Multiple Mechanisms Drive Calcium Signal Dynamics around Laser-Induced Epithelial Wounds

Abstract

Epithelial wound healing is an evolutionarily conserved process that requires coordination across a field of cells. Studies in many organisms have shown that cytosolic calcium levels rise within a field of cells around the wound and spread to neighboring cells, within seconds of wounding. Although calcium is a known potent second messenger and master regulator of wound-healing programs, it is unknown what initiates the rise of cytosolic calcium across the wound field. Here we use laser ablation, a commonly used technique for the precision removal of cells or subcellular components, as a tool to investigate mechanisms of calcium entry upon wounding. Despite its precise ablation capabilities, we find that this technique damages cells outside the primary wound via a laser-induced cavitation bubble, which forms and collapses within microseconds of ablation. This cavitation bubble damages the plasma membranes of cells it contacts, tens of microns away from the wound, allowing direct calcium entry from extracellular fluid into damaged cells. Approximately 45 s after this rapid influx of calcium, we observe a second influx of calcium that spreads to neighboring cells beyond the footprint of cavitation. The occurrence of this second, delayed calcium expansion event is predicted by wound size, indicating that a separate mechanism of calcium entry exists, corresponding to cell loss at the primary wound. Our research demonstrates that the damage profile of laser ablation is more similar to a crush injury than the precision removal of individual cells. The generation of membrane microtears upon ablation is consistent with studies in the field of optoporation, which investigate ablation-induced cellular permeability. We conclude that multiple types of damage, including microtears and cell loss, result in multiple mechanisms of calcium influx around epithelial wounds.

Introduction

Epithelial wound healing is a multistage process. Cells must detect the presence of a wound, migrate and proliferate in a coordinated fashion to close the defect, and then successfully reestablish tissue-wide epithelial architecture (1, 2, 3, 4, 5). An important early feature of the wound response is a rapid rise in cytosolic calcium. This rise initially occurs in cells near the wound margin and then spreads to more distant cells (6, 7, 8, 9, 10, 11) to act as a potent signal that regulates several cellular responses around wounds: JNK signaling (12), Rho GTPase activity (13), and remodeling of the actin cytoskeleton (7, 8, 13, 14). Nonetheless, it is not yet clear how wound-induced calcium signals are initiated in vivo. In cell-culture wounding models, calcium signals are generated downstream of a diffusible ligand released by damaged cells into the extracellular space (15, 16, 17). In vivo studies suggest another model, in which wound-induced changes in tissue tension open stretch-activated calcium channels or activate other mechanosensitive proteins (18). In support of this hypothesis, calcium waves can be generated in Drosophila wing disks by applying mechanical pressure (19) and are perturbed in both Caenorhabditis elegans and Drosophila wounding models after knocking out the putative stretch-activated calcium channel TRPM (7, 8, 10). Importantly, the diffusible-ligand and altered-mechanics hypotheses are not mutually exclusive: both could be upstream initiators of wound-induced calcium signals in vivo, each acting through specific regulated receptors or channels.

Here, we use pulsed laser ablation to create repeatable and controllable wounds in epithelial tissues in Drosophila larvae and pupae, and carefully measure the dynamics of the induced calcium response in surrounding cells over timescales from milliseconds to hundreds of seconds. We observe a complex spatiotemporal response with multiple phases: initial calcium influx beginning within milliseconds at discrete loci as far as 70 μm from the wound site; a rapid, intercellular expansion of calcium away from these loci; and a second, slower expansion of high cytosolic calcium to additional cells as far as 150 μm from the wound. We show that the initial influx and subsequent expansion phases can be described by different diffusive processes, suggesting different expansion mechanisms, and can be linked to different aspects of cellular damage around laser-induced wounds.

Although laser energy can be precisely focused to remove single cells or even subcellular elements (20, 21), additional cells near and far from the primary wound are affected through several mechanisms. The lasers typically used to investigate the mechanics of morphogenesis and wound healing in vivo are pulsed (22, 23, 24, 25), with femtosecond to nanosecond pulsewidths, and ablate tissue via plasma formation (26, 27, 28). The plasma directly destroys macromolecules within the laser focus (radius < 1 μm), but recombination of the plasma leaves the targeted location extremely hot, leading to the vaporization of water and expansion of a cavitation bubble (26). The bubbles associated with laser ablation vary with pulsewidth and fluence (26, 27, 28); those measured for near-threshold in vivo ablation using nanosecond pulse lasers have lifetimes on the order of microseconds, and inferred maximum radii of 10–100 μm (28, 29). In cell-culture studies, the rapid expansion and collapse of cavitation bubbles generates shear stresses that can lyse cells close to the ablation site, creating a primary wound, and can transiently permeabilize cells farther away (30, 31, 32), a process known as optoporation.

Here, we show that two mechanisms of laser-induced damage, a tissue-level primary wound and cellular microtears, each drive distinct calcium-signaling dynamics. By imaging wound-induced calcium dynamics with unprecedented time resolution, we show that calcium directly enters permeabilized cells 20–100 μm from the ablation site within milliseconds. This influx initiates a first calcium wave. In some cases, a second wave further expands the region of high cytosolic calcium, but the second wave is delayed by 30–60 s and its occurrence depends on the size of the primary wound. Through kinematic modeling of the calcium influx and expansion waves, we have identified parameters of the wave spread that narrow and inform the search for upstream mechanisms. Our findings have implications both specific to laser ablation and applicable to more general types of wounding. First, the precision lysis of laser ablation may be accompanied by cavitation bubble effects akin to a wider crush injury. Second, our general finding of two-phase initiation of calcium signaling is likely to be important for tissues sensing a wide range of traumatic damage.

Materials and Methods

Additional methods including fly genotypes, image processing, and computational analysis can be found in the Supporting Materials and Methods.

Laser ablation and live imaging

White prepupae expressing pnr-Gal4 and UAS-GCaMP3, UAS-GCaMP6m, or UAS-Arclight were aged for 12–18 h after puparium formation. Pupae were mounted with nota facing the coverslip. Wing disks expressing tubP-Gal4 and UAS-GCaMP6m were dissected from third-instar larvae and immediately mounted on coverslips for imaging and ablation.

