Melanization of the Drosophila clot

In Drosophila larvae, the release of PPO involves both rupture of crystal cells (Fig. 1A-D) and dissolution of the crystals (Fig. 1C,D) that contain PPO (Rizki et al., 1985) (Fig. 1E, see also supplementary material Movie 1). To study the activation of PPO after wounding, we prepared hemolymph clots from noninfected larvae using a previously described ex vivo method (Bidla et al., 2005). We found melanization to be strongest in folds of the clot matrix and around cellular fragments that were most likely derived from lysed crystal cells (Fig. 2A-D). Similarly some plasmatocytes in the clot were melanized, although less intensely than cellular fragments, whereas cells outside the clot were not affected (Fig. 2E,F). As expected, melanization of the clot correlates with phenoloxidase (PO) activity, which was measured using a dot blot assay with a PO substrate (Sorrentino et al., 2002) (see Fig. 2B inset). Both melanization and PO activity were detected in preparations from wild-type animals, in spite of the presence of Spn27A in the hemolymph at normal concentration. By contrast, melanization was completely inhibited when Spn27A was overexpressed either ubiquitously [using the daughterless (da) GAL4 driver, Fig. 2G,H and Table 1], or in crystal cells [using the lozenge (lz) GAL4 driver, see Table 1]. The absence of melanization after ectopic expression of Spn27A correlated with the complete loss of PO activity measured with the dot blot assay (Fig. 2H, inset). This indicates that PO activity in the clot is regulated proteolytically and can be inhibited by Spn27A if the inhibitor is present at sufficiently high concentrations. Note that Spn27A overexpression did not block coagulation per se; clot fibers still formed (Fig. 2G,H) and most of the characteristic clot components were not affected in clots from larvae overexpressing Spn27A (data not shown). This is consistent with our previous finding that PO is not required for clot formation (Bidla et al., 2005).

Although systemic PPO activation is known to depend on microbial elicitors, plasmatocytes as well as crystal cell fragments that are melanized in the clot (Fig. 2) do not display any microbial patterns on their surface. Thus some other, nonmicrobial feature must activate PPO on these cells in the clot. Since our mutant analysis (see below) also suggested that the recognition of microbial elicitors does not play an essential role in clot melanization (Table 1), we hypothesized that PPO activation might be triggered by endogenous signals exposed on dying crystal cells and plasmatocytes. To test whether hemocytes in the clot die, we analyzed clot preparations using the TUNEL assay and found some cells were labeled (Fig. 3A,B). We confirmed this with an independent assay for apoptosis based on the exposure of phosphatidylserine on the cell surface (Fig. 3C-E). Just as in the TUNEL assay, some cells in clot preparations showed a positive reaction whereas others did not. At an early point after bleeding, a few cells were necrotic (Fig. 3C-E), and the number of these cells increased over time (data not shown). Apoptosis of hemocytes was also observed in clot preparations from lz mutants which lack crystal cells and hemolymph PO, demonstrating that programmed cell death is not restricted to crystal cells and is not a consequence of melanization (data not shown).

Drosophila JNK, small GTPases and Eiger are required for PPO release from crystal cells

Having established a reliable assay for clot melanization, we went on to screen Drosophila strains bearing mutations in known immune pathway genes for possible effects on clot melanization (Table 1, see also Fig. 6). Our screen included Bc and lz mutants, which lack hemolymph phenoloxidase as negative controls. In hml mutants where hemolymph does not coagulate (Goto et al., 2003) we still observed melanization, demonstrating that clotting itself is not required for PPO activation. An RNAi mutant in a PPO-activating protease (CG3066/MP2), which is expressed in crystal cells (Castillejo-Lopez and Häcker, 2005; Tang et al., 2006), had reduced clot melanization, supporting the notion that PPO in the clot is regulated by the same proteases as in the hemolymph (see also Discussion). By contrast, clot melanization was not altered in larvae bearing mutations in genes coding for known PRRs. This included mutants in a gram-negative binding protein (GNBP1) as well as three PGRPs, including PGRP-LE, which mediates the transcriptional induction that leads to systemic activation of PPO (Table 1) (Takehana et al., 2004). PPO activation in the clot thus appears to be independent of transcriptional induction. This observation was also in line with results from mutants in the two main immune signaling pathways; mutants in both the Toll (spz) and the imd/Relish pathway (Relish), as well as dTAK mutants (Boutros et al., 2002; Park et al., 2004) all showed normal melanization (Table 1). Conversely, an RNAi knockdown mutant in Drosophila JNK (basket, bsk), which acts downstream of dTAK during transcriptional activation, showed a complete lack of clot melanization (Table 1 and supplementary material Fig. S1A-E). In accordance with this, most crystal cells from bsk mutants had not ruptured even 1 hour after bleeding, although at this stage some cells lysed, probably owing to incomplete inhibition of bsk by the RNAi construct (supplementary material Fig. S1A). JNK pathway involvement in clot melanization was supported by finding puc lacZ staining in crystal cells, which indicates JNK pathway activity in these cells (supplementary material Fig. S1F,G). Also, RNAi knockdown of hemipterous (hep, dJNKK), which encodes a Drosophila Jun N-terminal kinase (JNK) upstream of Bsk (Glise et al., 1995), prevented clot melanization (Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Local activation of phenoloxidase in the Drosophila clot. Drosophila hemolymph clots were prepared as described (Bidla et al., 2005) and examined with phase-contrast (A,C,E) and bright field microscopy (B,D,F). Melanization is mostly restricted to cell fragments (asterisks) and folds in the clot matrix (arrowheads indicate clot folds, the arrows indicate plasmatocytes). Bars, 10 μm. (E,F) Plasmatocytes inside the clot (filled arrows) show melanization beginning on their surface. Plasmatocytes outside the clot (open arrow) are not melanized. (G,H) Overexpression of Spn27A (da-GAL4 driven) leads to complete loss of melanization, although clot fibers are formed (visible in phase-contrast, also visible in bright field images G,H). The insets in B and H show PO activity analyzed with a dot-blot assay (see Materials and Methods).

