Abstract
Cells are exposed to frequent mechanical and/or chemical stressors that can compromise the integrity of the plasma membrane and underlying cortical cytoskeleton. The molecular mechanisms driving the immediate repair response that is launched to restore the cell cortex and circumvent cell death are largely unknown. Using drug-inhibition studies and microarray analyses in the Drosophila model, we find that initiation of cell wound repair is dependent on translation, whereas transcription is required for subsequent steps in the process. We identified 253 genes whose expression is up-regulated (80) or down-regulated (173) in response to laser wounding. A subset of these genes were validated using RNAi knockdowns and found to exhibit aberrant actomyosin ring assembly and/or actin remodeling defects. Strikingly, we find that disruption of the canonical insulin signaling pathway leads to abnormal wound repair, and that it controls actin dynamics through the actin regulators Girdin and Chickadee (profilin). Our results provide new insight for understanding how cell wound repair proceeds in healthy individuals and those with diseases involving wound healing deficiencies.
Introduction
Numerous cell types in the body are subject to high levels of stress daily. These stresses—physiological and/or environmental—can cause ruptures in the plasma membrane and its underlying cytoskeleton, requiring a rapid repair program to avert further damage, prevent infection/death, and restore normal function (Andrews and Corrotte, 2018; Bement et al., 2007; Draeger et al., 2014; Gurtner et al., 2008; McNeil and Kirchhausen, 2005; McNeil and Terasaki, 2001; Moe et al., 2015; Nakamura et al., 2018; Sonnemann and Bement, 2011; Tang and Marshall, 2017; Velnar et al., 2009). Injuries to individual cells can also occur as a result of accidents/trauma, clinical interventions, and disease conditions, including diabetes, skin blistering disorders, and muscular dystrophies, as well as in response to pore forming toxins secreted by pathogenic bacteria (Clarke et al., 1995; Cooper and McNeil, 2015; Coulombe et al., 1991; Galan and Bliska, 1996; Petrof et al., 1993). Repair of these cell cortex lesions can be particularly troublesome when occurring alongside these fragile cell disease states or in a non-renewing and/or irreplaceable cell type. Thus, the importance of cell cortex continuity and delineating the molecular mechanisms regulating cell wound repair is of considerable clinical relevance, and important for advancing our knowledge of the many critical cell behaviors and fundamental regulations underpinning normal biological events that are co-opted for this repair process.
Aspects of single cell wound repair dynamics have been studied in Xenopus oocytes, sea urchin eggs, mammalian tissue culture cells, and the genetically-amenable Drosophila syncytial embryo (Abreu-Blanco et al., 2011a, b; Bement et al., 1999; Kono et al., 2012; Nakamura et al., 2018; Steinhardt et al., 1994; Terasaki et al., 1997; Yumura et al., 2014). This repair is generally conserved among these organisms and occurs in four main phases (Figure 1A). In the first phase, the wound expands as the cell recognizes the membrane breach and subsequently forms a membranous plug to neutralize any flux between the extracellular space and cytoplasm. Second, the cell constructs an actomyosin ring that underlies the plasma membrane at the wound edge. Third, the actomyosin ring translocates inward to draw the wound area closed. Mechanistic variations exist during this step wherein the actomyosin ring in some models translocates through actin treadmilling (actin simultaneously polymerizes at the inner edge and depolymerizes at the outer edge of the actin ring), while others use myosin II for sarcomere-like contraction (anti-parallel actin filaments are directed past each other in opposing directions) (Abreu-Blanco et al., 2011a, b, 2014; Benink and Bement, 2005; Burkel et al., 2012; Nakamura et al., 2018). The final step of wound repair occurs in which the plasma membrane and the underlying cortical cytoskeleton are remodeled returning them to their pre-wounded composition and organization. The mechanisms deployed by the cell for this remodeling have not yet been delineated.
Previous studies have shown that Ca2+ is required for the initiation of cell wound repair and serves as a messenger to trigger downstream processes such as transcription: release of internal and/or external Ca2+ stores activates a number of intracellular pathways resulting in an uptick of gene expression (Fein and Terasaki, 2005; Grembowicz et al., 1999; Joost et al., 2018; Togo, 2004). Studies carried out in rat embryos and cultured bovine aortic endothelial cells showed a rapid increase in expression of the Ca2+-responsive element containing c-Fos protein as a direct result of plasma membrane damage (Grembowicz et al., 1999; Martin and Nobes, 1992; Verrier et al., 1986). c-Fos, a component of the Activator protein 1 (AP-1), serves as a transcription factor responsible for expressing a number of cytokines and growth factors required to drive the appropriate cellular responses necessary for wound recovery (Shaulian and Karin, 2002; Yates and Rayner, 2002).
Interestingly, though the Drosophila syncytial embryo functions under the developmental control of maternally-contributed mRNAs and proteins with minimal levels of zygotic transcription, it is still able to immediately recognize and repair breaches to its cortex. Here we show that translation, rather than transcription, is required for the initial stages of repair in this cell wound repair model. Although transcription does not serve as a “start” signal, disrupting transcription leads to impaired repair in subsequent steps of the process. Using microarrays to assess gene expression changes post-wounding, we have identified 253 genes with a potential role in cell wound repair, indicated by changes in their expression—either up or down—in response to laser wounding. A subset of these genes were analyzed using RNAi knockdowns to visualize spatio-temporal patterns that verified their involvement. Strikingly, we find that the canonical insulin signaling pathway is required for proper cell wound repair where it controls actin dynamics through the actin regulators Girdin (Hook-like protein family) and Chickadee (profilin). Thus, our study provides insight into the roles of transcription, translation, and insulin signaing in cell wound repair and provides new avenues for understanding how wound healing proceeds in healthy individuals and disease sufferers with wound healing impairments.