Laser ablation and live imaging were performed using a Zeiss LSM410 raster-scanning inverted confocal microscope with a 40 × 1.3 NA oil-immersion objective. Laser wounding used single pulses of the third harmonic (355 nm) from a Q-switched Nd:YAG laser (5 ns pulsewidth; Continuum Minilite II, Santa Clara, CA) at pulse energies ranging from 0.5 to 10 μJ. Kymographs were assembled and basic image measurements were taken in ImageJ (National Institutes of Health, Bethesda, MD).

Cavitation experiments

Pupae were mounted on coverslips in distilled water, and wing disks were mounted in calcium-free phosphate-buffered saline (PBS). A small hydrophone (Onda, 0.5 mm aperture, <20 ns rise time, 2.24 V/MPa sensitivity) was mounted to the confocal stage and lowered into the aqueous medium ∼1 mm from the focus of the laser. Bubble lifetime measurements were taken from hydrophone pressure traces displayed on an oscilloscope.

FM 1-43 and propidium iodide analysis in wing disks

Wing disks were mounted on a coverslip 20 min before imaging. FM 1-43 (#T35356; Thermo Fisher Scientific, Waltham, MA) was diluted in PBS to 5 μg/mL and propidium iodide (PI; #P3566; Thermo Fisher Scientific) was diluted in PBS to 20 μg/mL. Each were added to wing disks immediately before imaging, and experiments were completed within 10 min of adding dye.

Results

Fast Ca²⁺ signal dynamics from 2 ms to 2 s

To investigate wound-induced mechanisms of calcium (Ca²⁺⁾ release, we analyzed Ca²⁺ dynamics in the Drosophila pupal notum (following procedures in (7)). At 12–18 h after puparium formation, the notum is a continuous epithelial monolayer of diploid cuboidal cells that exhibit apicobasal polarity and sit atop a basement membrane (33). Pupae were wounded via laser ablation and imaged live simultaneously. Half of wounded pupae survived laser ablation and later eclosed (data not shown and (7)). Wound-induced Ca²⁺ waves were visualized with a genetically encoded, intracellular GCaMP3 reporter (34).

Before ablation, cytosolic Ca²⁺ remained at low basal levels (Fig. 1 A, t = −2 s), with single cells occasionally exhibiting transient Ca²⁺ increases. In the first frame (<2 s) after ablation, a dramatic increase in cytosolic Ca²⁺ was observed as bright GCaMP fluorescence, both in cells along the wound margin and in surrounding cells up to seven cell diameters away (∼50 μm) (Fig. 1 A; Movie S1). To analyze this initial response with improved temporal resolution, we performed line scans passing through the wound to increase our sampling rate to 2.1 ms per scan. For these experiments, we also employed a GCaMP6m Ca²⁺ indicator because it has faster kinetics with an increased signal/noise ratio. These line scans were assembled into kymographs, an example of which is shown in Fig. 1 B (additional examples shown in Fig. S1). Within milliseconds after ablation, Ca²⁺ entered the cytoplasm and increased GCaMP6m fluorescence in multiple cells up to ∼50 μm away from the ablated region. We call this rapid increase of intracellular Ca²⁺ the “initial response.”

Figure 1.

The initial response to wounding matches the radius of the cavitation bubble. (A) Confocal images of the Ca²⁺ reporter GCaMP3 in the pupal notum. Basal levels of cytosolic Ca²⁺ are low before wounding (i), but rise rapidly after laser ablation (ii). The rapid rise occurs in cells within five to seven rows from the ablation site (cross-hairs in i, central dark area in ii). Scale bars represent 50 μm. (B) Confocal kymograph of GCaMP6m. Fluorescence is low before wounding (kymograph lines above t = 0 ms), but rises within milliseconds of laser ablation in cells distant from the wound site (central dark area for t > 0 ms). Scale bars represent 50 μm. (C) The radius of the primary wound-induced Ca²⁺ signal (R_S,0) corresponds to the maximum radius of the laser-induced cavitation bubble (R_B,max). Linear regression yields a best fit with a slope of 0.99 (solid line; equation; R² = 0.87). Each data point corresponds to a single wound with initial Ca²⁺ signal radii measured from full-frame confocal images (<3 s after ablation) and bubble radii calculated from bubble lifetimes (t_B) measured via hydrophone (inset). Horizontal error bars represent estimated uncertainty in identifying the bubble collapse peak; vertical error bars are SDs of four radius measurements. To see this figure in color, go online.

Few biological signals could propagate away from the wound site quickly enough to drive this spatially distributed initial Ca²⁺ response. For example, the kymograph in Fig. 1 B shows rises in intracellular Ca²⁺ that begin within 5 ms for cells up to 50 μm from the wound. To reach this far so quickly, any ablation-induced diffusible signal would need a diffusion constant of α ∼1.3 × 10⁵ μm²/s, estimated from (Δx)²/4Δt. This estimated diffusion constant is two to three orders of magnitude faster than that of a small molecule like ATP (3.5 × 10² μm²/s), or of small ions (0.8–2.0 × 10³ μm²/s), or of even the self-diffusion of water (2.3 × 10³ μm²/s). Thus, a diffusible signal cannot trigger the initial rapid response.

The initial Ca²⁺ response is driven by cavitation

Pulsed laser ablation generates a plasma that destroys macromolecules in a nearly diffraction limited region (<1 μm diameter), and generates a rapidly expanding cavitation bubble with high shear stresses that lyse cells in a variably broader region. Importantly, the region of macromolecular destruction and region of lysis are both much smaller than the initial GCaMP response region we observe. The cavitation bubble is short-lived, but it can expand hundreds of microns beyond the region of lysed cells (29, 30, 31). To investigate whether the initial Ca²⁺ response could be driven by cavitation, we simultaneously measured initial Ca²⁺ response radii and cavitation bubble radii.