Small Rho GTPases are known regulators of JNK (Johndrow et al., 2004), so we also tested mutations in genes coding for these. We found that larvae expressing a dominant active form of RhoA (RhoA.DA) showed no clot melanization (Table 1, supplementary material Fig. S2A,B). Consistent with the lack of melanization, mutant crystal cells appeared unable to rupture and contained undissolved crystals even after prolonged incubation (supplementary material Fig. S2C,D). Although heat treatment to identify crystal cells (Galko and Krasnow, 2004) revealed the presence of comparable numbers of crystal cells in control and mutant larvae, melanization itself was much weaker in mutants, further supporting the idea that crystal cell activation is defective in these animals (supplementary material Fig. S2F,G, see also the Discussion for further details on the analysis of GTPase mutants). To test whether the TNF homolog Eiger, which is known to act upstream of the JNK pathway during eye development (Igaki et al., 2002; Moreno et al., 2002) is also required for crystal cell lysis, we analyzed two strongly hypomorphic egr alleles (egr¹ and egr³) (Igaki et al., 2002) as well as transheterozygotes between the stronger allele and a deletion uncovering the same region. All egr mutants showed reduced clot melanization, with the strongest effects in the transheterozygotes (Table 1) in which we also observed defects in crystal cell rupture (Fig. 4A,B). Consistent with these histological observations, we found less PO activity in egr mutants using a photometric assay. Overexpressing Eiger in the mutant background compensated or even overcompensated this defect (Fig. 4C). Eiger overexpression in wild-type larvae showed that protein expression is sufficient in hemocytes but not in the fat body to induce melanization, since expression driven by he-GAL4, but not the fat-body-specific lsp2-GAL4 activated melanization (Fig. 4D).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Hemocytes in the clot show hallmarks of apoptosis. Bright field (A) and fluorescence (B) microscopy of a TUNEL assay performed on a clot preparation showing TUNEL-positive (filled arrows) and TUNEL-negative plasmatocytes (open arrows) in the clot. The border of the clot is indicated by arrowheads. (C-E) Phase contrast (C), annexin V (D) and propidium iodide (E) labeling of plasmatocytes 7 minutes after bleeding showing live (^*) as well as apoptotic (open arrow, only annexin V positive) and necrotic cells (solid arrow, reacting with both annexin V and propidium iodide).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Melanization defects in egr mutants. (A,B) an egr¹/def mutant crystal cell is shown 10 minutes (A) and 60 minutes (B) after bleeding. Crystals are still visible in A (arrows) and not in B without leading to visible melanization. Bar, 10 μm. (C,D) Melanization in egr mutants was measured photometrically. (C) Melanization is reduced in egr¹/def mutants compared to wild-type (wt) activity (n=3, ^*P=0.014). When Egr is overexpressed using he-Gal4 in an egr¹ homozygous background melanization is restored to levels above wild-type activity (n=3, ^**P=0.007 for comparison with the wild type). (D) Melanization in UAS-egr control larvae and in larvae overexpressing Eiger in the fat body (lsp2-Gal4>UAS-egr) or in hemocytes (he-Gal4>UAS-egr). Only expression in hemocytes leads to significant induction (n=3, ^**P=0.006). Note that the crosses were kept at 25°C. Data are mean ± s.d.

