Circulating blood cells function as a surveillance system for damaged tissue in Drosophila larvae

Blood Cell Dynamics in the Larval Open Circulatory System.

To visualize blood-cell dynamics, we constructed a Drosophila strain whose blood cells express a yellow fluorescent protein (YFP) driven by the blood cell-specific Peroxidasin (Pxn) promoter (11). Live imaging of these larvae (see schematic, Fig. 1A) revealed two populations of blood cells (Fig. 1 B–D). One was a stationary population of cells bound to the surface of internal organs and the barrier epidermis (19, 20). Most of these tissue-bound cells were sessile and appeared well anchored to their targets, although some were loosely tethered at just a single point around which they could swivel [Fig. 1G, supporting information (SI) Movie S1]. The other population was circulating cells and cell clusters that flowed posteriorly through the open body cavity at rates of 83 +/− 17.6 μm/second (Fig. 1 B and C, Movies S2 and S3). The internal organs provide barriers to flow that channel the circulating cells along certain predominant routes (Movies S2 and S3), mostly in the ventral part of the body. However, peristaltic motions of larval body-wall muscles could redirect circulating cells (Movie S4). Many cells reaching the larval posterior entered the heart and were recirculated anteriorly (Movie S5) at speeds of up to 3.2 mm/second (Fig. 1D, Movie S6), 40 × faster than the posterior-directed flow and similar to the rate of blood cell flow through vertebrate microvessels (21). There was some pooling of blood cells in the larval posterior (Fig. 1 E and F) (20), presumably due to a circulation bottleneck at reentry into the heart. Little interconversion was observed between the tissue-bound and circulating cell populations, although occasionally a tissue-bound cell detached from a tissue and transiently entered circulation before reattaching to another tissue (Fig. 1H, Movie S7).

Tissue Damage Induces a Rapid Inflammatory Response.

When early third-instar larvae are pinched with dissecting forceps, a gap is created in the epidermal sheet (Fig. 2 A and B), but the overlying cuticle remains intact (Fig. 2 E–H), preserving barrier function and preventing entry of microorganisms. Over the next 24 hours, the wound heals by spreading of the surrounding epidermal cells to close the wound gap (Fig. 2 C and D) (5). Immediately after wounding, the gap in the epidermal sheet was devoid of cells (Fig. 2B) but littered with cell debris (Fig. 2 F and H). Then, as early as 5 min after wounding, individual and small clusters of blood cells began to accumulate in the wound gap (Fig. S1 and data not shown). Over the next several hours, the number of cells increased until there were hundreds or, in some cases, thousands of blood cells covering one-third and sometimes more of the wound surface (Fig. 2 C and I), similar to what is seen at wounds in other insects (22–24).

Recruitment of Blood Cells to Wound Sites by Direct Capture from the Circulation.

Blood cells in Drosophila (9–13) and zebrafish embryos (25, 26) are highly motile and rapidly attracted to dead and dying cells by signals released by such cells. However, this is not the mechanism of blood-cell recruitment to larval wound sites. First, the small number of tissue-bound blood cells in the vicinity of the wound site was not sufficient to account for the large number of cells that rapidly accumulated at the wound during the inflammatory response (Fig. 2 B, C, and I). Second, blood-cell-specific expression of dominant-negative versions of the small GTPases Rac or Rho, which are thought to be universally required for cell migration (27), and which block blood cell recruitment to wound sites in embryos when expressed with the same Gal4 driver (11), had little or no effect on blood cell accumulation at larval wound sites (Fig. S2). Third, no local migration of nearby tissue-bound cells into wound sites was detected in the live imaging studies described below. Indeed, tissue-bound cells near the wound remained sessile and appeared completely unresponsive to the injury (see Fig. 3B′). The live imaging studies revealed that blood cells arrive instead by direct capture from circulation.

Blood cell infiltration of pinch wounds was imaged in Pxn>YFP transgenic larvae that also carried a Neuroglian–GFP (Nrg-GFP) transgene (28) that labels epidermal cell membranes to permit simultaneous visualization of the wound site borders (Fig. 3 and Movies S8–S11). Time-lapse and real-time imaging studies showed that within an hour or two after wounding individual circulating blood cells (Movie S8) or, more commonly, circulating clusters of cells (Fig. 3B, Movie S9) abruptly arrived at the wound site. Arriving cells directly docked on the exposed cuticle and debris in the wound gap (Fig. 3, Movies S8–S11). A blood cell capture event recorded in real time (Fig. 3A, Movie S10) showed that a circulating cell initially bound loosely to the wound surface and pivoted around a single tether point for several seconds before attaching more securely. When a cluster of cells docked at a wound site, cells in the cluster subsequently dispersed (Fig. 3B Bottom) apparently by flow-induced shearing of the attached cluster and reattachment of the separated cells to more downstream (posterior) positions in the wound gap (Movie S9). However, dispersal was not always necessary, as one recorded capture event involved direct docking in the wound gap of several large clusters or sheets of blood cells that almost completely filled the gap (Fig. 3C, Movie S11). Inhibiting larval peristalsis and mobility dampened wound-induced inflammation (Fig. S3 and Fig. S4), providing evidence that blood cell circulation facilitates the inflammatory response.

Wound-Adherent Blood Cells Undergo Morphological Changes and Appear Phagocytically Active.