We estimated cavitation bubble radii using a hydrophone to detect the acoustic transients associated with bubble expansion and collapse. For water at standard temperature and pressure, maximum bubble radii (R_B,max) are theoretically related to cavitation bubble lifetimes (t_B), according to R_B,max ∼ (5.46 μm/μs) t_B (29). Validation of hydrophone measurements based on this relationship is provided by direct flash imaging of bubbles in Drosophila embryos, as shown in Fig. S2. For each experiment in pupae, we also collected full-frame confocal images of GCaMP6m fluorescence to record the initial Ca²⁺ response radius. With laser pulse energy intentionally varied to create both large and small cavitation bubbles, we found that cavitation radius matched the radius of initial response radius (R_S,0) with a slope very close to one: the best linear fit to the relationship was R_S,0 = (0.99 ± 0.08)R_B,max + (6.5 ± 6.0 μm) (Fig. 1 C). Our initial response measurements are an overestimate as the Ca²⁺ region can expand by as much as 17.6 ± 7.0 μm (mean ± SD) between the time of ablation and the capture of images used to determine the radius of Ca²⁺ influx (<2 s). Correcting for this overestimate would reveal a trend line that runs very close to the origin. We conclude that the cavitation bubble is tightly linked to the extent of the initial Ca²⁺ response, and we explore the mechanism below.

The initial response begins at discrete loci

Within milliseconds of ablation, kymographs of GCaMP6m fluorescence show the contemporaneous appearance of multiple high-Ca²⁺ loci (Fig. 2 A, arrowheads). This pattern cannot be due to Ca²⁺ flow from the wound margin through neighboring cells, but instead indicates Ca²⁺ entering the cytosol at multiple distinct locations around the wound. To quantify the spatial distribution of these Ca²⁺ influx loci, we first identified each kymograph’s signal front, i.e., the time points when the fluorescence signal in each kymograph column first exceeded the unwounded background by two SDs (Fig. 2 A, solid white line). Considering this signal front as a function t_front(x), we identified influx loci as its local minima, which appear as peaks because of the inverted time axis of our kymographs (Fig. 2 A, arrowheads). Although Ca²⁺ influx loci far from the wound appear as spatially distinct peaks, those close to the wound blend together into a wide band of high fluorescence. For this wide band of unresolved fluorescence near the wound, each pixel was counted as a Ca²⁺ influx site.

Figure 2.

Early wound-induced Ca²⁺ appears at discrete loci. (A) Confocal kymograph of GCaMP6m showing discrete sites of cytosolic Ca²⁺ entry in cells distant from the wound. These sites are marked by yellow arrowheads and were identified as local maxima along the signal front (outline in lower panel). Scale bars represent 50 μm. (B) Density of Ca²⁺ signal initiation sites as a function of relative distance from the wound center. Data were compiled from 25 kymographs with distances normalized to each kymograph’s maximum signal radius (R_max). Broad regions near the wound with rapid Ca²⁺ rise, but no discernable peaks were treated as having one initiation site per kymograph pixel. Brackets above the plot demarcate different zones of initiation site density. To see this figure in color, go online.

We plotted the radial density of influx sites around primary wounds by compiling data from 25 kymographs and normalizing peak locations to each kymograph’s maximum signal radius (R_max). Fig. 2 B shows the primary wound plus two distinct zones of Ca²⁺ influx site density: a high-density zone corresponding to the fast, unresolved band of high signal (0.1 R_max < r < 0.4 R_max); and a zone of lower density corresponding to the region of distinct kymograph peaks (0.4 R_max < r < R_max). Although the density in this latter region is ∼1 site per 10 μm, which is close to one per cell, the Ca²⁺ influx sites do not fall in a regular pattern with respect to cell borders (Fig. S3).

The spatially distinct kymograph peaks show that Ca²⁺ enters the cytoplasm of cells around the wound at discrete loci, and then spreads rapidly (Fig. 3 A). To quantify this spread, we individually fit 25 peaks to a three-dimensional diffusion model, i.e., the expected time-dependent signal along a sampled line due to diffusion from a point source of constant Ca²⁺ influx. An example fit for one peak is shown in Fig. 3 B, with selected temporal and spatial slices used to show overlays of the data and fit (Fig. 3, C and D). The complete set of kymograph peak fits is available as Fig. S4. From this complete set of fits, the interquartile range of estimated diffusion constants was 76–220 μm²/s, with a median of 120 μm²/s (Fig. 3 E). This range lies between the diffusion constants of cytosolically buffered Ca²⁺ (13 μm²/s) and free Ca²⁺ (220 μm²/s) (35). The breadth of the fitted diffusion-constant distribution can be attributed to a combination of expected random error, systematic errors based on using an infinitesimal point source equation that would slightly overestimate the diffusion constants from finite-sized influx sites, and cell-to-cell variations in the degree to which Ca²⁺ influx levels saturate cytoplasmic and GCaMP6 buffering capacity.

Figure 3.

Wound-induced intracellular Ca²⁺ diffuses from discrete loci at rates that appear to overcome cytosolic buffering capacity. (A) Kymograph of GCaMP6m fluorescence upon wounding. A single peak (yellow arrowhead) is examined in more detail in (B–D). Scale bars represent 50 μm. (B) Enlarged view of an isolated kymograph peak (left) and the best fit of that peak to a diffusion model (right; see main text for details). Scale bars represent 2 μm horizontal and 10 ms vertical. (C) Intensity versus time for two marked columns cutting through the selected kymograph section (at 0 and 2.5 μm from the center of the peak). (D) Intensity versus position for two marked rows cutting through the selected kymograph section (at 20.9 and 41.8 ms after laser ablation). (E) Box-and-whisker plot of the diffusion constants from best fits of 25 isolated peaks. Bars show median and interquartile range. Arrowheads indicate diffusion constants of free Ca²⁺ (black, 220 μm²/s) and cytosolically buffered Ca²⁺ (white, 13 μm²/s) as measured by (35). To see this figure in color, go online.