Results
Assessment of transcriptional contribution to cell wound repair
To investigate the role of transcription in cell wound repair using the Drosophila syncytial (nuclear cycle 4-6) embryo model, we performed a microarray screen on full-length cDNA arrays to compare changes of gene expression between laser wounded and non-wounded states at two time points: immediately after wounding (0-5 minutes post-wounding (mpw)) and at the end of the repair process (~30 mpw) (Figure 1B). We found that at the immediate timepoint, wounded embryos exhibited no significant changes in their expression profiles when compared to their non-wounded counterparts (Figure 1C). Interestingly, the later timepoint showed significant changes of gene expression in both the up and down directions (Figure 1D). Using a false discovery rate of 0.05, we identified 253 genes with statistically significant changes: 80 that are up-regulated and 173 that are down-regulated (Table 1; Table S1). The robustness of the differences we observe is striking given that only ~5-10% of the cell surface is wounded and undergoing repair.
We next determined if these genes were being co-differentially expressed by shared activating or regulatory elements within a localized region of the genome in response to wounding. Genome mapping of the 253 genes show no obvious clustering upon visual inspection (Figure 1E). Concomitantly, we mapped the 253 differentially-expressed genes onto the 1169 unique topologically associated domains (TADs) previously characterized in flies(Li et al., 2017), and found no difference in overall differentially-expressed genes between TADs (p=0.22), as well as when comparing just the down-regulated genes (p=0.81) (Figure 1F). Interestingly, we detected a slight difference in differentially-expressed genes by TAD for up-regulated genes (p=0.01), however the majority of this signal appears to be driven by TADs that were missing genes due to their poorer coverage on our arrays. The results from this TAD analysis suggest that the 253 genes are being regulated independently and deliberately in response to wound repair. Intriguingly, the 80 upregulated genes were, on-average, larger than gene products previously recorded during this stage of development (Figure 1G) (Artieri and Fraser, 2014; Fukaya et al., 2017; Sandler et al., 2018; Vastenhouw et al., 2019), implicating the existence of a wound-repair specific program (see Discussion).
Transcription is not required to initiate cell wound repair
We expected that if transcription served as an initiator for wound repair as previously proposed, then inhibition of transcriptional activity would result in disrupted actin dynamics throughout the wound repair process. From our initial microarray results, we observe significant changes in gene expression exclusively at the later timepoint indicating that transcriptional activity is not necessary for initiating the wound repair process in this model. To further delineate the role of transcription in this process, we wounded nuclear cycle 4-6 Drosophila embryos that were injected with α-amanitin, a transcription inhibitor that targets RNA polymerase thereby halting transcritional activity, and that also expressed an actin-GFP reporter (sGMCA) to observe wound repair dynamics using time-lapse microscopy. In control embryos, where only buffer was injected, actin became enriched in two distinct locations: 1) adjacent to the wound edge, forming a robust actin ring, and 2) in a “halo” or diffuse accumulation along the outer periphery of the ring (Figure 2A-A’,D-F; Video 1). Consistent with our microarray results, α-amanitin injected embryos initially showed actin dynamics similar to those observed in control embryos, however they exhibited disruptions to the repair process during the subsequent actin remodeling phases (Figure 2B-B’, D-F; Video 1). To ensure that we were getting efficient transcriptional knockdown, we verified the efficacy and duration of the α-amanitin treatment using the MS2-MCP system, a visual reporter of active transcription (see Methods) (Figure 2G-I’) (Forrest and Gavis, 2003; Weil et al., 2006). In Drosophila syncytial embryos, GFP appears as puncta within the nuclei of control embryos indicative of active transcription, whereas these GFP puncta are absent in α-amanitin injected embryos indicating that α-amanitin is effective in inhibiting transcription even beyond our initial wounding window (Figure 2H-H’; Video1). Thus, our results indicate that a transcriptional response is dispensable for the initiation of cell wound repair in the Drosophila model, but becomes important subsequently, potentially for replenishing and/or maintaining various factors necessary for establishing the wound repair response.
The initial steps of cell wound repair are translation dependent
Drosophila early embryonic development is mostly driven by maternally deposited mRNA and protein until the maternal-to-zygotic genome transition (MZT) at nuclear cyle 14 (cf. (Vastenhouw et al., 2019)). To explore what role translation might be playing in driving the wound repair process, embryos expressing a fluorescent actin reporter (sGMCA) were injected with the translation inhibitor puromycin prior to laser wound induction. After the initial formation of an actin ring (~2 mins), the ring began to disassemble while actin simultaneously became enriched inside the wound area (Figure 2C-C’). Quantitative measurements show a prolonged wound healing process compared to controls (Figure 2D), with significantly less wound expansion and slower wound closure (Figure 2E-F). Taken together, our results suggest that the Drosophila embryo requires active translation to initiate wound repair, as well as to regulate actin dynamics throughout the repair process.
