LECs of pupae at the indicated times APF. Mean and s.e.m. are shown (n=33 pupae at 20 h APF, n=37 pupae at 27 h APF, **P<0.01; ***P<0.001).
Dynamin is required for the extrusion of LECs
E-cadherin turnover depends in part on its endocytosis from the plasma membrane and recycling to the plasma membrane. Endocytosis (of E-cadherin) requires the activity of the large GTPase Dynamin (encoded by the shibire gene in Drosophila) (Jimah and Hinshaw, 2018). We first tested whether E-cadherin turnover depends on Dynamin function. To this end, we blocked Dynamin function using a temperature-sensitive allele of shibire, shibire1 (Grigliatti et al., 1973). Control and shibire1 homozygous mutant pupae were shifted to the restrictive temperature of 30°C at 16 h APF and FRAP experiments were performed 2 h later. The mobile fraction of E-cadherin-GFP was reduced by approximately 20% in shibire1 mutants compared with controls (Fig. 9A,B, Fig. S3), showing that Dynamin function is required for E-cadherin turnover at adherens junctions of LECs. To test whether Dynamin is required for LEC extrusion, we shifted control and shibire1 homozygous mutant pupae to the restrictive temperature of 30°C at 16 h APF. Histoblasts and LECs were visualized by E-cadherin-GFP and imaged by time-lapse microscopy. In the control, dorsal histoblast nests grew in size and LECs were extruded at both the dorsal midline and at the border of the histoblast nests (Fig. 9C,C′,E,E′,G, Movie 10). By contrast, histoblast nest growth was impaired in shibire1 mutants, and the number of extrusions of LECs at both the border of histoblast nests and at the dorsal midline was greatly reduced (Fig. 9D,D′,F-G, Movie 10). Notably, LEC extrusion was mainly impaired in the lateral region of each segment (Fig. 9D), whereas LECs in the medial region of each segment continued to extrude, even with a higher rate (Fig. S4). Integrity of the abdominal epidermis was maintained. These data reveal that Dynamin is required for LEC extrusion in the lateral region of the segments.
Dynamin is required for LEC extrusion in lateral but not medial regions of the segment. (A) E-cadherin-GFP fluorescence intensity of a FRAP experiment normalized to the average intensity before photobleaching as a function of time relative to bleaching of a single adherens junction of a LEC of control pupa and a shi1 mutant pupa shifted to 30°C at 16 h APF and analyzed at 20 h APF. Mean and s.e.m. are shown (n=10 photobleached adherens junctions for control; n=16 photobleached adherens junctions for shi1 mutant). Mean and s.e.m. of the data fitting curves are shown in black. (B) Mobile fraction of E-cadherin-GFP at adherens junctions of LECs for the experimental data shown in A. *P<0.05. (C-F′) Images from time-lapse movies (C-F) showing dorsal views of an abdominal epidermal segment of a control pupa (C,E) and a shi1 mutant pupa (D,F) shifted to 30°C at 16 h APF and analyzed at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. C′-F′ show segmentations, with color code (C′,D′) indicating cells that extrude at the border of histoblast nests (red), at the dorsal midline (green) or in between (cyan). Yellow indicates LECs that do not extrude. Dashed lines indicate the position of dorsal midline. Scale bars: 50 µm. (G) Cumulative number of LEC extrusions as a function of time for control and shi1 mutant pupa. Mean and s.e.m. are shown (n=3 hemisegments of 3 pupae for control and n=3 hemisegments of 3 pupae for shi1 mutant).
Dynamin is required for LEC extrusion in lateral but not medial regions of the segment. (A) E-cadherin-GFP fluorescence intensity of a FRAP experiment normalized to the average intensity before photobleaching as a function of time relative to bleaching of a single adherens junction of a LEC of control pupa and a shi1 mutant pupa shifted to 30°C at 16 h APF and analyzed at 20 h APF. Mean and s.e.m. are shown (n=10 photobleached adherens junctions for control; n=16 photobleached adherens junctions for shi1 mutant). Mean and s.e.m. of the data fitting curves are shown in black. (B) Mobile fraction of E-cadherin-GFP at adherens junctions of LECs for the experimental data shown in A. *P<0.05. (C-F′) Images from time-lapse movies (C-F) showing dorsal views of an abdominal epidermal segment of a control pupa (C,E) and a shi1 mutant pupa (D,F) shifted to 30°C at 16 h APF and analyzed at the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. C′-F′ show segmentations, with color code (C′,D′) indicating cells that extrude at the border of histoblast nests (red), at the dorsal midline (green) or in between (cyan). Yellow indicates LECs that do not extrude. Dashed lines indicate the position of dorsal midline. Scale bars: 50 µm. (G) Cumulative number of LEC extrusions as a function of time for control and shi1 mutant pupa. Mean and s.e.m. are shown (n=3 hemisegments of 3 pupae for control and n=3 hemisegments of 3 pupae for shi1 mutant).