The cavitation bubble creates plasma membrane microtears

Our results thus far show that Ca²⁺ enters the cytoplasm from discrete loci spread throughout the maximum extent of the cavitation bubble. Previous reports show that cavitation bubbles induce cellular damage (36, 37), and this cellular damage is associated with rises in intracellular Ca²⁺ (38, 39, 40, 41). Plasma membrane microtears would provide a simple direct mechanism for the inflow of Ca²⁺ into the cytoplasm from its high concentration in the extracellular space (∼10⁻³ M extracellularly vs. ∼10⁻⁷ M in cytoplasm). We tested for ablation-induced microtears in three types of assays described below: depolarization, dye internalization, and altered extracellular Ca²⁺.

First, we measured electrical depolarization. Epithelial cells maintain an electrical potential or voltage difference across their plasma membranes, and microtears would allow the free movement of Na⁺, Cl⁻, and K⁺ ions to eliminate this membrane potential and thus depolarize the cells. Ca²⁺ would also cross the plasma membrane, but Ca²⁺ ions do not contribute significantly to establishing or depolarizing the electrical potential (42). We visualized cavitation-induced changes in membrane potential using a genetically encoded voltage indicator, Arclight, whose fluorescence decreases upon depolarization (43, 44). Although Arclight kinetics are slower than those of GCaMP6m, we observed a slight decrease in fluorescence in the first frame upon wounding in a region centered around the wound (Movie S2), and this decrease became more pronounced over the next 30–60 s (Fig. 4 A). As shown above for the initial Ca²⁺ signal, the radius of the depolarized region also matched the radius of cavitation (Fig. 4 B), linking both to the maximum extent of the cavitation bubble. Over the course of 10 min, damaged but unlysed cells within the cavitation footprint repolarized, indicating that cells survive microtear damage (Fig. S5; Movie S3).

Figure 4.

The cavitation bubble creates microtears in plasma membrane. (A) Arclight fluorescence in the notum is high before wounding (i) and decreases around the wound afterwards, indicating depolarization (ii). The margin of depolarization is marked by red arrows. (B) The radius of wound-induced depolarization (R_Volt) corresponds to the maximum radius of the laser-induced cavitation bubble (R_B,max) with a best fit slope of nearly one (solid line; equation; R² = 0.91). R_Volt was measured from confocal images taken on the first frame (<3 s) after ablation. (C) Fluorescence labeling of an ex vivo wing disk with FM 1-43 is modest before wounding (i), and increases around the wound afterwards, indicating a region of increased cell permeability and dye influx (ii). The outer margin of dye influx is marked by yellow arrows. Scale bars represent 50 μm. (D) The radius of dye influx (R_Dye) also corresponds to R_B,max with a best fit slope of nearly one (solid line; equation; R² = 0.98). The radius of dye influx was measured from confocal images taken 60 s after ablation. (E) Again, the radius of dye influx (R_Dye) corresponds to R_B,max with a best fit slope of nearly one (solid line; equation; R² = 0.92). The radius of dye influx was measured from confocal images taken ∼5 min after ablation. (F) Wing disks expressing GCaMP6m were ablated ex vivo in Ca²⁺-free media. Fluorescence is low before wounding (i) and only increases afterwards in cells close to the wound (ii). Scale bars represent 50 μm. (G) The radius of the initial wound-induced Ca²⁺ rise in Ca²⁺-free media (R_0 mM) is significantly smaller than R_B,max, with a best fit slope of 0.25 (red, open triangles; R² = 0.60). In contrast, when wing disks are wounded in 2 mM Ca²⁺, the initial wound-induced Ca²⁺ radius (R_2 mM) corresponds to R_B,max with a best fit slope of nearly one (blue, closed circles; R² = 0.90). Best fit equations shown above graph. Horizontal error bars represent estimated uncertainty in identifying the collapse peak; vertical error bars are SDs of four radius measurements. To see this figure in color, go online.

Next, we used two robust and well characterized microtear assays based on two cell-impermeable fluorophores: the lipophilic dye, FM 1-43 (45, 46, 47, 48, 49, 50, 51, 52), and the DNA intercalating agent, PI (53, 54, 55, 56). FM 1-43 fluoresces only when bound to lipid membranes. Before wounding, dye binds only the outer leaflet of the cells’ plasma membranes. If wounding generates microtears, dye can then enter cells and label the inner leaflet of the plasma membrane, increasing its fluorescence intensity. We could not conduct these experiments in Drosophila pupae because they have an impermeable waxy cuticle that prevents dye from accessing notum cells and membranes. We thus used Drosophila wing disks, larval precursors of the pupal notum and wing that can be cultured ex vivo. We laser-wounded wing disks submerged in FM 1-43 and simultaneously imaged fluorescence and tracking cavitation with a hydrophone. After laser ablation, plasma membrane fluorescence increased, indicating inner-leaflet labeling and the presence of microtears in a circular region around the wound (Fig. 4 C; Movie S4). The radius of increased fluorescence matched the maximum radius of the cavitation bubble (Fig. 4 D) and did not expand with time. High-magnification imaging shows FM 1-43 preferentially labeling the plasma membrane with little labeling of endocytic vesicles, suggesting a route of entry that bypasses endocytosis (Fig. S6). This route could be microtears, but because FM 1-43 has been observed to pass through some plasma membrane channels (57, 58), we performed a similar experiment using the fluorophore PI. Upon laser ablation, the radius of PI-positive cells also matched the radius of cavitation (Fig. 4 E). These results show the internalization of two different fluorophores and suggest a nonspecific route of dye entry.

Finally, we removed extracellular Ca²⁺ and measured the effects on wound-induced Ca²⁺ signals. We could not alter extracellular Ca²⁺ concentrations in vivo in the notum, so we ablated wing disks ex vivo as above, but in media with or without Ca²⁺. Wounding in the presence of physiological [Ca²⁺] (2 mM) resulted in a Ca²⁺ influx radius that matched the cavitation radius (Fig. 4 G). Wounding in Ca²⁺-free media did not eliminate cytosolic Ca²⁺ influx, but the radius of this influx was just ∼1/4 the cavitation radius (Fig. 4, F and G). We conclude that, in vivo, most of the cavitation-induced Ca²⁺ influx is derived from an extracellular pool. Interestingly, the small region of Ca²⁺ influx observed in Ca²⁺-free media has similar dimensions to the zone of high influx-site density identified in kymographs in vivo in the notum (Fig. 2 B). We discuss this connection and possible explanations in the Discussion. Nonetheless, the strongly reduced extent of initial Ca²⁺ signals in Ca²⁺-free media, in concert with the cavitation-linked changes in membrane potential and dye permeability shown above, strongly support the hypothesis that most of the initial Ca²⁺ influx comes through cavitation-induced microtears in the cells’ plasma membranes.

