Requirement for and polarized localization of integrin proteins during Drosophila wound closure.

Wound reepithelialization is an evolutionarily conserved process in which skin cells migrate as sheets to heal the breach and is critical to prevent infection but impaired in chronic wounds. Integrin heterodimers mediate attachment between epithelia and underlying extracellular matrix and also act in large signaling complexes. The complexity of the mammalian wound environment and evident redundancy among integrins has impeded determination of their specific contributions to reepithelialization. Taking advantage of the genetic tools and smaller number of integrins in Drosophila, we undertook a systematic in vivo analysis of integrin requirements in the reepithelialization of skin wounds in the larva. We identify αPS2-βPS and αPS3-βPS as the crucial integrin dimers and talin as the only integrin adhesion component required for reepithelialization. The integrins rapidly accumulate in a JNK-dependent manner in a few rows of cells surrounding a wound. Intriguingly, the integrins localize to the distal margin in these cells, instead of the frontal or lamellipodial distribution expected for proteins providing traction and recruit nonmuscle myosin II to the same location. These findings indicate that signaling roles of integrins may be important for epithelial polarization around wounds and lay the groundwork for using Drosophila to better understand integrin contributions to reepithelialization.

INTRODUCTION

A skin wound exposes underlying tissues and the entire organism to further damage and infection and so must be healed quickly. “Reepithelialization” (RE), where sheets of skin cells migrate toward and reseal the wound, is a critical part of wound healing. Our knowledge of RE mechanisms is informed by studies of migratory single cells that must polarize, generating functionally different front and rear sides, as well as engage contractile mechanisms to exert force. Finally, controlled adhesion and deadhesion enable movement across a substratum, typically the ECM. Small GTPases of the Rho family mediate front/rear polarization (Ridley ; Charest and Firtel, 2007; Parsons ), and actomyosin supplies propulsive forces, with actin polymerization driving lamellipodia extension (Vicente-Manzanares ; Parsons ). Integrins are the major receptors for the extracellular matrix (ECM) and are essential for cell crawling (Scales and Parsons, 2011; Maartens and Brown, 2015). Extensive but poorly understood mutual regulation among the Rho-GTPases, actomyosin, and integrins directs the forward migration of cells (Ridley ; Gupton and Waterman-Storer, 2006; Parsons ).

Cells migrating as sheets employ similar polarization, contractile, and ECM adhesion mechanisms but must additionally coordinate these across multiple cell diameters and integrate adhesion to both substratum and adjacent cells (Friedl ; Collins and Nelson, 2015; Das ; Haeger ; Ladoux ; Mayor and Etienne-Manneville, 2016). In wound healing, further complexity is added by the transient dedifferentiation of epithelial cells to a motile phenotype (Haensel and Dai, 2018), the regulated mobilization of cells within several rows of the wound (Farooqui and Fenteany, 2005; Matsubayashi ; Richardson ), the need for the migrating epithelium to remodel the underlying ECM (DiPersio ; Nguyen ; Longmate ), and the inflammatory environment (Gurtner ; Shaw and Martin, 2009; Eming ).

The physiological complexity of wound healing indicates that the widely employed in vitro scratch wound assays (Boyko ) should be supplemented by in vivo studies. Drosophila is a valuable model for the genetic analysis of wound healing, as it has been studied at several developmental stages (Kiehart ; Wood ; Galko and Krasnow, 2004; Munoz-Soriano ). Fly epidermal cells are analogous to mammalian keratinocytes in their developmental specification, secretion of protective stratum corneum/cuticle, and regulation after wounding (Mace ; Ting ; Munoz-Soriano ). In both mammals and flies, cells in a restricted number of rows surrounding a wound polarize and, influenced by growth factor ligands from the wound, migrate as a sheet to heal the wound (Barrientos ; Wu ). Extensive ECM remodeling during wound repair also occurs in both (Stevens and Page-McCaw, 2012; Xue and Jackson, 2015).

Especially important in RE are the integrins, heterodimers of α and β transmembrane glycoproteins that bind the ECM. Integrins nucleate large cytoplasmic complexes that not only tune adhesion in response to both intracellular and extracellular cues (Horton ; Iwamoto and Calderwood, 2015) but also engage in bidirectional signaling, affecting cytoskeletal activities and gene expression inside the cell (Legate ), and shaping ECM composition on the outside (deHart ; Humphrey ). Integrin abundance and function are also regulated by transcriptional modulation (Homsy ), switching of different integrins (Truong and Danen, 2009; Elosegui-Artola ), protein clustering and localization (Yamaguchi ), vesicle trafficking (Paul ), and protein turnover (Webb ; Lopez-Ceballos ). Mammals have 18 α and eight β integrins, with extensive combinatorial diversity (Hynes, 2002), and their importance in wound healing is underscored by changes in expression (Grinnell, 1992; Gailit ), wound repair defects in animals with integrin-deficient keratinocytes (Grose ; Russell ; Egles ; DiPersio ), and the integrin dysregulation seen in human scarring disorders (Wang ). The wound phenotypes of many mouse keratinocyte integrin knockouts suggest some functional redundancy, which along with the complexity of the mammalian wound environment has hindered the determination of which adhesion, sensing, and signaling activities of which integrins effect the polarization and traction needed for RE (Margadant ; Martins-Green, 2013; DiPersio ).

