JNK-mediated Slit-Robo signaling facilitates epithelial wound repair by extruding dying cells

Multicellular organisms repair injured epithelium by evolutionarily conserved biological processes including activation of c-Jun N-terminal kinase (JNK) signaling. Here, we show in Drosophila imaginal epithelium that physical injury leads to the emergence of dying cells, which are extruded from the wounded tissue by JNK-induced Slit-Roundabout2 (Robo2) repulsive signaling. Reducing Slit-Robo2 signaling in the wounded tissue suppresses extrusion of dying cells and generates aberrant cells with highly upregulated growth factors Wingless (Wg) and Decapentaplegic (Dpp). The inappropriately elevated Wg and Dpp impairs wound repair, as halving one of these growth factor genes cancelled wound healing defects caused by Slit-Robo2 downregulation. Our data suggest that JNK-mediated Slit-Robo2 signaling contributes to epithelial wound repair by promoting extrusion of dying cells from the wounded tissue, which facilitates transient and appropriate induction of growth factors for proper wound healing.

Wound repair is an evolutionarily conserved process that maintains tissue homeostasis upon injury [1][2][3] . It has been reported that JNK signaling acts as an essential regulator of wound repair in Drosophila epithelial tissue 4-7 , planarians body 8,9 , and zebrafish tail fin 10 . Genetic studies in Drosophila have shown that JNK signaling contributes to (1) actin remodeling to close wound edges 6,11 , (2) reconstruction of lost tissue parts by activating growth promoters such as Yorkie (Yki, a YAP homolog) 12,13 , Wg (a Wnt homolog) 14 , Dpp (a TGF-β/BMP family member) 15 and Myc 14 , (3) facilitating cell reprograming via reducing the activity of polycomb-dependent silencing 16 , and (4) induction of developmental delay by upregulating Drosophila insulin-like peptide 8 (Dilp8) to prolong the developmental period for recovery 17 . Particularly, JNK-dependent induction of Wg promotes regenerative growth of Drosophila wing imaginal discs after genetic ablation of the tissue 14 . In addition, JNK-mediated upregulation of Wg and Dpp plays critical roles in compensatory proliferation of imaginal cells after induction of massive cell death 15,18,19 . JNK signaling also induces apoptosis 20,21 , which is required for regeneration of planarian body 9 or wound repair in Drosophila epithelial tissue [22][23][24] . Together, JNK regulates multiple steps of repair process from beginning to end.
Dying cells emerged in the epithelial tissue are extruded basally or apically by a coordinated mechanism 25 . For instance, overcrowding of cells within a limited space triggers extrusion of living or dying cells from Madin-Darby canine kidney (MDCK) epithelial monolayer 26 , developing zebrafish tail fin 26 , and Drosophila notum 27 . In Drosophila embryonic development, extrusion of apoptotic cells from amnioserosa promotes dorsal closure 28,29 , the process that shares common JNK-dependent events with epithelial wound repair, which include actin remodeling, cell migration, and epithelial zipping 30,31 . Similarly, JNK-dependent cell extrusion is required for tumor-suppressive cell competition, the process in which oncogenic polarity-deficient cells such as scribble (scrib) or discs large (dlg) mutant cells are actively eliminated from epithelia when surrounded by wild-type cells [32][33][34][35][36][37] . Importantly, extrusion of polarity-deficient cells by cell competition is driven by JNK-mediated activation of Slit-Robo2 axonal repulsive signaling that downregulates E-cadherin, as the ligand Slit, its receptor Robo2, and the downstream effector Enabled (Ena)/Vasp are all induced by JNK signaling 35 . During Drosophila Slit-Robo2 signaling promotes extrusion of dying cells from the wounded tissue. Our finding that Slit-Robo2 signaling plays a role in epithelial wound repair suggests that JNK-mediated cell extrusion is required for this process. We thus analyzed spatial locations of dying cells in the wounded tissue by immunostaining for the cleaved form of the effector caspase Dcp1 (c-Dcp1). In wild-type background, the number of dying cells in the wing pouch significantly increased at 6hrs after wounding ( Supplementary Fig. 4a,b, quantified in Supplementary Fig. 4i). Importantly, the number of dying cells within the disc was 4-fold higher at the earlier time point (Fig. 2g, 3hrs, quantified in Fig. 2j, compare to Fig. 2b, 6hrs, quantified in Fig. 2e), suggesting that dying cells are extruded from the tissue over time. Supporting this notion, the analysis of extruding/extruded dying cells in the wounded discs by classifying their locations into three classes ("in disc", "apically extruding", and "basally extruding";  Fig. 5f,g, quantified in Supplementary Fig. 5l) after wounding, while the tendency was not observed at 3hrs after wounding likely because extrusion has not proceeded sufficiently at this time point even in the wild-type tissue ( Supplementary Fig. 5a,b, quantified in Supplementary Fig. 5j).
Intriguingly, it has been reported that an axon guidance molecule PlexinA plays an important role in cell extrusion during epithelial wound repair in Drosophila and zebrafish 45 . In addition, Slit has been proposed to bind to PlexinA in mammals 49 , suggesting that multiple axon guidance signaling contribute to wound healing by promoting dying cell extrusion. Our findings suggest that dying cells remained in the tissue with excess growth factors need to be removed for proper wound healing, by promoting epithelial fusion and/or facilitating transient and appropriate production of growth factors.  55,56 , dpp d6 (DGGR #106644) 57 , dpp s11 (DGGR #106646) 57 , dpp hr92 (DGGR #106649) 58 , TRE-DsRed (BDSC #59012) 48 , P{PZ}dpp 10638 (dpp-lacZ, BDSC #12379) 52 . physical in situ wounding. 3 rd instar wandering larvae were randomly collected and anesthetized with ice-water for around 10 minutes. Then their wing discs (which were marked by fluorescent proteins GFP or Venus) were injured on ice with a sharpened 0.3 mm tungsten needle by performing aseptic in situ wounding (by pushing the wing pouch region using the needle) in living larvae without further damaging the animal. Wounding was performed under the fluorescence binocular microscope. After wounding, larvae were cultured in fresh food vials and kept at 25 °C again. Late 3 rd instar larvae before wondering were wounded only when we analyzed wing discs 24 hrs after wounding. See Supplementary Fig. 1 for further information.

Methods
Histology. Larval tissues were fixed and stained using standard immunohistochemical methods with rab-

Analysis of dying cells.
Apoptotic dying cells were detected by c-Dcp1 antibody. For the analysis of spatial locations of c-Dcp1-positive dying cells, the locations were classified into 3 classes: (1) "basally extruding", as dying cells located at the basal tip of the disc proper, (2) "apically extruding", as dying cells located at the apical tip of the disc proper, and (3) "in disc", as dying cells located within the disc proper (see Fig. 2a for further information). Wing discs with wound that crosses center of the wing pouches were analyzed for special locations of c-Dcp1-positive cells. c-Dcp1-positive cells classified as "in disc" were manually counted using xz or yz cross-section images, and pouch areas were manually measured with Fiji and calculated with Microsoft Excel for Mac. For the analysis of total number of c-Dcp1-positive cells in the pouch, the number of c-Dcp1-positive cells in the wing pouch and the size of the wing pouch areas were automatically counted using Z-stacked images with Fiji and calculated with Microsoft Excel for Mac. For the analysis of necrotic dying cells detected by PI staining, the number of PI-positive cells in the wing pouch were automatically counted using single xy cross-section images with Fiji and calculated with Microsoft Excel for Mac.