Knockdown of differentially expressed genes results in wound over-expansion and abnormal actin dynamics upon wounding
We next examined the effects of removing the differentially-expressed genes on cell wound repair. We generated knockdown embryos for 15 of the top 16 up-regulated genes and the 16 top down-regulated genes (based on their fold-change; Table 1) by expressing RNAi constructs in the female germline using the GAL4-UAS system (Brand and Perrimon, 1993; Rorth, 1998), then observing actin dynamics using a fluorescent actin reporter (sGMCA).
Up-regulated Genes
In all 15 cases of RNAi knockdown for up-regulated genes, wounded embryos exhibited abnormal actin dynamics that fell within three major visually distinct, but non-mutually exclusive, classes: 1) premature actin ring/halo disassembly, 2) failure of actin ring/halo dissassembly, and 3) abnormal actin ring/halo disassembly with concomitant accumulation of actin within the wound (Figure 3; Video 2; Figure S1). Among Class 1 members, exemplified by their incomplete formation and premature dissassembly of the actomyosin ring causing rifts at the initial injury site that remained open for the entire time of repair, are Imaginal morphogenesis protein-Late 2 (ImpL2) and Epidermial growth factor receptor (EGFR) (Figure 3B-C’,J-K; Video 2). ImpL2 has been proposed to work antagonisitically to the insulin/insulin-like (IIS) signaling pathway by interacting with receptor/ligand interactions to inhibit downstream signal transduction (Amoyel et al., 2016; Figueroa-Clarevega and Bilder, 2015; Kwon et al., 2015). EGFR encodes a receptor tyrosine kinase that works upstream of the c-jun N-terminal kinase (JNK) and decapentaplegic (dpp) pathways. Loss of EGFR results in down-regulation of JNK activity leading to the impairment of dorsal closure, a process sharing many features with epithelial (multicellular) wound repair (Kushnir et al., 2017).
Class 2 members, exemplified by Gp150, Inx3, and Thor, are unable to resolve actin structures or properly remodel cortical actin after wound closure (Figure 3D-F’,L-N; Video 2). Gp150 encodes a transmembrane glycoprotein that regulates Notch signaling during normal eye development in Drosophila (Li et al., 2003), whereas Inx3 encodes a gap junction protein involved in morphogenesis and nervous system development (Giuliani et al., 2013; Lautemann and Bohrmann, 2016). Thor encodes a translation inhibitor functioning downstream of insulin signaling that is sensitive to reactive oxygen species (Toshniwal et al., 2019). Interestingly, like ImpL2, Thor is a IIS pathway constituent and Gp150 has also been shown to physically interact with components of this pathway (Pten and S6k) (Vinayagam et al., 2016).
Class 3 genes, exemplified by Jitterbug (Jbug) and Nullo, are characteristically defined by the pronounced formation of actin inside the wound area (Figure 3G-H’,O-P; Video 3). Jbug is a filamin-type protein that serves as an F-actin crosslinker providing stability to the cytoskeleton, a system that has been proposed to utilize mechanical cues such as tension to modulate cellular processes (Martino et al., 2018; Nakamura et al., 2011). Nullo has been shown to establish cortical compartments during cellularization of the Drosophila embryo, suggesting an important role regulating actin stability at the cortex (Sokac and Wieschaus, 2008a, b).
Down-regulated Genes
Interestingly, in all 16 cases of RNAi knockdown for the down-regulated genes examined, wounded embryos exhibited abnormal actin dynamics that fell within the same three major distinct, but non-mutually exclusive, categories as described above for the up-regulated genes. A number of the genes that were downregulated have an unknown molecular function and/or associated biological processes (Table 1; Figure 4; Video 3; Figures S1q, S2). Of these unknown genes, CG31075, CG4960, and CG1598 exhibited characteristics of Class 1, Class 2, and Class 3 phenotypes, respectively. CG31075 underwent a mild expansion followed by a severly slowed closure rate and incomplete wound closure (Figure 4A-A’,H; Video 3), CG4960 exhibited a slight delay in wound repair dynamics but retained noticably enriched actin structures after closure (Figure 4E-E’,L; Video 3), and CG1598 developed a visually distinct, but transient, enrichment of actin inside the wound area prior to closure (Figure 4G-G’,N; Video 3). Of genes with known motifs/functions, Glutatione S transferases D2 (GstD2) and D1 (GstD1) mutants showed nearly identical phenotypes exhibiting a short-lived accumulation of actin inside the wound area and prolonged closure dynamics during the initial steps of repair (Figure 4B-B’,F-F’,I,M; Video 3), but in later steps GstD1 was unable to completely close (Figure 4B-B’,I; Video 3). Wound repair begins normally in exu knockdowns, however the leading edge and surrounding actin structures soon become static resulting in an open wound area and prolonged actin accumulation (Figure 4D-D’,K; Video 3). In addition to the phenotypes described above, many of these knockdowns exhibit wound over-expansion (Impl2, EGFR, Gp150, Thor, Dpn, l(3)neo38, GstD1, GstD2, CG1598, CG3652, Pex11, Cyp6a9, ReepB, P32, and RpL23) (Figures 3, 4; Videos 3, 4; Figures S1, S2). Thus, in all 31 cases of up- or down-regulated genes examined, knockdown using RNAi transgenes resulted in abnormal cell wound repair. Despite the molecular functions of many of these genes being unknown, they have been implicated in various cellular processes, but most notably insulin signaling.