Mechanical tension on adherens junctions of LECs increases during development
LEC extrusion at the dorsal midline requires neighboring LECs. Neighboring cells could provide biochemical or mechanical signals to promote LEC extrusion. To test whether mechanical signals (forces) contribute to LEC extrusion, we measured the mechanical tension on adherens junctions at two different time points: early, at 20 h APF, when LECs frequently extrude when located at the border of histoblast nests but only rarely when located at the dorsal midline, and late, at 27 h APF, when LECs extrude at both locations (see Fig. 1J). Mechanical tension was measured on adherens junctions of LECs located at the border of histoblast nests and of LECs at the dorsal midline. LECs in the process of extrusion were excluded from the analysis. To measure mechanical tension, we ablated single adherens junctions with focused laser light and recorded the resulting displacement of the two vertices at the ends of the ablated junction (Fig. 10A-M, Movie 11). We then calculated the average initial velocity of vertex displacement as a relative measure of mechanical tension acting on the adherens junction before ablation (Hutson et al., 2003; Ma et al., 2009). For both time points analyzed, the average initial velocity of vertex displacement upon laser ablation was similar for LECs located at the border of histoblast nests and LECs located in the vicinity of the dorsal midline (Fig. 10N). Interestingly, the average initial velocity of vertex displacement upon laser ablation of junctions for both subsets of LECs was approximately 5- to 10-fold increased at 27 h APF compared with 20 h APF (Fig. 10N). Moreover, the standard deviation of measured initial velocities was greater at 27 h APF compared with 20 h APF (Fig. 10O), indicating that the strength of the mechanical tension on adherens junctions was more heterogeneous at the later developmental time point. To test whether the high mechanical tension at 27 h APF depends on the growth of the histoblasts, we ablated single adherens junctions of LECs at this developmental time point in controls and in Cdk1E1-24 mutants, in which histoblasts do not divide (see Fig. 5). The average initial velocities of vertex displacement upon ablation were indistinguishable between controls and Cdk1E1-24 mutants (Fig. 10P; Fig. S5), demonstrating that the mechanical tension on LEC adherens junctions is independent of histoblast proliferation. Taken together, these results show that the mechanical tension on adherens junctions of LECs is drastically increased at late developmental time points when LECs extrude both at the border of histoblast nests and at the dorsal midline compared with an earlier time point when LEC extrusion is limited to the border of the histoblast nests.
Mechanical tension on adherens junctions of LECs increases during development. (A-L) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a pupa before and 0.2 and 20 s after ablation of a single LEC adherens junction at the edge of a histoblast nest (A-F) or in the vicinity of the dorsal midline (G-L) for the indicated times APF. 0.2 s is the first time point after ablation at which an image was acquired. Adherens junctions are visualized by E-cadherin-GFP. Scale bars: 10 µm. (M) Increase in distance between the two ends of the ablated cell junction upon laser ablation as a function of time relative to ablation. Adherens junctions of LECs at the edge of histoblast nests and in the vicinity of the dorsal midline were ablated at the indicated times APF. Mean and s.e.m. are shown. (N) Initial velocity of displacement of the two ends of the ablated adherens junction for LECs at the edge of a histoblast nest or in the vicinity of the dorsal midline at the indicated times APF. Mean and s.e.m. are shown. ***P<0.001. For N and M, n=16 ablations for nest at 20 h APF, n=15 ablations for nest at 27 h APF, n=18 ablations for midline at 20 h APF, n=15 ablations for midline at 27 h APF. (O) Standard deviations of the initial velocities shown in N. (P) Initial velocity of displacement of the two ends of the ablated adherens junction for LECs in the vicinity of the dorsal midline for control and Cdk1E1-24 mutant pupae. Pupae were shifted to 30°C at 0 h APF and analyzed at 23 h APF. Mean and s.e.m. are shown (n=19 ablations for control, n=20 ablations for Cdk1E1-24 mutant pupae). n.s., not significant.