While integrin roles in fly wound repair have not been emphasized, a conserved function is indicated by the induction of integrins in epithelial cells surrounding the wound (Lee ) and embryonic wound closure defects in αPS3 mutants (Campos ). Both integrins and their ECM ligands are conserved between flies and mammals (Broadie ), but the much smaller number of fly integrins greatly facilitates genetic analysis (Bokel and Brown, 2002). The availability of reagents for the tissue-specific knockdown of any gene in Drosophila, including in multiples, is particularly conducive to the study of integrins, which operate in multiple tissues during wound repair, function as heterodimers, and frequently exhibit redundancy. Therefore, we undertook to develop Drosophila as a model to study integrin function in wound RE and report here our systematic analysis of the epithelial requirements for specific α and β integrins and core adhesion components in wound closure, the pathways regulating their accumulation, and their contributions to cell polarization.

RESULTS AND DISCUSSION

αPS2/βPS and αPS3/βPS integrin dimers both contribute to epithelial wound closure

To examine integrin involvement in wound healing RE, we knocked down α and β integrins by expressing UAS-RNAi transgenes in the larval epidermis using A58-GAL4 (Galko and Krasnow, 2004). For each gene, we tested nonoverlapping RNA interference (RNAi) lines to exclude off-target effects (Mohr and Perrimon, 2012) and coexpressed Dicer-2 when necessary to enhance knockdown efficiency (Dietzl ). We used a pinch-wound protocol that creates a hole in the epidermis ∼30–40 cells wide but leaves the overlying cuticle intact, thus minimizing mortality from hemolymph leakage (Kwon ). As we and others have found, such wounds are closed within hours by RE (Galko and Krasnow, 2004; Lee ) (Figure 1, A and B).

FIGURE 1:

Integrins βPS, αPS2, and αPS3 contribute to epithelial wound closure. (A–H) Wound closure was examined in the epidermis of late third instar larvae. Cell boundaries were visualized by FasIII immunostaining in red, and cell nuclei were marked by DAPI staining in blue. (A, B) A58-GAL4-only controls, unwounded (A), and dissected at 16 h after wounding (B). (C, D) A58-GAL4 UAS-βPS RNA-i (βPS knockdown in the larval epidermis of third instar larvae, denoted A58>βPS-i, hereafter) at 16 h (C) or 40 h (D) after wounding. Asterisks indicate the wound hole. Dashed line indicates the original wound margin. (E–H) Epidermis was dissected at 14 h after wounding. (E) A58>Dcr-2 only control. (F) A58>αPS1-i. (G) A58>αPS1-i. (H) A58>αPS2-i. Scale bar: 100 μm (A–H). (I–N) Knockdown efficiencies were analyzed by immunostaining in RNAi-expressing larvae using anti-βPS (I, J), anti-αPS2 (K, L), and anti-αPS3 (M, N) antibodies in A58-only control (I, K, M), A58>βPS-i (J), A58>αPS2-i (L), and A58>αPS3-i (N) larvae. Yellow arrowheads indicate trachea. Scale bar: 25 μm (I–N). (O) Quantification of the wound closure defects in the larvae of indicated genotypes. At least six animals were analyzed for each genotype.

Epidermal knockdown of βPS (myospheroid) caused severe RE defects, with 100% or 87% of the wounds still open after 16 h, depending on the RNAi line (Figure 1, C and O, and Supplemental Figure 1A). By 40 h, 0% or 75% of the wounds in βPS knockdown larvae had closed, in a delayed manner (Figure 1, D and O, and Supplemental Figure S1B). Wound measurements 20 min after injury in wild-type (WT) vs. βPS knockdown larvae found no significant difference in size (Supplemental Figure S2). In further support for a model whereby slowed wound healing in βPS-deficient epithelia is due to delayed RE rather than excessive wound sizes, we observed that βPS-deficient epithelia failed to display the dynamic shape changes around the wound that normally accompany RE (Figure 1, B–D).

Immunostaining confirmed efficient knockdown of βPS (Figure 1, I and J, and Supplemental Figure S1C). Homozygotes for null alleles of the other β-integrin, βv, are viable and displayed normal wound closure (Figure 1O and Supplemental Figure S1D), indicating βPS is the only β subunit required for RE.