Activation of insulin/insulin-like (IIS) constituents during normal wound repair
The fact that ImpL2 and Thor, two of the most upregulated genes in our analyses, are constituents of the insulin/insulin-like growth factor signaling (IIS) pathway in Drosophila (Figure 5A) was unexpected. Deficiencies in insulin signaling have been implicated in multicellular (tissue) repair, where it is thought to impede growth factor production, angiogenic response, and epidermal barrier function (Brem and Tomic-Canic, 2007; Galiano et al., 2004; Kakanj et al., 2016; Manzano-Nunez et al., 2019), functions that might not normally be expected to govern regulation within individual cells.
To determine if the canonical IIS pathway was involved in individual cell wound repair, we first examined the recruitment pattern of a PIP3 (phosphatidylinositol (3,4,5)-triphosphate) GFP-reporter construct used as a reporter of insulin signaling activity (Britton et al., 2002) co-expressed with a Cherry fluorescently-tagged actin reporter (sChMCA) in a wildtype and chico RNAi knockdown background (Figure 5B-E). PIP3 is a phospholipid that composes a subset of specialized plasma membrane with various trafficking and signaling related functions (Rameh and Cantley, 1999). PIP3 is recruited to same region as the actomyosin ring in wildtype embryos (Figure 5B-B”,D), confirming the requirement for autocrine insulin pathway signaling. Importantly, this recruitment is dependent on the upstream activation of the insulin receptor (InR), as PIP3 recruitment is disrupted in a chico RNAi background (Figure 5C-C”,E).
We next examined the wound repair phenotypes in knockdown backgrounds for components spanning the IIS pathway by expressing RNAi constructs for pathway components in the female germline using the GAL4-UAS system (Brand and Perrimon, 1993; Rorth, 1998), then observing actin dynamics using a fluorescent actin reporter (sGMCA). The one ligand and six of the major IIS pathway components tested — Ilp4 (Insulin-like peptide), InR (Insulin receptor), Chico (IRS homolog), Pi3K21B (Phosphoinositide3-Kinase), Akt1 (Kinase), FoxO (transcription factor), and Reptor (transcription factor) — exhibited abherrant wound repair with notably similar phenotypes (Figures 5A, 6; Video 4; Figure S1Q). With the exception of InR, mutants for IIS pathway components exhibited wound overexpansion immediately after laser ablation that was visible as the outward retraction of the wound edge (Figure 6, yellow arrows; Video 4). Following this overexpansion, actin structures became transiently enriched inside the wound area, but dissassembled prior to complete wound closure (Figure 6, yellow arrowheads; Video 4). Lastly, progression of wound closure was signficantly delayed and/or incomplete, leaving openings around the actin ring as it translocated (Figure 6, red arrowheads; Video 4). Collectively, these results show that key components of insulin signaling are not only called to a wound, but have detrimental effects on actin dynamics upon knockdown, suggesting that there exists a tight association between the factors that regulate both insulin signaling and cell wound repair in the Drosophila model.
The IIS pathway effectors Profilin (Chickadee) and Girdin are required for cell wound repair
The IIS pathway has recently been shown to control actin dynamics independently of its role in growth control (Ghiglione et al., 2018). In particular, the IIS pathway has been found to activate the expression of the Drosophila profilin homolog, Chickadee, as well as the Akt substrate Girdin (GIRDers of actIN; also known as GIV) (Ghiglione et al., 2018; Hartung et al., 2013; Lopez-Sanchez et al., 2015). To determine if these actin regulators function as IIS pathway effectors during cell wound repair, we stained wounded embryos that expressed a GFP-tagged actin reporter (sGMCA) in a wildtype or chico RNAi knockdown background with antibodies to Profilin and Girdin (Figure 7A-D). Both proteins are recruited to wounds. Girdin exhibits a punctate recuitment at wounds with the highest accumulation overlapping the membrane plug inside the actin ring and with lower level diffuse accumulation overlapping the actin ring and the innermost part of the actin halo (Figure 7A-B). Profilin recruitment is internal to the actin ring and appears to be excluded from the actin ring region (Figure 7A-B). Importantly, the accumulation of both Girdin and Profilin at wounds requires a functioning IIS pathway as these accumulations are lost in a chico RNAi background (Figure 7C-D).
We next examined the effects of removing Girdin and Profilin on cell wound repair. Similar to knockdown of IIS pathway components described above, Girdin RNAi knockdown embryos exhibited aberrant wound repair including wound overexpansion, enrichment of actin structures inside the wound area, and signficantly delayed wound closure (Figure 7E-E’,G; Video 4; Figure S1Q). Unfortunately, Profilin RNAi knockdown females do not produce eggs. We therefore used the wimp mutation (Parkhurst and Ish-Horowicz, 1991; Verboon et al., 2018) to generate reduced Profilin expression in both the germline and soma (wimp reduces maternal gene expression such that, when in trans to the chickadee221 allele, it effectively generates a strong chickadee hypomorph, referred to as reduced Profilin). Similar to knockdown of Girdin and IIS pathway components, reduced Profilin embryos exhibited wound overexpansion, enrichment of actin structures inside the wound area, and signficantly delayed wound closure (Figure 7F-F’,H; Video 4; Figure S1Q). Thus, our results indicate that Girdin and Profilin are actin regulatory downstream effectors of the IIS pathway in cell wound repair.