Mechanical tension on adherens junctions of LECs increases during development. (A-L) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of a pupa before and 0.2 and 20 s after ablation of a single LEC adherens junction at the edge of a histoblast nest (A-F) or in the vicinity of the dorsal midline (G-L) for the indicated times APF. 0.2 s is the first time point after ablation at which an image was acquired. Adherens junctions are visualized by E-cadherin-GFP. Scale bars: 10 µm. (M) Increase in distance between the two ends of the ablated cell junction upon laser ablation as a function of time relative to ablation. Adherens junctions of LECs at the edge of histoblast nests and in the vicinity of the dorsal midline were ablated at the indicated times APF. Mean and s.e.m. are shown. (N) Initial velocity of displacement of the two ends of the ablated adherens junction for LECs at the edge of a histoblast nest or in the vicinity of the dorsal midline at the indicated times APF. Mean and s.e.m. are shown. ***P<0.001. For N and M, n=16 ablations for nest at 20 h APF, n=15 ablations for nest at 27 h APF, n=18 ablations for midline at 20 h APF, n=15 ablations for midline at 27 h APF. (O) Standard deviations of the initial velocities shown in N. (P) Initial velocity of displacement of the two ends of the ablated adherens junction for LECs in the vicinity of the dorsal midline for control and Cdk1E1-24 mutant pupae. Pupae were shifted to 30°C at 0 h APF and analyzed at 23 h APF. Mean and s.e.m. are shown (n=19 ablations for control, n=20 ablations for Cdk1E1-24 mutant pupae). n.s., not significant.
High mechanical tension on adherens junctions of LECs requires caspase activity
Caspase activity in LECs at late developmental stages correlates with high mechanical tension on adherens junctions (compare Fig. 3 and Fig. 10). To test whether high mechanical tension depends on caspase activity, we blocked caspase activity in LECs by expression of P35. Expression of P35 was limited to LECs of the pupal stage by combining the tsh-Gal4 driver with a temperature-sensitive form of the Gal4 repressor Gal80, encoded by gal80ts. We induced P35 expression at 0 h APF by shifting pupae to the restrictive temperature for gal80ts of 30°C and ablated single adherens junctions of LECs located at the dorsal midline at 18 h and 23 h APF (corresponding to the approximate developmental stages that are reached at 25°C by 20 h and 27 h APF, respectively). At 18 h APF, the average initial velocity of vertex displacement was similar for ablated adherens junctions of control and P35-expressing LECs (Fig. 11A-C,G-I,M,N). By contrast, at 23 h APF, the average initial velocity of vertex displacement was decreased by approximately 70% in P35-expressing LECs compared with controls (Fig. 11D-F,J-N). To test whether caspase activation is sufficient to increase mechanical tension, we expressed the pro-apoptotic gene reaper in LECs during early development. LECs underwent apoptosis prematurely (Fig. S6A,B), yet mechanical tension on LEC junctions was similar to that of controls (Fig. S6C-J). Taken together, these data show that caspase activity is required, but not sufficient, for high mechanical tension on adherens junctions of LECs during late development.
Increased mechanical tension at late development requires caspase activity. (A-L) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of control pupae (A-F) or pupae expressing P35 in LECs (G-L) before and 0.3 and 20 s after ablation of a single LEC adherens junction at the dorsal midline for the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. Scale bars: 10 µm. (M) Increase in distance between the two ends of the ablated cell junction upon laser ablation as a function of time relative to ablation for control pupae and pupae expressing P35 in LECs. Mean and s.e.m. are shown. (N) Initial velocity of displacement of the two ends of the ablated adherens junction for LECs of control pupae or pupae expressing P35 in LECs. Mean and s.e.m. are shown. n.s., not significant. ***P<0.001. For N and M, control: n=16 ablations at 18 h APF, n=16 ablations at 23 h APF; P35: n=16 ablations at 18 h APF, n=17 ablations at 23 h APF.