To identify the αPS subunit(s) that dimerize with βPS integrin in wound closure, we knocked down αPS1 (mew), 2 (if), and 3 (scab) in the epidermis, but none of these caused defects, with wounds closing efficiently by 14–15 h (Figure 1O and Supplemental Figure S1, E–J). Immunostaining detected no αPS1 in the WT epidermis, consistent with the lack of knockdown phenotype (data not shown). Expression of αPS1-RNAi in the wing disk phenocopied the wing blisters of null mutants (unpublished data; Brower and Jaffe, 1989), indicating the αPS1 RNAi constructs were functional. Anti-αPS2 and anti-αPS3 antibodies detected the corresponding proteins primarily in the plasma membrane of epidermal cells (Figure 1, K and M); in the knockdown larvae, expression levels were severely reduced (Figure 1, L and N, and Supplemental Figure S1, L and M). Altogether, these results indicate that the lack of phenotype was not due to knockdown failure.

We then examined double knockdowns for the three αPS genes. Of the three different combinations, only the αPS2-αPS3 RNAi pair showed defects, with 96%–82% of the larvae displaying open wounds at 14 h after wounding, (Figure 1, E–H and O, and Supplemental Figure S1K). This reveals that PS2 and PS3 integrins do function in larval epidermal wound closure but may compensate for each other. This is reminiscent of the contemporaneous but distinct roles that αPS2 and αPS3 play in glial cell migration and axon guidance (Tavares ).

Talin is required for wound closure, unlike other integrin adhesion components

Integrins function in complexes such as focal adhesions composed of over 50 different proteins (Horton ), so we tested 12 of the core complex members. RNAi knockdown of talin caused wound closure defects as severe as the βPS defects, with 100% of the wounds still open at 16 h (Figure 2, A and B, and Supplemental Figure S1, N–P).

FIGURE 2:

Talin is required for efficient wound closure, unlike other components and regulators of integrin-containing adhesive structures. Analysis of wound closure in the epidermis of late third instar larvae at 16 h. (A) A58>talin-i. Cell boundaries were visualized by FasIII immunostaining in red, and cell nuclei were marked by DAPI staining in blue. Asterisk indicates the wound hole. Scale bar: 100 μm. (B) Quantification of the wound closure defects at 16 h in epithelial RNAi knockdowns of 12 genes encoding components and regulators of integrin adhesive complexes. Asterisks indicate the cases where UAS-Dcr-2 was added to A58-GAL4 to enhance knockdown efficiency. For each genotype, at least six wounded animals were analyzed. (C) Analysis of the knockdown efficiency of each UAS-RNAi transgene using hs-GAL4 and qRT-PCR. Cross symbols indicate not tested (for talin, see Supplemental Figure S1, O and P; for pinch, see Wang ). Error bars represent standard errors of the mean.

In contrast, knockdown of 11 other integrin complex members failed to cause RE defects, with 100% wound closure at 16 h in all larvae examined. In most cases, we tested multiple RNAi lines, as well as coexpressed Dicer-2 to increase efficiency, but in no case observed a wound closure defect (Figure 2B). To further validate the knockdown efficiency, we drove each of the UAS-RNAi lines using hs-GAL4 and performed quantitative real-time PCR on the whole larvae. The results indicate that we had at least one RNAi line for each gene that caused greater than 70% reduction in expression, except in the case of Fak, CalpB, CalpC, and p130CAS (Figure 2C).

A lack of mutant phenotypes for core integrin components has been observed in many Drosophila tissues (Alatortsev ; Grabbe ; Bulgakova ; Moreira ; Maartens ) and suggests that a robust integrin adhesion structure exists in which components can substitute for each other (Bulgakova ). However, the requirement for PINCH and integrin-linked kinase (ILK) in maintaining integrity of the unwounded epidermis (Wang ) suggests that the requirement for individual integrin adhesion components is context dependent. In keeping with this idea, even structural domains within adhesome proteins have context-dependent requirements (Ellis ). Talin, however, is an indispensable component of integrin adhesions, in both Drosophila and mammals, necessary for virtually all adhesion and bidirectional signaling functions of integrins (Brown ; Klapholz and Brown, 2017).

Induction and subcellular localizations of α and β integrins after wounding

We recently reported wound induction of βPS integrin in the Drosophila epidermis (Lee ) and so sought to characterize this expression more thoroughly. βPS expression was markedly increased in the first four rows of cells surrounding the wound margin (Figure 3, A and B), with expression at 7 h post wounding induced 3.1-fold in the first row, 2.4-fold in the second row, and 1.6-fold in the fourth row (Figure 4C). Increased βPS expression lasted past wound closure at 16 h (Figure 3C). Quantitative real-time PCR (qRT-PCR) analysis indicates that at least some of the increased βPS abundance is due to transcriptional up-regulation, although we cannot rule out other mechanisms such as protein stability (Figure 4D). The accumulation of integrin in cells several rows from the wound is consistent with an important role for such “follower cells” in driving sheet migration toward the wound (Farooqui and Fenteany, 2005; Trepat ; Matsubayashi ; Richardson ). Close examination revealed an enrichment of βPS protein at the distal edge of cells surrounding the wound, as has been seen (Figure 3, B and H) (Stevens and Page-McCaw, 2012).