Discussion
Our study shows that cellular wound repair is not dependent on transcriptional activity to initiate wound repair programs, that dormant transcription pathways are activated in response to wounds, and that the insulin signaling pathway is an essential component of the repair process. A calcium influx-triggered transcriptional response is thought to be important to lead off the cell wound repair process, eliciting a downstream wound repair program. This proposed mechanism was at odds with the Drosophila syncytial embryo cell wound model that faithfully recapitulates the majority of features associated with other single cell wound repair models (Xenopus oocytes, tissue culture cells, sea urchin eggs) (Abreu-Blanco et al., 2011a, 2014; Bi et al., 1995; Miyake and McNeil, 1995; Nakamura et al., 2018; Nakamura et al., 2017; Sonnemann and Bement, 2011; Steinhardt et al., 1994; Verboon and Parkhurst, 2015), yet represents a special system running mostly off of maternally contributed products, highlighted by rapid cell cycles (~10 minutes/cycle) and minimal zygotic transcription (Foe and Alberts, 1983; Sandler et al., 2018; Vastenhouw et al., 2019).
Consistent with the closed nature of the Drosophila syncytial embryo cell wound model, we find no altered gene expression immediately upon wounding as assayed by microarray analysis of laser wounded versus unwounded embryos or following injection of the α-amanitin transcriptional inhibitor. We do detect alterations in gene expression at subsequent stages in the repair process: we identified 253 genes (out of ~8000 genes assayed) whose expression is significantly up (80 genes) or significantly down (173 genes) following laser wounding.
Polymerase rates in the early Drosophila embryo were reported to be 1.1-1.5 kb/min, leading to the suggestion that any genes transcribed in the early Drosophila embryo prior to the mid-blastula transition must be small with minimal introns due to the rapid (~10 min) cell cycles and limited transcription time (Ardehali et al., 2009; Garcia et al., 2013; O’Farrell, 1992; Sandler et al., 2018; Shermoen and O’Farrell, 1991; Thummel et al., 1990; Vastenhouw et al., 2019; Yao et al., 2007). Recent studies have revised this rate to 2.4-3.0 kb/min, lowering the size constraints on the zygotic genes that can be successfully transcribed prior to the mid-blastula transition(Fukaya et al., 2017). Therefore, genes up to ~20-25 kb could theoretically be transcribed during the early and rapid Drosophila embryo cell cycles. In this case however, the number of mRNA molecules would be likely limited by the overall nuclear cycle due to the lower number of nuclei present and thus copies of DNA.
We find that the average size of transcripts in syncytial Drosophila embryos is 2.5 kb, similar to the previously reported size of 2.2 kb (compared to the overall average length of coding genes in Drosophila of 6.1 kb) (Artieri and Fraser, 2014; Hoskins et al., 2011). Genes whose expression goes down during wound repair are, on average, 1.9 kb. It intriguing that these actively down-regulated genes negatively impact the wound repair process when knocked-down. These genes likely represent RNAs stored in the embryo that are used up during the repair process and not replaced. Alternatively, it is possible that wound repair itself may slightly delay development leading to a subset of zygotically expressed genes whose expression is lagging behind in wounded versus unwounded embryos such that this delayed developmental upregulation is read out as a down-regulation of genes.
Surprisingly, we find that genes whose expression is higher after wounding are much larger on average (3.7 kb) than the average sized transcript at that stage (2.5 kb). These genes likely encode cellular components that were expended during the repair process and are being replenished for normal developmental events to proceed, or that are activated specifically for the repair process. This subset of “up-regulated” genes includes genes that are not usually expressed in the early embryo (e.g., CG43693). Thus, our results suggest that, when wounded, the embryo may be able to activate a transcriptional program that is usually dormant during these stages.
Interestingly, 2 of the top 3 genes whose expression is significantly higher following wounding—Impl2 and Thor—are components of the Insulin signaling pathway. While it has been shown that defective insulin signaling impairs epithelial (multicellular) wound repair (Brem and Tomic-Canic, 2007; Galiano et al., 2004; Kakanj et al., 2016; Manzano-Nunez et al., 2019), this result was unexpected for wound repair in single cells. Using a combination of RNAi knockdowns and GFP reporters, we have shown that all major components of the IIS pathway are involved in cellular wound repair and upon knockdown display similar phenotypes, suggesting that in this context the canonical IIS pathway activation occurs in an autocrine-like manner. Previous studies have highlighted the necessity of calcium influx to facilitate vesicle exocytosis and subsequent fusion of the plasma membrane during wound repair (Bi et al., 1995; Miyake and McNeil, 1995; Steinhardt et al., 1994; Terasaki et al., 1997). Similarly, this influx has also been shown to modulate insulin secretion in β-islet cells via the opening of L-type channels by establishing calcium microdomains along the cortex (Lee et al., 2018; Rutter et al., 2006). Insulin/insulin-like peptides are secreted into the extracellular space where they bind to InR thereby activating the heavily conserved IIS pathway that is known to regulate a number of downstream processes that range from transcription via phosphorylation events on the FOXO family of transcription factors to translation via the regulation of the 4E-binding protein, Thor (Brunet et al., 1999; Haeusler et al., 2018; Puig and Tjian, 2005; Teleman et al., 2005). Recently emerging evidence has also shown that the activated IIS pathway can control actin dynamics through activation of actin regulators including Chickadee (profilin) and Girdin (Ghiglione et al., 2018; Hartung et al., 2013; Houssin et al., 2015; Lopez-Sanchez et al., 2015; Wang et al., 2018a).