Increased mechanical tension at late development requires caspase activity. (A-L) Images from time-lapse movies showing dorsal views of an abdominal epidermal segment of control pupae (A-F) or pupae expressing P35 in LECs (G-L) before and 0.3 and 20 s after ablation of a single LEC adherens junction at the dorsal midline for the indicated times APF. Adherens junctions are visualized by E-cadherin-GFP. Scale bars: 10 µm. (M) Increase in distance between the two ends of the ablated cell junction upon laser ablation as a function of time relative to ablation for control pupae and pupae expressing P35 in LECs. Mean and s.e.m. are shown. (N) Initial velocity of displacement of the two ends of the ablated adherens junction for LECs of control pupae or pupae expressing P35 in LECs. Mean and s.e.m. are shown. n.s., not significant. ***P<0.001. For N and M, control: n=16 ablations at 18 h APF, n=16 ablations at 23 h APF; P35: n=16 ablations at 18 h APF, n=17 ablations at 23 h APF.
DISCUSSION
We have analyzed the replacement of LECs by adult histoblasts in the pupal abdominal epidermis in Drosophila. Previous work has shown that the interaction between the growing nest of histoblasts and LECs is required for LEC extrusion. LEC extrusion is closely preceded by caspase activation and involves contraction of a supracellular actomyosin cable (Teng et al., 2017). Here, we have identified a second hotspot of cell extrusion at the dorsal midline of the pupal abdomen. Cell extrusion at the dorsal midline occurs mainly late during development and is independent of histoblasts, but depends on the presence of neighboring LECs. LEC extrusion is preceded by a uniform rise in caspase activity throughout the larval tissue and involves pulsatile contractions of a medial actomyosin network. Mechanical tension on adherens junctions of LECs increases during pupal development. High mechanical tension during late pupal development depends in part on caspase activity and correlates with a high turnover of E-cadherin. E-cadherin turnover depends on Dynamin-dependent endocytosis and blocking Dynamin severely affects LEC extrusion. This work reveals a novel, histoblast-independent mechanism by which LECs extrude during late pupal development.
The mechanism of LEC extrusion at the dorsal midline
Extrusion of LECs located at the border of histoblast nests during early pupal development involves the remodeling and weakening of adherens junctions between the extruding LEC and its neighboring cells, and the accumulation of Myosin II at the cortices of both the extruding LEC and its neighboring cells. Cortical Myosin II accumulation in neighboring cells results in the formation of a supracellular actomyosin cable. Contraction of this cable facilitates the extrusion of the LEC (Teng et al., 2017). Our data indicate that extrusion of LECs at the dorsal midline during late pupal development involves a distinct mechanism. In addition to a cortical pool of Myosin II and F-actin, we find Myosin II and F-actin also in an apical medial region of the extruding LECs. Similar accumulations of Myosin II and F-actin have been reported in several other epithelia, including in the Drosophila embryo in ventral furrow cells of the epidermis and in amnioserosa cells (Martin et al., 2009a; Solon et al., 2009). The medial actomyosin network is pulsatile in these cells resulting in concomitant fluctuations in apical cell area. Similarly, we find that in extruding LECs located in the vicinity of the dorsal midline, apical cell area fluctuates and apical constriction correlates with increased medial Myosin II intensity. Over longer time periods, apical cell area shrinks, indicating a ratchet-like mechanism. A similar ratchet-like contraction mechanism is involved in the Drosophila embryo during neuroblast delamination (Simões et al., 2017) and the extrusion of amnioserosa cells (Solon et al., 2009). The reason for these different extrusion mechanisms of LECs during early and late pupal development remains unclear. We note, however, that mechanical tension on LECs increases approximately 5- to 10-fold between early (20 h APF) and late (27 h APF) pupal development, suggesting that overall tissue mechanical tension influences the mechanism by which cells extrude.