FIGURE 3:

Induction and interdependence of α- and β-interins after wounding. Analysis of βPS, αPS2, and αPS3 induction on wounding via immunostaining using anti-βPS (A–C, H–J), anti-αPS2 (D, E), and anti-αPS3 (F, G) antibodies, respectively. Cell nuclei were marked by DAPI staining in blue. Asterisks indicate the wound hole. (A–C) βPS expression in A58-only controls at 0 h (A), 7 h (B), and 16 h (C) after wounding, respectively. (D, E) High-magnification images of αPS2 induction at 7 h in cells surrounding a wound in A58-only controls (D) and A58>βPS-i (E). (F and G) αPS3 induction at 7 h in cells surrounding a wound in A58-only controls (F) and A58>βPS-i (G). (H–J) βPS induction at 7 h in cells surrounding a wound in A58-only controls (H), A58>αPS2-i (I), and A58>αPS3-i , Dcr-2 (J). Scale bar: 100 µm (A-J).

FIGURE 4:

JNK-dependent induction and distal localization of βPS integrin in cells surrounding wound. (A, B, E, F) Analysis of βPS induction (A, B) and distal localization (E, F) via immunostaining using anti-βPS antibody in red at 7 h after wounding in A58-only controls (A, E) and A58>bsk (B, F) larvae. Cell nuclei were marked by DAPI staining in blue, and asterisks indicate the wound hole but not necessarily the wound center. Yellow arrowheads indicate occasional strong staining in cell debris at the wound edge. These regions were excluded in measurement, as they did not reflect gene expression in living cells. Scale bar: 100 µm. (C) Quantification of the fluorescence intensities of βPS immunostaining in each of the first four rows of cells from the wound margin in WT vs. JNK-deficient epithelia. (D) qRT-PCR quantification of fold transcriptional induction of βPS gene in epithelial fillets at 7 h after wounding. In the first two bars, the βPS induction level at 7 h was compared with that of msn, which is also induced prominently in the first three to four rows of cells from the wound margin (qRT-PCRs were performed using RNAs prepared from the same epidermal fillets). In the remaining four bars, βPS mRNA expression in A58-only controls was compared with that in A58>bsk at 7 h after wounding. Error bars represent standard errors of the mean. (G, H) Plotting of the fluorescence intensities of βPS immunostaining along the front-to-rear axes of individual cells, as marked in white lines in E and F. (I) Quantitative comparison of βPS protein localization in each row of cells between A58-only control and A58>bsk larvae using a “rear localization index.” See Materials and Methods for the statistical analysis in C and I.

Both the α-integrins that contribute to wound closure were also induced by wounding, although this increased expression was in a narrower band of cells than βPS, approximately two rows around the wound (Figure 3, D and F). Notably, αPS2 had a polarized expression pattern similar to βPS, including enrichment in posterior protrusions (Figure 3D).

We then tested the dependence of the α and β subunits on each other for expression and localization. βPS knockdown drastically reduced the plasma membrane localizations of αPS2 and αPS3 (Figure 3, E and G). For αPS2, wound induction of the protein expression was also greatly reduced (Figure 3E). In contrast, levels of βPS induction were normal when either αPS2 or αPS3 were singly knocked down, which we attribute to the redundancy of these two α-integrins in wound RE (Figure 3, H–J). Interestingly, the distal accumulation of βPS was impaired in αPS2 but not αPS3 knockdowns (Figure 3, H–J). In summary, although wound closure phenotypes were only observed in double knockdowns of αPS2 and αPS3, suggesting some functional redundancy, αPS2 colocalizes with βPS at the posterior of cells surrounding the wound more than does αPS3 and shares a mutual dependency on βPS for this colocalization, possibly indicating a more important role for αPS2 than for αPS3 in βPS dimers in the epidermal wound response.

αPS2/βPS dimers are homologous to mammalian α5β1, α8β1, and αVβ1 (Bokel and Brown, 2002; Johnson ). Although αPS3 does not have clear mammal orthologues, it most resembles mammal α4 and α9, and like αPS2 is predicted to bind RGD-containing ligands such as laminin and collagen (Johnson ). Most of these β1-containing integrin dimers are implicated in mammalian wound healing (Koivisto ; DiPersio ). It is not clear whether flies, like mammals, deposit a provisional matrix on the wound bed that is different in composition from the unwounded ECM (Barker and Engler, 2017), but increased synthesis of collagen and deposition of matrix is apparent in fly wounds, and surrounding epithelia migrate over both this and cellular debris (Galko and Krasnow, 2004; Stevens and Page-McCaw, 2012). Notably, the distal accumulation of integrin we observed is opposite to the localization in lamellipodia adhesions at the front of singly migrating cells (Scales and Parsons, 2011).

The JNK pathway is required for the polarized accumulation of βPS integrin around the wound

We next sought to characterize the pathway(s) mediating the induction of integrin and its localization to the distal edges of cells around the wound. The Jun N-terminal kinase (JNK) pathway is activated in a gradient extending away from Drosophila wounds, downstream of Rho-family small GTPases (Ramet ; Galko and Krasnow, 2004; Baek ). Since JNK hyperactivation in unwounded epidermis upregulated βPS integrin (Wang ), we asked whether JNK is required for the wound induction of βPS. Epithelial expression of a dominant negative allele of the JNK-encoding gene bsk (bsk) that blocks JNK signaling (Wu ) greatly reduced βPS accumulation (Figure 4, A–C). Similarly, qRT-PCR analysis in the epithelium revealed JNK contributes to βPS transcriptional induction (Figure 4D).