Observation of actin dynamics in mutants for a number of the IIS pathway components show common phenotypes of impaired cytoskeleton dynamics, most notably an immediate over-expansion of the wound leading edge and a transient actin structure forming inside the wound area suggesting that normal wound repair processes are heavily reliant on a functioning IIS pathway. We propose that the initial inrush of calcium generates microdomains that trigger the secretion of the Drosophila insulin-like peptide 4 (Ilp4) into the perivitelline space where it recognizes and binds to the extracellular face of the Insulin receptor (Figure 7I, steps 1-4). Subsequently, the InR is activated and initiates a signaling cascade that regulates a number of downstream processes, including cytoskeletal dynamics. Chickadee/profilin binds to actin and affects the formation/remodeling of actin-rich structures (Ghiglione et al., 2018). Girdin also binds to actin, as well as the catenin-cadherin complex and the Exo-70 subunit of the exocyst complex, where it has been proposed to coordinate cytoskeleton organization, cell adhesion, membrane trafficking events, and serves as an indicator for poor prognosis with invasive breast cancers (Choi et al., 2017; Hartung et al., 2013; Houssin et al., 2015; Lopez-Sanchez et al., 2015; Wang et al., 2018b). Interestingly, girdin and Profilin knockdown embryos exhibit wound repair phenotypes consistent with defects in actin structure assembly/remodeling, actomyosin ring attachment to the overlying plasma membrane, and membrane trafficking. In addition to the genes involved in the IIS pathway, our microarray analyses identified numerous other genes that show phenotypes associated with actin dynamics regulation. For example, Nullo is a known regulator of actin-myosin stability and has been proposed to affect actin-actin and actin-membrane interactions at the cortex, suggesting a role in cortical remodeling during actomyosin ring contraction (Sokac and Wieschaus, 2008a, b).
In summary, our understanding of the mechanisms that trigger cell wound repair remain incomplete, but here we show functional translation is essential for initiating a normal and processive wound repair process, suggesting that the first responders are likely mRNA and protein already present in the cell. While transcription is not immediately necessary in the Drosophila cell wound model, it is needed for the repair process. The requirement for insulin signaling in the single cell wound repair context highlights the conservation of repair mechanisms employed. Given its prominence in the single cell, as well as multicellular (tissue), repair pathways, it is not surprising that impaired insulin signaling leads to major wound repair defects in diseases such as diabetes where chronic wounds are symptomatically observed. As many of the top up- and down-regulated genes we identified are evolutionarily conserved genes, but of currently unknown function, the challenge for the future is to determine their roles in normal cellular maintenance and/or development, in addition to their effects in a cell wound repair context, thereby allowing the establishment of a network of cellular processes involved to better aid in treatments of disease involving wound healing impairments, or in disciplines such as regenerative medicine.
Materials and Methods
Fly stocks and genetics
Flies were cultured and crossed at 25°C on yeast-cornmeal-molasses-malt extract medium. The flies used in this study are listed in Table S2A. RNAi lines were driven using the GAL4-UAS system using the maternally expressed driver, Pmatalpha-GAL-VP16V37. All genetic fly crosses were performed at least twice. All RNAi experiments were performed at least twice from independent genetic crosses and ≥10 embryos were examined unless otherwise noted.
An actin reporter, sGMCA (spaghetti squash driven, moesin-alpha-helical-coiled and actin binding site bound to GFP reporter) (Kiehart et al., 2000) or the Cherry fluorescent equivalent, sChMCA (Abreu-Blanco et al., 2011a), was used to follow wound repair dynamics of the cortical cytoskeleton.
wimp (RpL140wimp) reduces maternal gene expression of a specific subset of genes in the early Drosophila embryo (Liu et al., 2009; Parkhurst and Ish-Horowicz, 1991; Verboon et al., 2018). Reduced chickadee embryos were obtained from trans-heterozygous females generated by crossing chickadee221 to RpL140wimp.
We attempted InR knockdown in three ways: 1) expressing one shRNA (GL00139) using one maternal-GAL4 driver (BDSC #7063), 2) expressing two shRNAs (HMS03166 and GL00139) using one maternal-GAL4 driver (BDSC #70637063), and 3) expressing one shRNA (GL00139) using one maternal-GAL4 (BDSC #7063) in an InR05545 heterozygous mutant backgrounds. We achieved only 50% knockdown with approach (1), and no eggs were produced by approach (3). We achieved 87% knockdown with approach (2) and this condition was used for the phenotypic analyses included here.
For the MS2-MCP system (Forrest and Gavis, 2003; Weil et al., 2006), female virgins maternally expressing MCP-GFP and Histone-RFP were crossed with males expressing 24xMS2 stem loops and lacZ driven by hunchback P2 enhancer and promoter. F1 embryos (MCP-GFP, Histone-RFP/+; 24xMS2-lacZ/+) at NC9-10 stages were used for imaging where the 24xMS2-lacZ mRNA is contributed zygotically.
Localization patterns and mutant analyses were performed at least twice from independent genetic crosses and ≥10 embryos were examined unless otherwise noted. Images representing the average phenotype were selected for figures.