Additional wound-induced Ca²⁺ signaling on timescales of seconds to minutes

After the initial influx, the region of high intracellular Ca²⁺ around the wound typically undergoes two stages of radially symmetric expansion (Movies S1 and S5). The first stage follows directly from the microtears generated in the footprint of the cavitation bubble and expands the high-Ca²⁺ region radially outward ∼20 μm over a period of 15–20 s (Fig. 5 A, i and ii). This first expansion is eliminated by expressing RNAi against the Drosophila gap junction proteins Inx2 or Inx3 (Fig. 5, C and D; Movie S6), demonstrating signal propagation via gap junctions, consistent with some previous findings on other Ca²⁺ waves (10, 19, 59). The signal diffusing through gap junctions could simply be the leakage of Ca²⁺ from damaged to undamaged cells; alternatively, the first expansion could be mediated by IP3, similar to Ca²⁺ waves in wing disks and in other wounding systems (8, 19, 59, 60). Cytoplasmic Ca²⁺ buffering makes direct Ca²⁺ transport through gap junctions less common than IP3 transport (61), but the estimated diffusion constants in Fig. 3 E suggests the presence of saturating concentrations of Ca²⁺ that could overcome cytoplasmic buffering and thus flow directly through gap junctions. After the brief expansion, this high-Ca²⁺ region then shrinks modestly before a second expansion becomes evident 40–200 s after ablation (Fig. 5 A, iii and iv; Fig. S7). The second expansion is typically larger than the first and does not rely on gap junctions (Fig. 5, C and D). After the second expansion, the high-Ca²⁺ region begins to shrink radially and send off localized flares (Fig. 5 A, v–viii).

Figure 5.

Two wound-induced Ca²⁺ signal expansions occur on different timescales via different mechanisms. (A) Stills, taken from Movie S5, of in vivo live imaging of GCaMP6m in the notum. In the first frame after laser ablation, an increase of cytosolic Ca²⁺ is observed in an ∼5–7 cell radius around the wound (i). Seconds after wounding, the region of increased fluorescence expands to adjacent cells (ii), before contracting slightly (iii). The region of increased fluorescence then expands concentrically again (iv), before breaking into propagating, anisotropic Ca²⁺ flares (v–viii). The flares continue initiating for >30 min after wounding whereas the central region cyclically expands and contracts. Scale bar represents 50 μm. (B) The radial expansion of Ca²⁺ signaling is plotted over time for 24 samples: each line represents a different sample with radii reported from the wound center. The initial region expands briefly in all samples before contracting. A second expansion may then occur and does so more frequently in larger wounds. No second expansion occurs when no wound is present. (C) The radial expansion of Ca²⁺ signaling over time for representative samples expressing either Inx2 or Inx3 RNAi in the pnr domain. A similar pattern was seen with a second Inx2 RNAi construct (data not shown). Knocking down gap junctions blocks the first expansion but not the second. (D) Kymographs showing the expansion of Ca²⁺ signals over time for control (i) and Inx3 knockdown samples (ii). The curved and straight yellow lines in controls indicate the normal shape of the first and second expansion. In Inx3 knockdowns, the first expansion is absent; the second is perturbed but still occurs. Wound location and time are indicated by X, horizontal scale bars represent 50 μm, and vertical scale bars represent 15 s. To see this figure in color, go online.

To further analyze the expansion of Ca²⁺ signals over time, we condensed each movie to a graph representing the expansion of the Ca²⁺ wave with respect to time. This condensation involves radial averaging to identify the average edge of the Ca²⁺ wave (see Supporting Materials and Methods), and is thus most appropriate for analyzing time periods in which the high-Ca²⁺ region is radially symmetric, e.g., during the first and second expansions, but not once flaring commences. A collection of Ca²⁺ signal radius graphs for 24 laser ablation experiments is shown in Fig. 5 B. All of these experiments have a first signal expansion, but the presence of a second expansion depends on the presence and size of the primary wound. Large wounds (>35-μm diameter, blue) always exhibited a second expansion, and intermediate wounds (15–35 μm, green) sometimes did, but small wounds (<15-μm diameter, magenta) either had no second expansion or an atypical weak one. Furthermore, some samples exhibited a first expansion but no visible wound likely due to slight mistargeting of the laser pulse, which could create cavitation without ablation; these samples never exhibited a second expansion. Thus, although the first expansion is tightly linked to cavitation-induced microtears, the second requires some cells to be damaged beyond repair and becomes more likely as the primary wound size increases.

We fitted each expansion of the Ca²⁺ signal radius to a two-dimensional diffusion model to assess the kinematics and to begin to elucidate the signals driving each stage. This model assumes that some unknown signal (X) is released at a specific time (t₀) over a distributed area, given by an axisymmetric two-dimensional Gaussian model, with 1/e² radius σ₀ and total signal amount M. This signal diffuses with diffusion constant α and triggers a Ca²⁺ influx wherever the local concentration exceeds a threshold (C_th). We acknowledge that the source is likely more complicated than a two-dimensional Gaussian model, but details of the distribution shape beyond its root-mean-square radius ($\sqrt{2}$σ₀) have no discernable effects over the timescales of our measurements. With this model, the time-dependent radius of the high-Ca²⁺ region (R_S) is thus given by

$$R_{\text{S}}^2=2{\sigma }^2\text{ln}\left[\frac{1}{2\pi {\sigma }^2\left(C_{\text{th}}/M\right)}\right]\mathrm{with}{\sigma }^2={\sigma }_0^2+2\alpha \left(t-t_0\right).$$ (1)