It appeared that the reduced βPS accumulation in JNK-deficient epithelia was also accompanied by a loss of the polarized distribution of βPS around the wound. To quantify this, we assessed polarized βPS localization with a “rear localization index” (see Materials and Methods). In WT, the values were 0.42 and 0.29 in the first and second rows of cells surrounding the wound, compared with 0.12 and 0.04 in the corresponding cells in bsk-expressing epithelia (Figure 4, E–I). While this supported a role for JNK in generating the polarized accumulation of βPS, it remained possible that the apparent loss of polarized distribution was merely an artifact of the overall reduced βPS expression level. However, when we used RNAi conditions that partially reduce βPS levels, close to the level seen in the bsk epithelia (Supplemental Figure S3B vs. Figure 3B), polarized βPS distribution was still apparent, with only a slight, and not statistically significant, decline in rear localization, analyzed in the first and second rows of cells (Supplemental Figure S3E). Our findings thus indicate that in addition to promoting βPS accumulation, JNK signaling may additionally contribute to the polarized distribution of βPS around the wound.

Since integrins are suggested to regulate JNK through Rho-family GTPases (Schwartz and Shattil, 2000; Homsy ; Matthews ; Pereira ; DiPersio ), we then tested whether integrin function is reciprocally required for JNK pathway activation by wounding but found that knockdown of βPS or talin did not impair the wound induction of msn-lacZ, a JNK pathway reporter (Supplemental Figure S4). Interestingly, in both fly epithelia and cultured cells, loss of integrin adhesions activates JNK (Pereira ; Wang ), suggesting that integrin regulation of JNK during wound healing warrants further investigation.

Our data add to the diverse roles of JNK in wound healing in Drosophila, which include cytoskeletal reorganization and cell shape changes (Galko and Krasnow, 2004; Bosch ; Pearson ; Campos ; Kwon ), transcriptional induction of many genes (Pearson ; Lesch ; Brock ; Stevens and Page-McCaw, 2012; Alvarez-Fernandez ), cytoprotection of neuronal tissues (Nam ), cell death-induced regeneration (Bergantinos ), inhibition of polyploidization (Losick ), cell fusion (Lee ), and distal accumulation of nonmuscle myosin II (Kwon ), to include mediating induction and possibly polarized accumulation of integrins. Requirement for the JNK pathway has also been extensively documented in developmental epithelial sheet movements in Drosophila (Xia and Karin, 2004), and the few investigations into JNK in vertebrate wound healing are consistent with an evolutionarily conserved RE role (Yates and Rayner, 2002; Li ; Schafer and Werner, 2007; Richardson ).

β-Integrin is required for the correct distally polarized accumulation of myosin II in cells surrounding the wound

JNK is required downstream of Rho small GTPases for distal localization of nonmuscle myosin II Zip protein in wounded epidermis (Baek ), so we analyzed the regulatory relationships between βPS and myosin II. Knockdown of βPS impaired the normal accumulation of a GFP-Zip/myosin II fusion protein after wounding; instead of clustering in arcs along the distal membranes as seen in WT, in βPS knockdown cells, GFP-Zip either displayed a polarized membrane accumulation that was somewhat randomized relative to the wound location or failed to localize at the membrane (Figure 5, A and B). For example, in the first row of cells, 62% of βPS-depleted cells displayed normal accumulation of myosin II, 32% displayed an incorrectly polarized membrane accumulation, and 6% displayed no membrane accumulation, contrasted with 92% of the WT cells around the wound that exhibited the typical distal recruitment of myosin (Figure 5D). A similar apparent randomized polarization of membrane recruitment of myosin II was observed after wounding in cells knocked down for talin gene expression (Figure 5C).

FIGURE 5:

βPS is required for rear localization of nonmuscle myosin II during wound closure. Localization of the nonmuscle myosin II heavy chain was examined using GFP-Zip in green at 7 h after wounding. (A, A’) A58-only. (B, B’) A58>βPS-i. (C, C’) A58>talin-i. (A’–C’) Enlarged view of the areas marked in white lines in A–C. Cell boundaries were visualized by FasIII immunostaining in red, and the cell nuclei were marked by DAPI staining in blue. Arrows represent the directionality of individual cells assessed by rear localization of GFP-Zip protein. White arrows indicate a normal direction, and yellow arrows indicate the wrong directions. Yellow dots indicate no response. Asterisks indicate the wound hole. Scale bar: 100 μm. (D) Quantification of the results in A–C, analyzing nonmuscle myosin localization in each of the first three rows of cells from the wound margin. See Materials and Methods for the statistical analysis.