Quantification of mRNA levels in RNAi mutants
To harvest total RNA, 100-150 embryos were collected after a 30 min incubation at 25°C, treated with TRIzol (Invitrogen/Thermo Fisher Scientific) and then with DNase I (Sigma). 1 μg of total RNA and oligo (dT) primers were reverse transcribed using the iScript gDNA Clear cDNA Synthesis Kit (Bio-Rad). RT-PCR was performed using the iTaq Universal SYBR Green Supermix (Bio-Rad) and primers obtained from the Fly Primer Bank listed on Table S2B. We were unable to identify primer sets that would work for qPCR for Geko, Ama, l(3)neo38, danr, and CG4960.
Each gene in question was derived from two individual parent sets and run in two technical replicates on the CFX96TM Real Time PCR Detection System (Bio-Rad) for a total of four samples per gene. RpL32 (RP-49) or GAPDH were used as reference genes and the knockdown efficency (%) was obtained using the ΔΔCq calculation method compared to the control (GAL4 only).
Embryo handling and preparation
NC4-6 embryos were collected for 30 min at 25°C, then harvested at room temperature (22°C). Collected embryos were dechorionated by hand, desiccated for 5 min, mounted onto No. 1.5 coverslips coated with glue, and covered with Series 700 halocarbon oil (Halocarbon Products Corp.) as previously described (Abreu-Blanco et al., 2011a).
Drug Injections
Pharmacological inhibitors were injected into NC4-6 staged Drosophila embryos, incubated at room temperature (22°C) for 5 min, and then subjected to laser wounding. The following inhibitors were used: α-amanitin (1 mg/ml; Sigma-Aldrich); puromycin (10 mg/ml; Sigma-Aldrich). The inhibitors were prepared in injection buffer (5 mM KCl, 0.1 mM NaP pH6.8). Injection buffer alone was used as the control.
Laser Wounding
All wounds were generated with a pulsed nitrogen N2 micropoint laser (Andor Technology Ltd.) set to 435nm and focused at the lateral surface of the embryo. A circular targeted region of 16×15.5 μm was selected along the lateral midsection of the embryo, and ablation was controlled by MetaMorph software (Molecular Devices). Average ablation time was less than 3 seconds and time-lapse image acquisition was initiated immediately after ablation. Upon ablation, a grid-like pattern is sometimes observed (fluorescent dots within the wound area), as a result of the laser scoring the vitelline membrane that envelops the embryo. This vitelline membrane scoring has no effect on wound repair dynamics.
Immunostaining of wounded embryos
Embryos (1-2 min post-wounding) were fixed in formaldehyde saturated heptane for 40 min. The vitelline membrane was removed by hand and the embryos were then washed 3 times with PAT [1x PBS, 0.1% Tween-20, 1% bovine serum albumin (BSA), 0.05% azide], then blocked in PAT for 2h at 4°C. Embryos were incubated with mouse anti-chickadee antibody (chi 1J; 1:10; Developmental Studies Hybridoma Bank) and guinea pig anti-Girdin antibody (1:500; provided by Patrick Laprise) (Houssin et al., 2015) and for 24h at 4°C. Embryos were then washed 3 times with XNS (1x PBS, 0.1% Tween-20, 0.1% BSA, 4% normal goat serum) for 40 min each. Embryos were incubated with Alexa Fluor 568- and Alexa Fluor 633-conjugated secondary antibodies (1:1000; Invitrogen) overnight at 4°C. Embryos were washed with PTW (1x PBS, 0.1% Tween-20), incubated with Alexa Fluor 488-conjugated Phalloidin at 0.005 units/μl (Molecular Probes/Invitrogen, Rockford, IL) at room temperature for 1 h, washed with PTW, and then imaged.
Live Image Acquisition
All imaging was done using a Revolution WD systems (Andor Technology Ltd.) mounted on a Leica DMi8 (Leica Microsystems Inc.) with a 63x/1.4 NA objective lens under the control of MetaMorph software (Molecular devices). Images were acquired using a 488 nm, 561 nm, and 633 nm Lasers and Andor iXon Ultra 897 EMCCD camera (Andor Technology Ltd.). All time-lapse images were acquired with 17-20 μm stacks/0.25 μm steps. For single color, images were acquired every 30 sec for 15 min and then every 60 sec for 25 min. For dual green and red colors, images were acquired every 45 sec for 30-40 min.
Image processing and analysis
Image processing was performed using FIJI software (Schindelin et al., 2012). Kymographs were generated using the crop feature to select ROIs of 5.3 x 94.9 μm. To generate fluorescent profile plots by R, 10 pixel sections across the wound from a single embryo were generated using Fiji as we described previously (Nakamura et al., 2017). The lines represent the averaged fluorescent intensity and gray area is the 95% confidence interval. Line profiles from the left to right correspond to the top to bottom of the images unless otherwise noted. Wound area was manually measured using Fiji and the values were imported into Prism 8.2.1 (GraphPad Software Inc.) to construct corresponding graphs. Figures were assembled in Canvas Draw 6 for Mac (Canvas GFX, Inc.).
Microarray Preparation and Processing
Expression profiles were obtained using the FHCRC Fly 12k spotted array (GEO platform, GPL 1908). Embryos, prepared for wounding, were either wounded 8 times or left unwounded, then collected for total RNA extraction. Sample labeling and hybridization protocols were performed as described by Fazzio et al (Fazzio et al., 2001). Specifically, cDNA targets were generated from total RNA using a standard amino-allyl labelling protocol where 30 ug of total RNA from each wounding condition (wounded vs non-wounded) were coupled to either Cy3 or Cy5 fluorophores. Targets were co-hybridized to microarrays for 16 hours at 63C and sequentially washed at room temperature (22C) in: 1 x SSC and 0.03% SDS for 2 mins, 1 x SSC for 2 mins, 0.2 x SSC with agitation for 20 mins, and 0.05 x SSC with agitation for 10 mins. Arrays were immediately centrifuged until dry and scanned using a GenePix 4000B scanner (Molecular Devices, Sunnyvale, CA). Image analysis was performed using GenePix Pro 6.0.