The trigger for LEC extrusion at the dorsal midline
How is the extrusion of LECs at the dorsal midline initiated? Several mechanisms have been described that drive cell extrusion, including cell competition and tissue crowding (Eisenhoffer et al., 2012; Levayer et al., 2016; Marinari et al., 2012). These two mechanisms, however, appear not to play a major role in the extrusion of LECs located at the dorsal midline of the pupal abdominal epidermis. Tissue crowding induces, for example, caspase-dependent cell extrusion at the dorsal midline in the pupal notum (Levayer et al., 2016). Experimental induction of clones expressing an activated form of Ras, RasV12, leads to the crowding of neighboring wild-type cells and to their extrusion (Levayer et al., 2016). In the abdominal epidermis, the growth or migration of histoblasts (Bischoff and Cseresnyes, 2009) could result in crowding of the LECs. During the analyzed developmental time period, the average area of non-extruding LECs located at the dorsal midline, however, remains nearly constant, indicating that these LECs do not become crowded (Fig. 5K). Moreover, LECs located at the dorsal midline still extrude when histoblast division is halted or histoblasts are genetically ablated, providing further evidence that tissue crowding is not involved in the extrusion of these cells. Finally, cell competition refers to the out competition of slow-growing cells by fast-growing cells (Martin et al., 2009b). Again, our finding that LEC extrusion at the dorsal midline is independent of histoblast division, strongly argues that cell competition does not have a major influence on the extrusion of these LECs.
During early pupal development, local interactions between histoblasts and LECs trigger caspase activation in LECs (Nakajima et al., 2011). Caspase activation then closely precedes extrusion of LECs, cell by cell, at the border of histoblast nests, indicating that caspase activation triggers extrusion (Teng et al., 2017). In LECs at the dorsal midline, by contrast, caspase activity, as measured by Apoliner cleavage (Bardet et al., 2008), shows a population-wide fluctuation during pupal development. Caspase activity peaks during the observed developmental time period twice, around 20 h APF and around 26-29 h APF. Caspase activation in these cells does not appear to trigger their extrusion, because the first peak of caspase activity at around 20 h APF takes place several hours before LECs at the dorsal midline start to extrude. Interestingly, the second rise of caspase activity between 24 h and 26 h APF precedes the extrusion of LECs at the dorsal midline (compare Fig. 1J and Fig. 3M). Taken together with our observation that LEC extrusion at the dorsal midline requires caspase activity, these results indicate that the second rise of caspase activity is a prerequisite for LECs to extrude at the dorsal midline. How the population-wide fluctuations of caspase activity in LECs at the dorsal midline arise will be interesting to explore.
We find that mechanical tension acting on LEC adherens junctions increases during pupal development. Increased mechanical tension is independent of the proliferation of histoblasts, but requires caspase activity, which in turn is necessary for LEC extrusion. We speculate that LEC extrusion during late pupal development may, in part, facilitate the increased mechanical tension on adherens junctions of the remaining LECs. Mechanical tension may influence cell extrusion in different ways and by different mechanisms. In the pupal notum, for example, a similar developmental increase in mechanical tension has been observed, but, unlike the situation in the pupal abdomen, increased mechanical tension correlates with fewer cell extrusions (Curran et al., 2017). This different behavior may result from the different characteristics of the two epithelia. In the notum, cell extrusion occurs within a diploid, proliferating tissue with columnar cells. On the other hand, LECs of the pupal abdomen are polyploid, non-proliferating and squamous in shape. In these respects, the abdominal LECs resemble more the amnioserosa cells of the embryo.
How could an increase in mechanical tension on LEC adherens junctions facilitate extrusion? LEC extrusion involves the remodeling of adherens junctions of the extruding cell (Teng et al., 2017). Adherens junction remodeling depends on E-cadherin. We find that the turnover of E-cadherin is increased at 27 h APF compared with 20 h APF, thus positively correlating with increased mechanical tension. The turnover of E-cadherin depends on lateral diffusion on the plasma membrane and on trafficking between a cytosolic, presumably vesicular, pool of E-cadherin and a junctional pool of E-cadherin (Le et al., 1999). The exchange between these two pools requires Dynamin-dependent endocytosis (Le et al., 1999). Consistent with this notion, we find that E-cadherin turnover is reduced in LECs of flies mutant for shibire1, which encodes Dynamin. Interestingly, extrusion of LECs in the lateral region of the abdominal segments is nearly blocked in shibire1 mutants. Although we cannot exclude the possibility that Dynamin-dependent endocytosis of proteins other than E-cadherin are important for LEC extrusion, these data lend support for the idea that high E-cadherin turnover is required for extrusion of these LECs. Surprisingly, a subset of LECs, those located in medial regions of the segment, do extrude in shibire1 mutants. These findings reveal an unexpected complexity of LEC extrusion that will require future work to be resolved.