We note that the defects in myosin II accumulation after wounding in βPS or talin knockdown cells are distinct from those in which JNK signaling is impaired, where myosin II remains cytoplasmic (Kwon ). This suggests that JNK signaling promotes the membrane accumulation of myosin II after wounding, possibly directing a polarized accumulation. β-integrin, on the other hand, is required to impart the correct directionality to this polarized recruitment of myosin II.

Functions of integrins in reepithelialization

In summary, we report that αPS2/βPS and αPS3/βPS dimers are required in the fly larval epidermis for RE following wounding, that talin appears alone among integrin adhesion components to be required for this process, that the induction of integrins following wounding requires JNK, and that integrin accumulates at the distal margin of cells surrounding the wound, where it is required for the similar localization of nonmuscle myosin.

Integrins contribute to wound RE in mammals and fish, but their specific molecular contributions have often been elusive (DiPersio ). Our findings offer some clues. While the obvious role for integrins would be adhering to the ECM to provide traction for the migrating epithelium, our findings suggest that this may not be the only, or even the primary, role. Traction for RE that is contributed by cells several rows from the wound is thought to be via “cryptic lamellipodia” (Farooqui and Fenteany, 2005; Trepat ; Matsubayashi ; Richardson ), which is hard to imagine being effected from the distal margin of surrounding cells. The posterior enrichment of integrin in cells surrounding the wound that we and others (Stevens and Page-McCaw, 2012) observed may reflect a role in monitoring the extracellular environment. The ECM undergoes marked changes in tension and composition following wounding, including through the initial rupture, and then through the deposition of replacement components, as well as the action of stiffening enzymes (Kobayashi ; Alatortsev ; Stramer ; Wong ; Barker and Engler, 2017). A role for integrins in sensing chemical and mechanical ECM features is well established in cell culture and beginning to be appreciated in wound repair (Kenny and Connelly, 2015; DiPersio ).

Possibly relevant to understanding the molecular function of integrins in RE is our finding that talin, but not other “core components,” are required for wound healing. Different adhesion components are required for different subfunctions of integrin complexes (Stutchbury ), with talin, in particular, implicated in sensing function of adhesions (Kumar ; Yao ; Rahikainen ). Our findings do not preclude an adhesion role for integrins in RE which seems inescapable; it could be that the structural components mediating the adhesive role are somewhat interchangeable (Bulgakova ).

The requirement we found for integrin in mediating the distal localization of myosin could simply reflect a localized recruitment of myosin to adhesions, as has been seen (Gupton and Waterman-Storer, 2006; Choi ; Vicente-Manzanares ). Alternatively, the integrin dependence of distal myosin accumulation might represent a primary role for integrin in polarizing cells relative to the wound; Rho-family GTPase-dependent front-rear polarization of cells by integrins has been well documented (Etienne-Manneville and Hall, 2001; Ridley ; Frank and Carter, 2004; Choma ) and could be consistent with a sensing role for integrins.

Clearly, there is much yet to learn about integrin contributions to wound healing. The small number of integrin dimers needed for RE in the fly, coupled with the array of genetic tools available, suggest that studies in Drosophila may point the way forward.

MATERIALS AND METHODS

Fly strains and genetics

The following lines were obtained from the Bloomington Stock Center: UAS-bsk-lacZ, and UAS-Calpain-B-i [BL25963 (denoted as #2 in superscript)]. The following lines were obtained from the Vienna Drosophila Resource Center (for the strains with numerical-only IDs) or the National Institute of Genetics in Japan (for the strains with IDs containing “R”): UAS-βPS-i [1560R-1 (#1), 29619 (#2), and 8070R-1 (#3)], UAS-talin-i [6831R-1 (#1) and 40399 (#2)], UAS-αPS1-i [1771R-1 (#1) and 44890 (#2)], UAS-αPS2-i [100770 (#1) and 44885 (#2)], UAS-αPS3-i [4891 (#1) and 100949 (#2)], UAS-Calpain-A-i [101294 (#1), 18152R-1 (#2), and 7563R-3 (#3)], UAS-Calpain-B-i [23037 (#1)], UAS-Calpain-C-i (107844), UAS-Fak-i (108608), UAS-Paxillin-i [31794R-C-1 (#1) and 107789 (#2)], UAS-parvin-i (105356), UAS-PINCH-i [7954R-1 (#1) and 100582 (#2)], UAS-P130CAS-i [1212R-3 (#1) and 41479 (#2)], UAS-tensin-i [9379R-3 (#1) and 22823 (#2)], UAS-Vinculin-i [3299R-2 (#1) and 34586 (#2)], and UAS-wech-i [1642R-2 (#1) and 106390 (#2)]. The following lines were obtained from private collections: A58-GAL4 (M. Galko, MD Anderson Cancer Center), UAS-GFP-Zip (D. Kiehart, Duke University), and βν2 (Y. Nakanishi, Kanazawa University, Japan).