We fit models of this form to each expansion independently. The intervening and final shrinkage phases were not fit to this model because they involve additional mechanisms and timescales governing the return to baseline Ca²⁺ levels. The model fit the first expansion well with t₀ set to the time of laser ablation. This left three fit parameters: σ₀, α, and C_th/M. The second expansion could not be fit well with t₀ = 0, so we instead set t₀ to equal the time at which each second expansion became apparent. Example fits are shown in Fig. 6 A, with the first expansions highlighted in red and the second in blue. The complete set of fits is shown in Fig. S7. Note that the fitted equation has a degeneracy with respect to σ₀ and t₀, i.e., broadly distributed sources releasing a signal at late times would be equivalent to narrowly distributed sources releasing the same signal earlier. This becomes a particular issue when fitting the second expansion because choosing earlier values for t₀ would lead to equivalent fits with smaller values for σ₀. Considering this degeneracy, fits to the second expansion have been back propagated (dashed lines, Fig. 6 A) to indicate that the driving signal could have been released before the second expansion became apparent. Notably, the back propagations do not go all the way back to t = 0. They instead intersect the time axis at a median time of 47 s (interquartile range of 29–61 s), strongly suggesting that the unknown signal driving the second expansion is either released tens of seconds after ablation or has a delay introduced by cellular signal transduction.

Figure 6.

Parameterization shows distinct characteristics for each expansion. (A) Four examples showing the expansion of Ca²⁺ over time. The initial and secondary expansions are highlighted in red and blue, respectively. The solid lines show diffusional fits to each expansion; the dashed blue lines show back projections of fits to the second expansion. The inset shows an expanded view of the fit to the data from the outlined region. The complete set of analyzed curves (N = 28) is shown in Fig. S7. As a measure of goodness-of-fit, the SEs of the regression for the four fitted first expansions were (from top to bottom) 0.19, 0.18, 0.95, and 1.04 μm, and for the four fitted second expansions were (from top to bottom) 1.62, 1.76, 1.21, and 1.30 μm. (B) Box-and-whisker plots of the best fit parameters for all first and second expansions: α is the diffusion constant; C/M is the ratio of the signal threshold to the amount released; and σ₀ is the 1/e² radius of initial signal distribution. Median and interquartile range are displayed. (C) Variability among the fitted diffusion constants for each expansion is partially explained by a dependence on the wound diameter. To see this figure in color, go online.

The best fit parameters for both expansions are compiled in the box-and-whisker plots of Fig. 6 B. The 1/e² radii of the initial signal distributions are comparable (24 vs. 22 μm, assuming the second expansion starts when it first becomes apparent). On the other hand, the second expansion yields significantly smaller values for both the diffusion constant of the signal (median values for α of 53 vs. 20 μm²/s; p = 1.5 × 10⁻⁵, Mann-Whitney U test), as well as for the relative signal threshold (median values for C_th/M of 2 × 10⁻⁵ μm⁻² vs. 6 × 10⁻⁶ μm⁻²; p = 7 × 10⁻⁸). These results suggest that each expansion is driven by a different diffusive signal.

Interestingly, the diffusion constants for both the first and second expansion increase with wound size (Fig. 6 C). In fact, for large wounds, the diffusion constant of the first expansion approaches the value for free diffusion of Ca²⁺. These observations suggest a model in which larger wounds release more of both signals, and thus yield larger, effective diffusion constants by overcoming more of the buffering or binding capacity of the environment.

Discussion

When a pulsed laser creates wounds in the epithelium of the Drosophila pupal notum, these wounds trigger a complex series of Ca²⁺-signaling dynamics: a rapid influx into nearby surviving cells that matches the footprint of the laser-induced cavitation bubble; followed by a spreading of the high-cytosolic-Ca²⁺ region via two sequential and concentric waves. These dynamics and a model that explains them are summarized in Fig. 7 and Movie S7.

Figure 7.

Summary of Ca²⁺ signal dynamics after laser wounding. Plasma formation and cavitation: pulsed laser ablation generates a localized plasma at the laser focus (X), which then recombines and leads to expansion of a cavitation bubble (dashed black outline denotes maximum bubble radius). Cavitation induced Ca²⁺ entry: the cavitation bubble expands, damaging cells as it spreads. Close to the point of ablation, cavitation-induced shear stresses lyse cells, creating the primary wound (black). Cells close to the primary wound undergo extensive damage (dark orange). Microtears on the plasma membrane and in organelle membranes result in Ca²⁺ influx from the extracellular space and internal stores milliseconds after wounding (black arrows in dark orange cells). The shear forces applied to cells are attenuated as the cavitation bubble expands and slows. Cells far from the wound (light orange) exhibit plasma membrane microtear damage, which allows Ca²⁺ entry from the extracellular space (black arrows in light orange cells). Cells that were not lysed to create the primary wound survive cavitation damage. First expansion: seconds after wounding, high concentrations of Ca²⁺ that have just entered the cells spread intracellularly through gap junctions (black arrows) to neighbors (yellow). Second expansion: after wounding, a delayed second expansion of Ca²⁺ spreads to cells distant from the wound (blue). This expansion is driven by extracellular ligand diffusion (blue arrows), but gap junctions may still have some role in generating a smooth wavefront. Modeling of the second expansion shows that it is characteristically different from the first expansion. To see this figure in color, go online.

The initial Ca²⁺ influx occurs at hundreds of distinct loci and spreads throughout the affected cells in <0.1 s. It spreads intracellularly with diffusion constants near that of free Ca²⁺, suggesting that high concentrations of Ca²⁺ are flooding the cells and overcoming cytosolic buffering capacity. We propose that each locus corresponds to a cavitation-induced, plasma membrane microtear that allows Ca²⁺ influx from the extracellular environment. In support of this model, cells in the cavitation footprint also become electrically depolarized and permeable to entry of two cell-impermeable dyes. Further, wounding in Ca²⁺-free media strongly reduces the extent of the initial rise in cytosolic Ca²⁺. These results could also be explained by cavitation-induced shear stresses opening mechanosensitive channels rather than microtears; however, the variety of observed effects implies a nonspecific route of entry. We thus consider microtears more likely.

