Wounds induce epithelial syncytia via cell fusion.

A) Syncytia form within 2 h post wounding, evident by the clustering of multiple nuclei within cell borders. B) The number of nuclei per syncytia increases over time after wounding. Number of nuclei was estimated based on area and nuclear density (see text) for the 3 largest syncytia of 3 different wounds, mean and SD. C-D) Larger wounds generate larger syncytia. Images are at 3 h post wounding, single syncytium outlined. E) Syncytial apical area is proportional to initial wound size. Each dot represents a wound, made with either low or high laser energy as sown. For each wound, the mean area of the three largest syncytia is plotted. Bars represent SD. F) A time course of six cells fusing within 30 min after wounding. Apical borders are lost (white arrowhead) as syncytia form. Original cells are numbered. G) All borders lost to cell fusion (white) mapped to cells in the first frame after wounding. The leading edge of wound closure will form at dashed line; cells within the shaded area were damaged by the wound and will be dismantled. H) Distance from the wound center vs time for all border breakdowns in 3 wounds. Each symbol represents a cell border. Leading-edge locations indicated by solid lines. I) Cytoplasmic GFP is expressed in cell 1 before wounding and mixes with neighboring cells 2-4 by 2 min after wounding. Cytoplasmic sharing is followed by the lagging fusion indicator of visible border breakdown (white arrow). The fates of cells 3 and 4 are shown at later times in Figure 2D. Maximum intensity projections in A, C-D, G; single Z slices in F, I. Scale bars: A,C,D,G = 20 µm, F,I = 10 µm. W and red star indicate wound.

Cell fusion often appears as apical shrinking.

A) The apical footprint of an epithelial cell shrinks in the epithelial plane after wounding (yellow arrow). B) Same sample as A, showing that the nucleus (asterisk) of the apically shrinking cell enters a neighboring syncytium, outlined in Biv. C) Cell 1 expresses actin-GFP before wounding. After wounding, cell 1 fuses with cells 2-6 evidenced by GFP sharing, then the apical surface of cell 1 shrinks. D) Two cells undergoing apical shrinking are identified in the X-Y plane with white and yellow arrows, shown at three times after wounding. Below, two X-Z projections are shown for each, at Y1 and Y2 as indicated, with the Z-plane of the adherens junctions (ZAJ) indicated by dotted lines. Rather than extrude, cytoplasm moves to the right as the cell depth in the Z-axis diminishes, so that cytoplasm joins with the neighboring wound-proximal syncytium. These video frames are a continuation of the sample shown in Figure 1I. E) All cells displaying apical shrinking mapped to first frame after wounding. F) All apical shrinking and border breakdowns were tracked in 3 wounds. Border-breakdown fusions happen sooner after wounding than apical-shrinking fusions. A, B, C, D (top), and E show maximum intensity projections. Lower panels in D show X-Z projections. Scale bar for all panels A and B shown in Biv. A, B, C, D (top),E = 10 µm; D (X-Z projections) = 5 µm.

Half the cells near the wound fuse to form syncytia, demonstrated by tracking individual cell fates.

A) All GFP labeled cells in the region of fusion (80 µm) were tracked in 5 wounds over 6.5 h after wounding to determine frequency of fusion initiation (GFP mixing) and later morphological changes into syncytia. All morphological fusion events were preceded by GFP sharing, and 90% of GFP sharing events were followed by morphological fusion. Untrackable cells lost GFP, see Figure 3—figure supplement 1A. B-C) Pre-wound location of persisting and fusing cells from panel A is shown with respect to the wound center. Panel B shows that fusion is common within 70 µm. Panel C shows that apical-shrinking fusion and border-breakdown fusion occur at similar distances from the wound.

Atg1 promotes wound-induced fusion and faster wound closure.