Two distinct mechanisms drive cell extrusion in the pupal abdominal epidermis
We propose the following model of how LECs in the pupal abdominal epidermis extrude (Fig. 12). During early pupal development (20 h APF) mechanical tension on LEC junctions is low and E-cadherin turnover is low. LECs extrude only if they receive a specific ‘extrusion signal’ from the histoblasts (Nakajima et al., 2011; Ninov et al., 2007). Extrusion is mediated by cortical actomyosin, which forms a contractile supracellular cable driving apical cell constriction (Teng et al., 2017). Later during pupal development (27 h APF), mechanical tension on LEC junctions, E-cadherin turnover and caspase activity are high. Stochastic extrusions of LECs commence, which are mediated by a pulsatile medial actomyosin network that leads to ratchet-like apical cell contractions.
Two distinct mechanisms drive LEC extrusion. Schematic of the differences between early and late pupal development and the resulting change in LEC extrusion. During early development, mechanical tension and E-cadherin turnover are low and LECs extrude at the border of the histoblast nest (red) by the contraction of a cortical actomyosin cable. At late development, mechanical tension (arrows) and E-cadherin turnover are high and LECs extrude at the dorsal midline (green) through a pulsatile actomyosin network.
Two distinct mechanisms drive LEC extrusion. Schematic of the differences between early and late pupal development and the resulting change in LEC extrusion. During early development, mechanical tension and E-cadherin turnover are low and LECs extrude at the border of the histoblast nest (red) by the contraction of a cortical actomyosin cable. At late development, mechanical tension (arrows) and E-cadherin turnover are high and LECs extrude at the dorsal midline (green) through a pulsatile actomyosin network.
MATERIALS AND METHODS
Fly stocks and genetics
The following fly stocks were used: E-cadherin-GFP (Huang et al., 2009), esg-Gal4 [Bloomington Drosophila Stock Center (BDSC) #26816], tsh-Gal4 (BDSC #3040), tub-Gal80ts (BDSC #7019), UAS-p35 (BDSC #5072), UAS-Apoliner (Bardet et al., 2008), shi1 (BDSC #7068), Cdk1E1-24 (Stern et al., 1993), indy-GFP (Quiñones-Coello et al., 2007), UAS-reaper (BDSC #5824), sqh-Utr::GFP (Rauzi et al., 2010), sqhAX3; sqh-sqh::GFP (Bertet et al., 2004).
The genotypes of pupae and the temperature regime they were raised at is as follows (if not stated otherwise pupae were raised at 25°C). Fig. 1: E-cadherin-GFP; Fig. 2: tsh-Gal4, tub-Gal80ts, E-cadherin-GFP/UAS-p35, E-cadherin-GFP and E-cadherin-GFP as control. Flies were raised at 18°C, transferred to 25°C 2 days before pupae collection and transferred to 30°C at 0 h APF; Fig. 3: tsh-Gal4/+; UAS-Apoliner/+; Fig. 4: sqhAX3; sqh-sqh::GFP; Fig. 5: Cdk1E1-24, E-cadherin-GFP and E-cadherin-GFP as controls. Flies were raised at 18°C, transferred to 25°C 2 days before pupae collection and transferred to 30°C at 0 h APF; Fig. 6: esg-Gal4, tub-Gal80ts, E-cadherin-GFP/UAS-reaper, E-cadherin-GFP and E-cadherin-GFP as control. Flies were raised at 18°C and transferred to 25°C 2 days before pupae collection. On each of the 2 days before pupae collection the pupae were transferred for 6 h to 30°C to induce reaper expression and then back to 25°C. Pupae collection and imaging was done at 25°C; Fig. 7: indy-GFP; Fig. 8: E-cadherin-GFP; Fig. 9: shi1; E-cadherin-GFP and E-cadherin-GFP as control. Flies were raised at 25°C. Collected pupae were first put at 25°C and transferred to 30°C at 16 h APF; Fig. 10: (A-O) E-cadherin-GFP. (P) Cdk1E1-24, E-cadherin-GFP and E-cadherin-GFP as controls. Flies were raised at 18°C, transferred to 25°C 2 days before pupae collection and transferred to 30°C at 0 h APF; Fig. 11: tsh-Gal4, tub-Gal80ts, E-cadherin-GFP/UAS-p35, E-cadherin-GFP and E-cadherin-GFP as control. Flies were raised at 18°C, transferred to 25°C 2 days before pupae collection and transferred to 30°C at 0 h APF. For information on genotypes shown in supplementary figures, see Supplementary Materials and Methods.