Wounding, immunohistochemistry, and β-galactosidase staining

For epidermal wounding, third instar larvae were pinched at the dorsal side using a pair of forceps (Fine Science Tools, Cat. No. 11295-00). A wound with the typical size of 30–40 cells was generated in the abdominal segment A3-A4 (Kwon ). Larvae were allowed some time for recovery in food and were pinned down at their heads and tails on a silicon plate for dissection. Each larva was immersed in several drops of phosphate-buffered saline (PBS), and the outer integuments were torn open on the ventral side. Internal organs were removed, and the remaining epidermal fillet was spread wide using four more pins at the rectangular corners. In some cases, muscles were carefully removed by forceps for a better result in immunostaining. The tissue was fixed in 4% paraformaldehyde for 30 min, and the remaining fixative was washed out using PBS twice.

For immunohistochemistry, a fixed sample was first incubated in a primary antibody plus 2% normal goat serum in PBS supplemented with 0.2% Triton X-100 (PBST) overnight at 4°C. The sample was then incubated in a secondary antibody plus 0.2% PBST for 2 h at room temperature. For nuclear staining, 4′,6-diamidino-2-phenylindole (DAPI) was used at 0.01 μg/ml. Washed samples were mounted in 80% glycerol or Vectashield (Vector Laborartories, Cat. No. H-1200) and subjected to fluorescent microscopy (Olympus BX40 and Zeiss Imager.A2, with AxioCamMRc5) or confocal imaging (Zeiss LSM 510 META). Anti-FasIII (1:100, 7G10), anti-βPS (1:50, CF6G11), anti-αPS1 (1:50, DK.1A4), anti-αPS2 (1:10, CF.2C7), and anti-Talin (1:100, A22A or E16B) were obtained from the Developmental Studies Hybridoma Bank. Chicken anti–β-galactosidase antibody (1:1000, ab9361) was purchased from Abcam. Anti-αPS3 (1:100) and anti-Talin (1:100) were gifts from S. Hayashi (Riken Center for Developmental Biology, Japan) and N. Brown (University of Cambridge, UK), respectively. Cy3-conjugated goat anti-mouse immunoglobulin G (IgG) (1:200, 115-165-003) was purchased from the Jackson ImmunoResearch. Alexa 488-conjugated goat anti-mouse IgG (1:200; A-11001), goat anti-rat IgG (1:200; A-11006), and goat anti-rabbit IgG (1:200; A-11008) and Alexa 546-conjugated goat anti-chicken IgG (1:200; A-11040), goat anti-mouse IgG (1:200; A-11003), and goat anti-rabbit IgG (1:200; A-11035) were all purchased from Invitrogen (ThermoFisher Scientific). For analysis, some images are stitched using “pairwise stitching” plugin of Fiji (ImageJ2) (Preibisch ).

For β-galactosidase staining, dissected samples were fixed in 2% glutaraldehyde for 15 min at room temperature. The samples were washed with PBS two times and incubated in 10 mM NaPO4, 150 mM NaCl, 1 mM MgCl2, 3.1 mM K4[FeII(CN)6], 3.1 mM K3[FeIII(CN)6], 0.3% Triton X-100, and 0.2% X-Gal for 2 h at 37°C.

Quantitative real-time PCR

Total RNA was extracted using Trizol from 15 epidermal fillets in Figure 4D or from 20 to 30 whole larvae in Figure 2C in triplicates for each experimental condition. For the heat shock condition in Figure 2C, larvae were given a heat shock at 37°C for 30 min at 72 h after egg laying, incubated at 25°C for 24 h, and given another heat shock for 30 min, incubated at 25°C for 24 h, and then lysed. cDNAs were synthesized from 2 μg RNA using Moloney murine leukemia virus reverse transcriptase (M-MLV RT) (Promega, Cat. No. M1701). Quantitative real-time PCR was performed using SYBR Premix Ex Taq (Tli RNaseH Plus; Takara, Cat. No. RR420A) and each of the values was normalized by the corresponding value of rp49. Relative mRNA levels were calculated using the comparative cycle threshold (Ct) method. The following primers were used: 5′-CAGTCGGATCGAT­ATGCTAAGCTGT-3′ and 5′-TAACCGATGTTGGGCATCAGATACT-3′ for rp49; 5′-CCTGGCTAAAGTCGAAGATCTTG-3′ and 5′-GTTCCGACGCTCCACTTGAT-3′ for msn; 5′-TGGCGAGTGTCACTTGA­GTC-3′ and 5′-CAACCACATTGGATGAATCG-3′ for mys; 5′-AGGA­GACCGTTTTCCATGTG-3′ and 5′-AGTGCTTCTAGCCAGCCAAA-3′ for Fak; 5′-CATCCACATCGACATCCAAG-3′ and 5′-GCTGTTGGCCATTGGTATTT-3′ for tensin; 5′-TAACGGATCTGCGCATAAAT-3′ and 5′-GCCATCAGATCATCCAGTTC-3′ for Pax; 5′-ACCCGATTGGCTAAGGAGTT-3′ and 5′-ACGGTGGACAGGATTTTCAG-3′ for Vinc; 5′-CGAGCGTCTCCAGTGTTAAT-3′ and 5′-AGCTATTGTCTCGCAACAGG-3′ for wech; 5′-TTGAGGAGAGGGTGGTAACG-3′ and 5′-ACATCGCCTCCAATA ACCTG-3′ for pinch; 5′-CACCAAAAGATTCCCA AGTG-3′ and 5′-TCGTCGTACTCGGACGTTAT-3′ for parvin; 5′-AAATCGTCAGG GAAAGATCC-3′ and 5′-GCGATCGA­AGTGGTTAAGAA-3′ for CalpA; 5′-GCTGGCATCTTTCACTTCAA-3′ and 5′-TAAGATCCATGCAGCT TTGC-3′ for CalpB; 5′-GAAGCGTATCATGGA TTTGG-3′ and 5′-CA CCACGCGATAGAAGAGAT-3′ for CalpC; 5′-CTGA TCTTAGGCAGCAACCA-3′ and 5′-AAGTCGTAGGTCAGCC CACT-3′ for p130CAS.