Ultrastructural analysis showed that in contrast to the rounded morphology of sessile blood cells bound to the epidermis in unwounded larvae (Fig. 4A), blood cells bound to epidermal wound sites had a spread morphology (Fig. 4B). Cells bound at wound sites where more tightly adherent than tissue-bound cells as they persisted during dissection and staining whereas tissue-bound cells generally did not (data not shown). Cells bound at the wound were commonly arranged in loose clusters, with each cell extending many long cytoplasmic processes (Fig. 4 C and D), some of which contacted neighboring cells in the cluster (Fig. 4 C″ and D″). The layering of attached cells suggests that later capture events might occur by binding of blood cells to previously attached cells, as occurs in the response to parasitoid wasps (17, 18). Cell debris were present at most sites of contact between blood cells and the wound (Fig. 4 C and D′), and the debris were typically surrounded by cellular processes (Fig. 4C) or present within intracellular vesicles (Fig. 4C′), implying that the blood cells were phagocytically active (16, 29, 30).

Dispersal of Wound-Adherent Cells Requires Wound Closure.

As wounds healed and the epidermis spread to close the wound gap, the number of blood cells at the wound site gradually declined (Fig. 2 D and I). This decline was not likely due to programmed cell death, because no antiactivated caspase 3 immunoreactivity was detected in any blood cells up to 24 h after wounding (data not shown). When wound closure was blocked by expression in epidermal cells of a dominant-negative basket (Jun N-terminal kinase) transgene (31), blood cells persisted within the wound gap for at least 24 h (Fig. 4 E and F). Thus, spreading of the epidermis is necessary for turnover of blood cells at the wound site, suggesting that the spreading epidermal sheet may physically displace blood cells or induce their release from the wound site. Genetic ablation of the blood cells before wounding did not impair wound closure (Fig. S5).

Fly Stocks and Genetics.

The GAL4/UAS system (35) was used to drive expression of UAS transgenes in blood cells (Pxn-Gal4) (11) and larval epidermis (A58-Gal4) (5). w;Pxn-Gal4 8.1.1, UAS-GFP (Pxn>GFP) (11) was used to label blood cells and drive expression of UAS-rhoL.N25^DN, UAS-rac1.N17^DN, UAS-cdc42.N17^DN, which encode dominant-negative forms of the respective proteins (36). For TEM analysis, Pxn>GFP drove expression of a UAS-lacZ.NZ transgene (Bloomington) to allow identification of wound-associated blood cells in X-Gal- (5-bromo-4-chloro-3-indolyl-d-galactopyranoside) stained samples (see below). For live-imaging experiments, w, Nrg-GFP^G00305;Pxn-Gal4, UAS-2xeYFP larvae were used that carry a UAS-2xeYFP transgene (37) to label blood cells and a protein-trap GFP insertion (Nrg-GFP^G00305) in neuroglian (28), which expresses a GFP fusion protein localized to epithelial septate junctions. w;UAS-src-GFP, A58-Gal4/TM6b line was used to express a dominant-negative form of Basket (31) in GFP-labeled larval epidermis.

Wounding Procedure.

Pinch wounds were done as described (5), except that wounds were centered in a single abdominal segment, usually A4, -5, or -6. Larvae were maintained at 25°C except during wounding or live imaging, which was performed at room temperature.

Whole-Mount Immunofluorescence.

Dissection and immunostaining of larval epidermal whole-mount preparations were done as described (5). Primary antibodies were anti-Fasciclin III (38) (Developmental Studies Hybridoma Bank, 1:50), anti-Peroxidasin (39) (1:3,000), and anti-GFP (Molecular Probes, 1:500). Secondary antibodies (Jackson ImmunoResearch) were goat-anti-mouse Cy3 (1:1,000) and goat-anti-rabbit-FITC (1:300).

Live Imaging.

Third-instar w, Nrg-GFP;Pxn-Gal4, UAS-2xeYFP larvae were wounded and mounted dorsal side up on a glass slide with the anterior and posterior ends of the larvae taped down to varying degrees to prevent locomotion and constrain body peristalsis. Larvae were imaged on a Leica MZ16FA microscope using a Planapo 1.0× objective, and images were captured on a monochrome Leica DFC350FX digital camera. For real-time recording of larval circulation or heartbeat, frames were captured every ≈400 ms over a 4- to 40-s period and for time-lapse imaging, frames were captured every 2 or 5 min over a 2- to 4-h period. Images were obtained by using the Scope-Pro Advanced Acquisition plug-in, and image analysis was carried out with Image-Pro AMS ver. 5.1 Software (Media Cybernetics). The object-tracking tool was used to manually track specific blood cells in a time series. Velocities of tracked cells were calculated by using coordinates provided by the software.

Electron Microscopy.

For TEM, larvae were prepared as described (5), except that SPURR resin (Electron Microscopy Sciences) was used. Ninety-nanometer sections were observed in a JEOL JEM 1010 transmission electron microscope. Digital photos were obtained by using an AMT (Advanced Microscopy Techniques) imaging system. For SEM, dissected larvae were fixed in 3% glutaraldehyde/2% paraformaldehyde with 2.5% DMSO in 0.2 M sodium phosphate buffer, dehydrated in graded ethanol concentrations, and immersed in hexamethyldisilazane before vacuum drying, mounting on conductive carbon tabs (Electron Microscopy Sciences), and sputter-coating with gold to 0.1 kÅ. Samples were imaged with a Philips 525 scanning electron microscope and photographed with a Semicaps digital camera.

Blocking Larval Peristalsis and Mobility.

Two-hour immobilization of w,Nrg-GFP;Pxn-Gal4,UAS-2xeYFP larvae was accomplished by anesthetizing with 250 μl Flynap (Carolina Biological Supply) for 5 min and mounting securely on double-sided tape subsequent to wounding. Wounded and mounted larvae were maintained at room temperature under light humidity. Control groups were anesthetized with ether and put on food in a 25°C incubator.