Time-lapse imaging
Flies were raised under standard conditions at the temperatures indicated for each experiment. To stage the pupae, white pupae (0 h APF) were collected and put at the indicated temperatures. The late pupae were put on a piece of double-sided tape, a window in the pupal case was opened, covered with halocarbon oil 700 and mounted on a glass-bottom dish (MatTek corporation) with the opening facing towards the cover slip (Ninov and Martín-Blanco, 2007). Imaging was performed using an inverted confocal laser-scanning microscope (Leica SP5, Zeiss LSM 880, Zeiss LSM 710) with 40× or 63× water- or oil-immersion objectives. Images of abdominal segments 2-4 were acquired using different time intervals. At every time point for each pupa several tiles of z-stacks were recorded.
Image processing and analysis
Images were projected via the ‘maximum intensity/sum slices projection’ in Fiji (Schindelin et al., 2012) or using a customized PreMosa script (Blasse et al., 2017). Projected tiles were stitched together using Fiji and segmented using the Fiji plugin Tissue Analyzer (Aigouy et al., 2016). Tracking of single cells was carried out either manually with Fiji or using the Fiji plugin Tissue Analyzer. Comparable regions of the dorsal hemisegment were analyzed in controls and experimental conditions. Areas and intensities were measured using a custom-made script based on the segmentations. Data were analyzed and plotted using custom-made scripts with the R package (R Foundation for Statistical Computing; https://www.R-project.org/). The cortical region of a cell was defined as a four-pixel-wide area centered on the cell outline. The medial region was defined as the non-cortical region of the cell. The dorsal midline was identified by the curvature of the global LEC alignment. Extrusions at the dorsal midline were defined as extrusions not in contact with the histoblast nest that were not more than 50 µm away from the dorsal midline. Extrusions at the histoblast nest were defined as extrusions at least 50 µm away from the dorsal midline and in contact only with the histoblast nest of the respective hemisegment and not with nests from the other hemisegments. In any case only extrusions of cells from a single hemisegment were counted. In the plots cumulative extrusions are shown, which corresponds to the number of all extrusions of this type until the time point plotted on the x-axis.
Laser ablation (Figs 10 and 11; Figs S5 and S6)
Laser ablation was performed as described previously (Umetsu et al., 2014). Pupae carrying E-cadherin-GFP were staged and prepared as for the time-lapse imaging. They were imaged on an inverted spinning disc microscope with a 63×/1.2 water-immersion objective. Images were recorded every 0.2-0.3 s over a total time of 1 min. Ablation was carried out using a UV laser focused on the level of the adherens junctions. Tissue relaxation was analyzed manually using Fiji and plotted as relative distance increase over time or as initial velocity.
Laser ablation (Fig. 7)
Pupae were staged and prepared as for the time-lapse imaging. They were imaged on an inverted confocal microscope (Zeiss LSM 710) with a 63×/1.4 oil-immersion objective. The indicated region was ablated with an IR laser (800 nm) and imaging continued afterwards.
Cross-correlation analysis (Fig. 4)
Mean and total fluorescence intensities of medial and cortical Myosin were measured with a time interval of 30 s. Using the R package the cross-correlation of total cell area and the indicated fluorescence intensities were analyzed.
FRAP
Statistical analysis
Statistical significance was tested with the Mann–Whitney U-test using the R package.
Acknowledgements
We thank Christian Bökel, Suzanne Eaton, Elisabeth Knust and the Bloomington Drosophila Stock Center for fly stocks; Corinna Blasse and Eugene Myers for providing a customized version of PreMosa; the Light Microscopy Facility of the MPI-CBG and the CMCB Technology Platform for providing imaging equipment; Klaus Reinhardt for providing access to the Zeiss LSM 880 microscope; Stephan Grill for providing access to the laser-ablation equipment; and Christian Bökel and Elisabeth Knust for comments on the manuscript.
Footnotes
Author contributions
Conceptualization: M.M., C.D.; Formal analysis: M.M.; Investigation: M.M.; Data curation: M.M.; Writing - original draft: C.D.; Writing - review & editing: M.M., C.D.; Supervision: C.D.
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
References
Competing interests
The authors declare no competing or financial interests.