Microarray Analysis
Wounded and non-wounded samples were independently replicated 4 times each at the 0 min and 30 min time point. For each array, spot intensity signals were filtered and removed if the values did not exceed 3 standard deviations above the background signal, if the background subtracted signal was <100 in both channels, or if a spot was flagged as questionable by the GenePix Pro Software. Spot-levels ratios were log2 transformed and loess normalized using the Bioconductor package limma(Smyth et al., 2005). Differential gene expression between wounded and non-wounded states was determined using the Bioconductor package limma, and a false discovery rate (FDR) method was used to correct for multiple testing (Reiner et al., 2003). Significant differential gene expression was defined as |log2 (ratio)| ≥ 0.585 (± 1.5-fold) with FDR set to 5%. Gene ontology enrichment scores were determined using DAVID with significance based on EASE scores corrected for multiple testing (Huang da et al., 2009a, b). The microarray datasets are available at GEO (NCBI Gene Expression Omnibus) under accession numbers: GSE39481, GSE39482, and GSE39483.
TAD analysis
Genes were mapped to previously described TADs (Li et al., 2017). A TAD by up/down regulated gene versus unaffected gene expressed on the microarray contigency table was assembled. Fisher’s exact test of independence was used to test the null hypothesis that porportion of differentially expressed genes was different per TAD.
Gene Size Analysis
Gene size was determined as the size of the largest expressed transcript per gene (dm6 build) expressed on the arrays. The median plus 95% CI was determined using the bootstrap procedure and 1000 iterations.
Author Contributions
All authors contributed to the design of the experiments, performed experiments, and analyzed data. JJD and JMV performed the bioinformatic analyses. MN, ANMD, JMV, and SMP wrote the manuscript with input from all authors.
Competing Interests
The authors declare no competing or financial interests.
Supplementary Video Legends
Video 1 | Translation and transcription are needed for different aspects of cell wound repair. a-c, Time-lapse confocal xy images from Drosophila NC4-6 staged embryos expressing an actin marker (sGMCA): control (buffer only) (a), alpha-amanitin injected (b), and puromycin injected (c). Time post-wounding is indicated. UW: unwounded.
Video 2 | Knockdown of up-regulated genes results in wound over-expansion and abnormal actin dynamics. a-h, Time-lapse confocal xy images from Drosophila NC4-6 staged embryos expressing an actin marker (sGMCA): control (w1118/+; sGMCA, 7063/+) (a), ImpL2RNAi/+; sGMCA, 7063/+ (b), EGFRRNAi/+; sGMCA, 7063/+ (c), Gp150RNAi/sGMCA, 7063 (d), Inx3RNAi/sGMCA, 7063 (e), ThorRNAi/sGMCA, 7063 (f), JbugRNAi/sGMCA, 7063 (g), NulloRNAi/sGMCA, 7063 (h).). Time post-wounding is indicated. UW: unwounded.
Video 3 | Knockdown of down-regulated genes results in wound over-expansion and abnormal actin dynamics. a-g, Time-lapse confocal xy images from Drosophila NC4-6 staged embryos expressing an actin marker (sGMCA): CG31075RNAi/+; sGMCA, 7063/+ (a), GstD1RNAi/+; sGMCA, 7063/+ (b), dhdRNAi/+; sGMCA, 7063/+ (c), ExuRNAi/+; sGMCA, 7063/+ (d), CG4960RNAi/+; sGMCA, 7063/+ (e), GstD2RNAi/+; sGMCA, 7063/+ (f), CG1598RNAi/+; sGMCA, 7063/+ (g). Time post-wounding is indicated. UW: unwounded.
Video 4 | Actin dynamics of IIS pathway mutants. a-i, Time-lapse confocal xy images from Drosophila NC4-6 staged embryos expressing an actin marker (sGMCA): insulin-like peptide 4RNAi (Ilp4RNAi; a), InRRNAi(1)/+; InRRNAi(2)/sGMCA, 7063 (b), ChicoRNAi/sGMCA, 7063 (c), Pi3K21BRNAi/sGMCA, 7063 (d), Akt1RNAi/sGMCA, 7063 (e), FoxORNAi/sGMCA, 7063 (f), ReptorRNAi/sGMCA, 7063 (g), GirdinRNAi/+; sGMCA, 7063/+ (h), and sGMCA; chickadee221/+ sGMCA, wimp/+ (reduced chickadee) (f). Time post-wounding is indicated. UW: unwounded.
Acknowledgements
We thank Ryan Basom, Patrick Laprise, Scott Somers, the Bloomington Stock Center, the Kyoto Stock Center, the Harvard Transgenic RNAi Project, the Vienna Drosophila RNAi Center, the Drosophila Genomics Resource Center, and the Developmental Studies Hybridoma Bank for advice, antibodies, DNAs, flies, and other reagents used in this study. This research was supported by NIH GM111635 (to SMP) and NCI Cancer Center Support Grant P30 CA015704 (for Shared Resources).