Membrane permeabilization is known to occur when laser-induced cavitation bubbles are used to introduce DNA, drugs, or microparticles into cultured cells via a process known as optoporation (62, 63, 64, 65, 66, 67). The shear stress associated with cavitation-bubble expansion and collapse can induce zones of acute cell lysis, cell necrosis, and membrane disruption (30, 31). We note similar zones of damage in our studies, observed as a primary wound surrounded by regions with a high and then low density of Ca²⁺ influx sites (Fig. 2 B). This suggests more severe cellular damage close to the wound, which may be linked to the limited region of cytosolic Ca²⁺ influx observed after wounding in Ca²⁺-free media. In fact, the radius of this limited region and the radius of high-density damage identified in kymographs represent similar fractions of the cavitation bubble radius (0.25 and 0.3, respectively). In this region proximal to the wound, the cavitation bubble and its shear stresses may generate additional microtears in organelle membranes that release Ca²⁺ from the endoplasmic reticulum, mitochondria, and Golgi, or may trigger mechanosensitive GPCR-induced release of Ca²⁺ from these same intracellular stores (68).

After the initial influx, the high-Ca²⁺ region expands via two sequential waves. The first wave spreads with a diffusion constant of 32–87 μm²/s, with faster diffusion observed in bigger wounds. These rates still exceed that of cytosolically buffered Ca²⁺ and suggest a simple model for the first wave based on the diffusion of excess Ca²⁺ into neighboring cells through gap junctions. Indeed, variations in how well the excess Ca²⁺ is buffered could explain this wave’s wide range of fitted diffusion constants. Furthermore, this expansion is short-lived and short-ranged, which would be predicted as the Ca²⁺ concentration decreases as it spreads and would thus be well buffered further from the wound. Although we favor this model in which the first expansion is caused by the direct diffusion of Ca²⁺ to neighboring cells, the diffusion of IP3 has been identified as important for signal expansion in other wounding systems (8, 19, 59, 60) and remains an alternative possibility.

The second wave begins ∼45 s after wounding, but is not always present. The second wave occurs more frequently after larger wounds, suggesting its critical dependence on the extent of primary wound-induced damage. Interestingly, the second wave has a delayed start (at least 29–61 s after ablation), which may represent a delayed release of signal or time required for cells to transduce the signal into a Ca²⁺ response. The second wave spreads with a diffusion constant of 7–30 μm²/s, which is much slower than the first wave. This suggests distinct signals driving the two expansions, but further experiments are needed to identify the second wave’s time delay and spread mechanism. After the second expansion, the high-Ca²⁺ region begins sending off asymmetric directional flares, likely representing waves of Ca²⁺-induced Ca²⁺ release moving throughout the tissue. Each flare lasts tens of seconds, but new ones continue starting for more than 30 min after wounding. These flares are similar to Ca²⁺ oscillations reported after epithelial wounds in zebrafish (69, 70) and fly embryos (10).

Previous studies have identified two models for wound-induced Ca²⁺ wave initiation: mechanosensitive Ca²⁺ channels (7, 8, 10, 18, 19, 68, 71) and extracellular diffusible ligands (9, 15, 16, 72, 73, 74). In many cell-culture models, the driving signal for the Ca²⁺ wave propagates extracellularly and very fast, with speeds ranging from 4.6 to 49.3 μm/s (6, 9, 75). This is much faster than the speeds of Ca²⁺-signal spread in our in vivo experiments: the median speeds were 2.9 and 1.7 μm/s for the initial portions of the first and second expansions, and these slowed in a diffusive manner. These rates are more similar to other, in vivo wound-healing models, in which Ca²⁺ signals propagate via gap junctions and spread at rates of 0.4–6.9 μm/s (10, 19, 59). Signal propagation via extracellular ligand diffusion may occur in the Drosophila notum during the second wave as this expansion is not dependent on gap junctions, but the ligand would have to diffuse much more slowly than those identified in cell-culture models.

Once Ca²⁺ enters a cell, regardless of how, it is a master regulator of wound healing. Ca²⁺ regulates Rho (13, 76, 77) and the actin cytoskeleton (7, 14), activates JNK signaling (12), prevents apoptosis (11), and increases hydrogen peroxide and inflammation around wounds (10). Ca²⁺ waves alert cells to the presence of a wound and permit the activation of healing programs (70). Given the variety of types of wounds that may need to be healed, multiple mechanisms may have evolved to initiate Ca²⁺-signaling cascades. Our study of laser wounding has identified at least three different mechanisms. The first Ca²⁺ response is extracellular Ca²⁺ entering the cytosol through plasma membrane microtears. Although this effect is driven here by laser-induced cavitation bubbles, similar cellular and tissue damage are inflicted simultaneously during naturally occurring trauma wounds (78, 79). In fact, pulsed laser ablation is similar to a localized puncture accompanied by a wider crush injury. In either case, the Ca²⁺ influx through microtears is effectively a single-cell response to cell-level damage; it becomes a tissue-level response as high cytosolic Ca²⁺ spreads to neighboring cells by diffusion through gap junctions. This so-called first wave thus involves a second distinct mechanism by which surrounding undamaged cells experience a rise in cytosolic Ca²⁺. Importantly, this expansion of a single-cell wound signal to neighboring cells suggests a continuum between single-cell and tissue-level wound responses.

Finally, if the region of ablated cells is large enough, an unknown signal initiates another wave of cytosolic Ca²⁺ increases, which we term the second expansion. The unknown signal driving this wave represents a third mechanism governing Ca²⁺ increases after wounding. Because this third mechanism occurs only after a discontinuity or hole appears in the epithelium, and not simply after cellular microtear damage, this signal may arise from cell mechanics, or cell lysate or dying cells around the primary wound margin. It is noteworthy that separate, but overlapping Ca²⁺ signals emanate from the wounded region due to both cell-level damage and tissue-level damage, albeit by different mechanisms. Our results suggest the interesting possibility that signaling from many types of damage may converge on increasing cytosolic Ca²⁺ levels to regulate both single-cell and multicellular wound-healing programs.