A) Overview: Atg1 was knocked down in central pnr domain (pink), visualized by nls-mCherry (not shown). A laser pulse was targeted to the edge of the pnr domain (right or left) so that the wound extended into both Atg1 knockdown and control domains, allowing the symmetry of response to be analyzed. B-C) Inhibition of Atg1 decreased the frequency of wound-induced fusion in the pnr domain. For border loss, n = 11 pupae, p < 0.0001. For apical shrinking, n=9 pupae, p = 0.0008 (paired T-test). D-E) Wound closed more slowly when Atg1 was knocked down. (D) Radius of the wound was calculated by the distance between the wound center and the leading edge in pnr or control domains. (E) Difference in the wound radius between pnr and control domains over time (i.e., wound asymmetry) is shown. Each dot represents one wound. With no genetic manipulation in either domain (black), wound radius was slightly larger in pnr than in control domain. When Atg1 was knocked down in the pnr domain, the asymmetry grew over time indicating slower wound closure on the Atg1 knockdown side. p = 0.0031 (5 hr) calculated from two-way ANOVA, fit full model, comparing each cell mean with the other cell mean in that row. error bar = 1 +/- SEM. F-G) Larger cells (i.e. syncytia) move further during wound closure, evident in control wounds (F) or with Atg1 knockdown (G). Each data point represents a cell or syncytium present at the leading edge when the wound is half closed. The curve of square root area was overlaid for comparison, as it represents the amount of movement expected based on reshaping of a cell/syncytium from an initial round state to one strongly elongated towards the wound.

Syncytia outpace mononuclear cells.

A) Before wounding, two clusters of cells are labeled with GFP, 1-2 and 3-4 (white numbers). After wounding, cells 2 and 3 fuse with neighbors 5-9 to form a syncytium (dashed leading edge in Aiii), which advances toward the wound, passing unfused cell 4 (arrow). B-C) Unfused cell (outlined in orange and white) is replaced at leading edge by neighboring syncytia (S and yellow outline). D) Leading-edge perimeter was analyzed over time for three wounds, also analyzed in panel G. Percent leading edge occupied by syncytia increased over time to 100%. E) Images of sample 1 from graph D. Unfused cells (arrows) were tracked over the course of wound closure; 30 min after wounding, 10 unfused cells at the wound leading edge are indicated by arrows. F) At 120 min, the last unfused cell was ejected from leading edge (arrow). The wound closed at 140 min. G) The loss of unfused cells was analyzed over time in the three wounds from panel D. All unfused cells are excluded from leading edge by syncytia well before each wound closed 20-160 min later. Images are single Z slices in Ai-Aii and maximum intensity projections in Aiii-F. Scale bar for B,C shown in Civ and for E,F shown in F. Scale bars in A,E,F = 20 µm, B,C = 10 µm.

Cells fuse along different axes at different frequencies, reducing the number of wound-proximal cell intercalations.

A) Illustration of tangential vs radial border breakdown. B) More tangential than radial borders break down after wounding. Data are from four wounds, with a total of 235 border breakdowns: 39 radial and 196 tangential. Individual counts are shown per wound with mean and SEM. C) Fusions across radial borders reduce intercalation at a wound. D) Quantification of intercalations and fusions around the three wounds of Figure 5D,G.

Computational modeling indicates that fusion speeds the rate of wound closure because of reduced cell intercalation.

(A-B) Experimental observations of cell shape index (perimeter divided by square root of area). By two hours after wounding, the cell shape index has increased for both syncytia and non-syncytial cells within two rows of the wound margin. These increases indicate an increase in tissue fluidity. (A) Cells and syncytia were segmented, and cell shape index calculated, as shown for this image at 115 min after wounding; cell shape index indicated by color-coding. (B) Pre-and post-wound cell-shape distributions were compiled from 6 pupae. The red dashed line represents the critical cell shape parameter, 3.81, for the solid-to-fluid transition in vertex models of epithelia. Grey boxes span the interquartile range with a line at the median; whiskers extend out 1.5x the median-to-closest-quartile range; and outliers beyond the whiskers are plotted as individual points. Distributions include 1304 pre-wound mononucleate cells, 168 post-wound mononucleate cells, and 52 post-wound syncytia; **** denotes p < 10-6 level; *** denotes p < 0.001. (C-E) Vertex model simulations of wound closure with cell fusion allowed or suppressed. When the spatial and temporal probabilities for cell fusion match experimental observations, the model yielded 55 ± 22 fusions per wound (mean ± standard deviation), with 15 ± 6 of these occurring at radially aligned edges. As shown in C, fusions in the model speed wound closure: four simulations per condition; shaded regions denote standard deviation; time normalized to the average time to closure when fusion is suppressed. Panels D-E show a series of still frames from a matched pair of simulations in which fusion is allowed or suppressed. Corresponding videos are available as Video 5.