Quantification of βPS protein induction

For quantification of anti-βPS immunofluorescence, Fiji (ImageJ2) was used. Immunohistochemical micrographs for βPS were converted to eight-bit images, and cells were sorted into different rows manually using Selection and XOR tools: cells (excluding cell debris) located immediately adjacent to the wound margin were designated to the first row category, and cells in the next row to the second row category, and so on. The value of fluorescence intensity per area (V1) was read for each of the rows using the Measure tool. The value for the noninduced state (V2) was obtained using the Threshold and Creative Selection tools by selecting cells in the neighboring segment adjacent to the wounded segment, or in some occasions, cells in the same wounded segment but located farthest away and were still polygonal in shape (Kwon ). Last, the value of the background (V0) was determined by reading fluorescence intensities of cell-free regions and was used for background subtraction. The fold increase of βPS protein in a specific row was then calculated as V = (V1-V0)/(V2-V0). Eight WT and six bsk animals were analyzed. For statistical analysis in Figure 4C, Mann–Whitney U test was used. Rank sum values of the first to fourth rows of WT samples were 38, 36, 39, and 48, and those of bsk samples were 67, 69, 66, and 57, respectively. The p values of two-sided test of each row were 0.0045, 0.0019, 0.0067, and 0.1213.

Quantification of the rear localization of βPS

A line was drawn along the frontal-rear axis of a cell in the wounded area, and the intensity value of anti-βPS immunofluorescence was obtained along the axis using ImageJ2. Owing to the large size and nonsymmetrical shape of a typical wound hole, the exact determination of the frontal-rear axis was sometimes arbitrary. We, however, tried to take the longest line that passes over the nucleus and lies within the degree of ±45° from the wound center. The rear localization index was calculated in the following way: the intensity value of the rear half of the cell minus that of the frontal half of the cell divided by the intensity value of the frontal half of the cell. Five WT and six bsk larvae were analyzed, and 19 cells from each row were analyzed on average (that was >70% of the total cells in a row and was essentially all the cells that were analyzable), which was presented in Figure 4I. For statistical analysis, both Student’s t test and Mann–Whitney U test were used. For Student’s t test, the rear localization index of each cell was regarded as an independent value. The p values of two-side test of each row were 4.3 × 10-8, 6.8 × 10-11, 2.7 × 10-9, and 0.214. For Mann–Whitney U test, the mean of the rear localization index of each row was calculated and regarded as a value for rank calculation. Rank sum values of the first to fourth rows of the WT samples were 17, 16, 16, and 22, and those of bsk were 49, 50, 50, and 44, respectively. The p values of the two-sided test of each row were 0.0176, 0.0106, 0.0106, and 0.1441. In Supplemental Figure S3E, five larvae of each of WT and βPS-i were dissected, and 54 cells of WT and 68 cells of βPS-i, collectively combined from the first and second rows were analyzed. The p values were 0.30 and 0.34 in Student’s t test and Mann–Whitney U test, respectively.

Quantification of GFP-Zip localization

The directionality and polarization of GFP-Zip was measured in the first three rows of cells from the wound margin. A symbol of compass was overlaid on each and every cell within the area using a circle with a "×” mark in the middle to determine the directionality of a cell. The compass was set toward the wound center or set perpendicular to the tangent line of the wound leading edge. If GFP-Zip was polarized in the correct direction, that was within the degree of 90°, then the cell was sorted as “normal.” If GFP-Zip was not polarized and remained around the nucleus as if the epidermis was not wounded, then these cells were sorted as “no response.” Cells that were ambiguous in directionality or polarization due to tissue damage, cell fusion, or other reasons were carefully excluded from the analysis. For statistical analysis in Figure 5D, Mann–Whitney U test was used. Six animals for each of WT and bsk were analyzed. Rank sum values of the rear localization index of the first to third rows of the WT samples were 22, 25, and 27, and those of bsk were 56, 53, and 51, respectively. The p values of two-sided test of each row were 0.0065, 0.025, and 0.0547.

Click here for additional data file.