Conclusions

Wounds created via laser ablation contain single-cell damage and tissue damage, similar to naturally occurring puncture or crush wounds. Each damage mechanism drives its own Ca²⁺ signal dynamics. Single-cell damage arises from cavitation-induced, plasma membrane microtears, and results in direct Ca²⁺ influx from the extracellular environment followed by diffusive expansion into neighboring cells. Tissue damage in the form of a primary wound at the ablation site results in a delayed Ca²⁺ wave that expands well beyond the cavitation bubble footprint. This second wave occurs more frequently after larger wounds. Our kinematic analysis narrows and informs the search for Ca²⁺ wave initiation and propagation mechanisms by showing that both expansion events fit diffusive models that are consistent with previously observed rates of Ca²⁺ wave propagation in vivo, but are driven by different signals. Finally, this laser-wounding model exhibits exaggerated, but controllable single-cell damage and may be useful for future investigations at the intersection of single-cell wound healing and tissue wound healing.

Author Contributions

E.K.S., A.P.-M., and M.S.H. designed the study. E.K.S., W.E., and A.K.J. performed experiments and collected data. W.E., A.S., Y.Z., and M.S.H. developed analytical tools. E.K.S., A.S., W.E., Y.Z., and M.S.H. analyzed data and made figures. E.K.S., A.P.-M., and M.S.H. wrote the article. E.K.S., A.S., A.P.-M., and M.S.H. edited the article.

Editor: Ruth Baker.

Supporting Material

Movie S1. There Are Two Distinct Calcium Expansion Events upon Laser Ablation in the Notum

GCaMP3 was expressed in the pnr region of the notum using the Gal4-UAS system. Upon wounding at 0 s, intracellular calcium levels rise (1 s). Calcium levels then rise in neighboring cells (3–16 s) before fading again (18–31 s). A second, delayed calcium expansion event is then observed (starting at 33 s) that spreads farther than the first. After this expansion event, calcium expands into asymmetric flares (starting at around 63 s), which propagate around the wound for the duration of the movie (228 s).

Movie S2. Cells around the Wound Depolarize upon Laser Ablation in the Notum

The genetically encoded voltage indicator, Arclight, was expressed in the pnr region of the notum using the Gal4-UAS system. Before wounding, Arclight fluorescence is high in the notum. Upon wounding (at 0 s), cavitation-induced microtears allow cells to depolarize, which causes a conformational change in Arclight that reduces its fluorescence. The darkened area corresponds to the region of depolarization and the region of microtears; this dark region is obvious by 60 s after wounding.

Movie S3. Cells around Laser-Induced Wounds in the Notum Repolarize

The genetically encoded voltage indicator, Arclight, was expressed in the pnr region of the notum using the Gal4-UAS system. After wound-induced depolarization, cells repolarize over the course of 10 min, indicating they survive and repair cavitation-induced damage. To prevent photobleaching, scans were taken every 10 s.

Movie S4. A Nonpermeable Dye Enters Cells upon Laser Ablation in the Wing Disk

Wild-type wing disks were dissected and mounted in FM 1-43, a cell-impermeant lipophilic dye, which fluoresces on binding lipid membranes. Upon wounding in Ca⁺⁺-free PBS at 0 s, the cavitation bubble creates microtears in the plasma membrane and allows the dye to enter cells. The internalized dye binds the inner leaflet of the plasma membrane, leading to an increase in fluorescence in the region of microtears. This increase is gradual, but obvious by the end of the movie (525 s).

Movie S5. Cytosolic Calcium Levels Fluctuate around the Wound for >30 min after Wounding

Upon wounding (at 0 s), cavitation-induced microtears allow extracellular calcium to enter cells in the footprint of cavitation. Intracellular calcium levels then briefly rise in neighboring cells (0–20 s) before fading again (20–40 s). At ∼45 s after wounding, the high-calcium region undergoes a second expansion. This calcium expansion event spreads beyond the footprint of cavitation before breaking into asymmetric flares (∼100 s). The high-calcium region then fluctuates, expanding and contracting, for at least 30 min after wounding, even as the wound begins to close. Photobleaching contributes to the loss of signal intensity over time, and the movie gradually shifts out of focus and is manually refocused at 995, 1213, and 1481 s.

Movie S6. Knocking Down Gap Junctions Blocks the First Expansion and Modifies the Second

UAS-Inx3 RNAi is expressed in the pnr domain of the Drosophila notum using the pnr-Gal4 driver. This knocks down gap junctions and blocks the first postwound expansion of the high-calcium region. The first expansion is thus dependent on intercellular diffusion through gap junctions. The second delayed expansion still occurs, but it appears “spotty” and does not have a smooth wavefront. The second expansion thus relies on gap-junction communication to coordinate cellular responses, but such communication is not absolutely required for the signal to spread, suggesting a primary role for diffusion through the extracellular space. The same effects were observed with UAS-Inx2 RNAi.

Movie S7. Summary of Cavitation-Induced Calcium Release and Calcium Expansion Events

Animation shows a cartoon cross section of the epithelium (top) and a corresponding en face image (bottom) over the course of laser-induced wounding, on three timescales. First, in “Ablation and Cavitation,” the laser (purple) acts over nanoseconds to induce a primary wound (black spot) and a cavitation bubble (black circle) that expands and collapses over microseconds; the dashed circle indicates the maximum footprint of the cavitation bubble. Second, in “Primary Ca++ Influx,” calcium (white) enters cytosol through microtears over the course of milliseconds. Third, in “Two-Stage Signal Expansion,” calcium expands over the course of seconds. The red and blue curves above the cartoon images represent concentration profiles of two hypothesized diffusible signals. The red signal rises immediately after ablation; the blue arises 60 s later. Diffusive spread of these signals triggers calcium influx wherever either signal is above its threshold (dashed horizontal lines; arrowheads mark point where each signal concentration equals its threshold). These signals are hypothesized to drive the first (red) and second (blue) calcium signal expansions.