Recapitulation of morphogenetic cell shape changes enables wound re-epithelialisation

Wound repair is a fundamental, conserved mechanism for maintaining tissue homeostasis and shares many parallels with embryonic morphogenesis. Small wounds in simple epithelia rapidly assemble a contractile actomyosin cable at their leading edge, as well as dynamic filopodia that finally knit the wound edges together. Most studies of wound re-epithelialisation have focused on the actin machineries that assemble in the leading edge of front row cells and that resemble the contractile mechanisms that drive morphogenetic episodes, including Drosophila dorsal closure, but, clearly, multiple cell rows back must also contribute for efficient repair of the wound. Here, we examine the role of cells back from the wound edge and show that they also stretch towards the wound and cells anterior-posterior to the wound edge rearrange their junctions with neighbours to drive cell intercalation events. This process in anterior-posterior cells is active and dependent on pulses of actomyosin that lead to ratcheted shrinkage of junctions; the actomyosin pulses are targeted to breaks in the cell polarity protein Par3 at cell vertices. Inhibiting actomyosin dynamics back from the leading edge prevents junction shrinkage and inhibits the wound edge from advancing. These events recapitulate cell rearrangements that occur during germband extension, in which intercalation events drive the elongation of tissues.

To quantify the pulses of junction shrinking, junction length was tracked manually from movies (commencing 10 minutes post-wounding) of wounds made to E-cadherin-GFP-expressing embryos. Pulses were considered to have occurred when the change in junction length exceeded one standard deviation above the mean change for three consecutive time points.
To plot the location of myosin pulses in wound edge cells relative to the AP junction, we measured the distance between the centre of each myosin pulse from the nearest vertex as a proportion of the length of the AP junction (x-axis). The same was performed along the y-axis to measure the pulse distance from the AP junction as a proportion of the cell width. For wound edge cells, the myosin pulses were measured relative to the AP junction between cell rows one and two. For cells back from the wound edge, measurements were made relative to the AP junction in these cells closest to the wound.
The junction shrinkage as a result of each myosin pulse was calculated by measuring the length of the junction one time point before the myosin pulse and again after the pulse had resolved. The Pearson r value of myosin pulses versus junction shrinkage was assessed by drawing a 3 µm ROI around the AP junction and using this to measure the Sqh-GFP average intensity. The junction shrinkage was tracked manually and the change in Sqh-GFP intensity and junction length between timepoints used to calculate the r value. The time shifts were achieved by shifting change in the myosin intensity data set relative to change in the junction length data set.
To measure Bazooka breaks and E-cadherin intensity at vertices, 15-minute periods of wound closure were analysed (movies started 20 minutes post-wounding). A 0.58 µm ROI was drawn around the vertex and used to measure average Bazooka-GFP intensity, and a 1.54 µm ROI to measure average mCherry-moesin intensity. The average intensities are expressed relative to a linear regression analysis of the data. A break was identified if the Bazooka-GFP intensity dropped one standard deviation of the data set below a relative Bazooka-GFP intensity of 1 for three consecutive time points. The intensity of Bazooka-GFP and mCherry-moesin during the break period and four time points either side was used to calculate Pearson r values. For identifying pulses of actin in Bazooka mutant embryos, GFP-moesin intensity was analysed in 20-minute movies, starting 20 minutes post-wounding. The GFP-moesin average intensity at cell vertices was monitored with a 2 µm ROI drawn around the vertex and expressed relative to the linear regression analysis of the data. GFP-moesin pulses were identified if the relative intensity exceeded one standard deviation above a relative GFP-moesin intensity of 1 for three consecutive time points.  Figure 3Di with addition of pulses from cells in the second row (blue) (n = 33 myosin pulses from 17 cells from 9 different wounds). (Di) Stills from a movie of the wound edge in an mCherry-moesin (magenta), Sqh-GFP (green, grey in bottom panel) expressing embryo showing no significant accumulation of myosin at the junctions of cells at the wound edge, nor further back from the wound (examples of the junctions followed are highlighted by arrowheads). Graph represents average Sqh-GFP intensity at junctions relative to that in the cell over 60 minutes of wound closure (n ≥ 12 junctions from 5 movies for each). (Dii) Close up still images from a time-lapse of a junction at the wound edge of an mCherry-moesin (magenta), Sqh-GFP (green, also shown in the bottom panel, grey) expressing embryo showing that individual junctions may accumulate myosin during myosin pulses, but this subsequently resolves (the junction observed is highlighted by the arrowheads, and the myosin pulse by the arrow). Graph shows the intensity of myosin at the junction and in the myosin pulse relative to the whole cell intensity, highlighted by the arrowheads/ arrows in the adjacent still images. Wound (W) margins marked by dashed lines. Error bars represent SEM. Scale bars represent 5 µm in all except 15 µm in Di. Time is in seconds for Ai-B and minutes for Di-ii.

Supplementary Figure S3
(Ai) Graph shows relative intensity of Bazooka-GFP (green) and mCherry-moesin (magenta) over several minutes of imaging showing multiple breaks at the same vertex. The grey arrowhead on the graph points to the break shown in Fig. 3Ei and the black arrowhead points to the break represented in Fig. S3Ai. The second break in Bazooka-GFP (green, and grey in the bottom panel) is shown in the accompanying still images and indicated by the arrowheads (mCherry-moesin shown as magenta). (Aii) A related graph to Ai but showing no breaks in vertices back from the wound edge. The small decrease in Bazooka-GFP (green, and grey in bottom panels) intensity at the vertex (highlighted by the arrowhead) is shown in the accompanying time series stills and shows no break in Bazooka. (Aiii) Graph of junction shrinkage of AP junctions associated with vertices showing either no Bazooka break or one break (vertices from 6 independent wounds). (B) Graph shows a representative plot of E-cadherin-GFP intensity at a vertex of a junction that recruits a pulse of actin (arrowhead) but is not preceded by a break in E-cadherin. The time points surrounding the actin pulse are shown in the accompanying time-lapse stills (E-cadherin-GFP in green, and grey in bottom panels, mCherry-moesin in magenta). (Ci-Cii) Graphs represents relative Moesin-GFP intensity at a vertex in a control embryo versus a Baz Xi106 mutant embryo. Peaks of Moesin-GFP intensity represent pulses of actomyosin at the vertex, the number of which are reduced in Baz Xi106 mutants. Arrowheads on graphs indicate the time points shown in the accompanying time-lapse stills showing an actin pulse at vertices (arrowheads) in control embryos but not in Baz Xi106 mutants. (D) Graph representing wound edge advancement in Control (black) vs Baz Xi106 mutant (red) embryos (n ≥ 5 wounds for each genotype). Wound (W) margins marked by dashed lines. Times are in seconds for all. Scale bars represent 5 µm in all except in close up views of Bazooka-GFP and E-cadherin-GFP at the vertices which represent 2 µm. Error bars represent SEM. * denotes p<0.05 from a Student's t-test in Aiii and via a Two-way ANOVA with a Bonferroni post test in D.