Syncytia concentrate pooled resources at the leading edge.

Random scattered cells expressing Actin-GFP were generated by heat-shock mediated flip-out expression of Gal4. A-B) Labeled actin is expressed in cell 1 before wounding (Ai,Bi). These cells are not close to the wound (inset in Aii). By 28 min after wounding, actin-GFP equilibrated between cells 1 and 2 (Aii,Bii), demonstrating cytoplasmic fusion. The resulting syncytium had no access to the leading edge, and actin remained equilibrated, as shown in the kymograph (Biii) generated from actin-GFP intensity over time at the yellow line in Bii. Fusion was confirmed at 180 min after wounding by apical shrinking of cell 2 (not shown). C-D) Labeled actin is expressed in cell 1 before wounding (Ci,Di) and equilibrates between cells 1 and 2 by 6 minutes after wounding, demonstrating cytoplasmic fusion (Cii,Dii). The resulting syncytium contacts the leading edge, and by 28 min after wounding actin from cell 1 is redistributed to the wound margin (Ciii,Diii), as shown in the kymograph (Div) of actin intensity over time at the yellow line in Diii. Fusion was confirmed at 55 min after wounding, by apical shrinking of cell 1 (not shown). E-F) Before wounding, actin-GFP in cell 1 is three cells away from the future leading edge. After wounding, fusion of cells 1-4 allows actin-GFP to be subcellularly localized to the leading-edge actin cable. Border breakdown is visible between cells 1-2 at 30 min after wounding (Ev) and apical shrinking occurs later (see Video 6). G) Mean profile plot of actin-GFP comparing the syncytia in Fv at 30 min after wounding (dark line) with the cells in Fi before wounding (dotted line) demonstrating that nearly all actin-GFP has been relocalized to the leading edge from its starting position 20-30 µm away. Single Z slices for Ei - ii, Fi - ii; maximum intensity projections for A-D, Eiii - v, Fiii - v. Scale bar for panels in A-B shown in Bii, 10 µm. Scale bar for panels in C-D shown in Diii, 5 µm. Scale bar for panels E-F shown in Fv, 10 µm.

Characteristics of the pupal notum epithelium.

A-C) An example of mitosis occurring ∼80 µm from the wound. Scale bar for A-C shown in Cv, 10 µm. D) Z-slices at different depths of GFP-labeled cells reveal that the apical area (red outline) does not reflect the position of cells at basal slices (yellow arrows). The nuclear shadow in panel Dii demonstrates the location of the nucleus. Scale bar, 10 µm. E-H) Ecad-GFP and p120ctnRFP colocalize and behave similarly during border breakdown. Arrowheads in F-H points to borders breaking down in first hour after wounding. Scale bar for F-H shown in Hv, 10 µm.

Temporal analysis of fusion events

A) An example of an untrackable cell is shown. After wounding, cell 1 loses cytoplasmic GFP, but there is no obvious recipient cell. Cell 1 then shrinks. Scale bar = 10 µm. B) Cells individually labeled with GFP reveal the timing and frequency of fusion or GFP loss (untrackable). C) Fusion cells from panel B were divided into two types of fusion events, shrinking cells and border breakdowns, to compare the temporal onset of each type of fusion. D) Cell shrinking was a lengthy and variable process, lasting up to hundreds of minutes. 274 shrinking cell events from three wounds were identified by p120ctnRFP, mean and SD indicated.

Comparisons of tangential and radial border breakdowns

A) The distance of tangential and radial border breakdown events from the wound, compared over time binned into 10 min intervals. The data is the same as in Figure 6B, from four wounds, with a total of 235 border breakdowns: 39 radial and 196 tangential. Border breakdown events were identified by p120ctn. B) The timing of tangential vs radial border breakdown fusion events observed in single GFP-labeled cells.