Since plasmatocytes in the clot show some hallmarks of apoptosis (Fig. 3), we also tested the influence of the viral apoptosis inhibitor p35 after ectopic expression in hemocytes and found strong reduction in clot melanization comparable with the strongest egr alleles (Table 1). To obtain in vivo evidence for a role of hemocyte death during PO activation and to exclude microbial contamination as a cause of melanization in ex vivo preparations, we expressed the apoptotic inducer Grim ectopically in larval hemocytes (Wing et al., 2001). Consistent with our previous results, induction of apoptosis led to the formation of melanotic aggregates that were confirmed to contain hemocytes (Fig. 5A,D,E). When Grim expression was induced using heat-shock driver lines, extensive melanization was observed throughout the animal (data not shown). Just as in our ex vivo experiments (Fig. 2G,H), the number and extent of the aggregates induced in hemocytes was regulated by Spn27A, since the severity of the melanotic phenotype was enhanced in a spn27A mutant background (Fig. 5B). Conversely, when Spn27A was co-expressed with Grim in hemocytes, surviving larvae showed almost no melanization (Fig. 5C). In addition, Grim-induced melanization was completely inhibited when p35 was co-expressed (data not shown), paralleling the reduced melanization in clots from larvae that expressed p35 in hemocytes in a wild-type background (Table 1).

Fly stocks

Flies were kept at 25°C in a 12 hour light/12 hour dark cycle on potato sucrose medium. Washed third instar larvae were used. The Drosophila strains used were: w¹¹¹⁸ (wild type), y¹ w^*; P{UAS-mCD8::GFP.L}LL5 (GFP-tagged mouse transmembrane protein CD8, which labels the cell surface), w^*; wg^Sp-1/CyO; ry⁵⁰⁶Sb¹P{Δ2-3}99B/TM6B,Tb⁺ (transposase stock), P{GawB}lz^gal4, P{UAS-GFP.S65T}T2, P{lArB}puc^A251.1F3 ry⁵⁰⁶/TM3, Sb¹ (JNK pathway reporter), Df(2R)stan1, P{neoFRT}42D cn¹ sp¹/CyO (deficiency that uncovers egr), w^*; P{UAS-p35.H}BH1, w^*; P{UAS-p35.H}BH2, Act5C-GAL4/TM6B, Tb¹, hml^f03374 and Bc¹ (Bloomington Stock Center); dTAK1 (Vidal et al., 2001), UAS-bsk-IR (Ishimaru et al., 2004) and daGAL4 (kindly provided by B. Lemaitre, Gif-sur-Yvette); UAS-spn27A (Ligoxygakis et al., 2002), GNPB-1^osi, PGRP-SA^seml and PGRP-LC^E12 (Gobert et al., 2003) (kindly provided by J.-M. Reichhart, Strassbourg); Hemese-GAL4, Hemese-GAL4 UAS-GFP.nls (Zettervall et al., 2004) (kindly provided by D. Hultmark, Umeå); PGRP-LE¹¹² (Takehana et al., 2004) (kindly provided by S. Kurata, Sendai); lz^r15/X^ÙX (kindly provided by M. Meister, Strassbourg); UAS-RhoA V14/CyO-GFP (Billuart et al., 2001), CG3066-IR (Castillejo-Lopez and Häcker, 2005), UAS-grim (Wing et al., 2001) and egr¹/egr¹, egr³/egr³, UAS-egr (Igaki et al., 2002) (kindly provided by M. Miura, Tokyo). Flies bearing a UAS-grim and UAS-egr insertion on chromosome three were generated by standard genetic techniques after crossing UAS-grim (Wing et al., 2001) and UAS-egr (Igaki et al., 2002), respectively to the transposase stock. UAS-grim, UAS-spn27A flies were obtained by recombining UAS-grim and UAS-spn27A (Ligoxygakis et al., 2002) on the third chromosome. Gal4>UAS experiments were performed at 29°C unless noted otherwise.

Histology

The preparation of clot samples was essentially as previously described (Bidla et al., 2005). Hemolymph from five larvae was incubated upside down for 30 minutes in a humid chamber at 25°C. The clot was then caught on a coverslip, and images were taken with a COOLPIX 4500 digital camera (Nikon) adapted to a Leitz Orthoplan microscope.

PO activation

For the measurement of PO activity by dot blots, 5 μl hemolymph was applied to a filter paper pre-soaked with 20 mM DOPA in phosphate buffer pH 6.6 (Sorrentino et al., 2002), incubated for 20 minutes at 37°C and briefly heated in a microwave to dry the paper. For photometric measurement of PO activity, 30 μl hemolymph from each strain was pooled on ice by quickly bleeding 10-12 larvae and withdrawing 6 μl hemolymph each time into 10 μl of insect Ringer with EDTA instead of Ca²⁺ (anticoagulant) (Scherfer et al., 2004). After pipette mixing, 8 μl of each hemolymph was aliquoted and activated at 25°C for 10 minutes. Optical density (OD) was read at 490 nm with a Vmax™ Kinetic Microplate Reader after adding 25 μl DOPA at 10 and 30 minutes. Activation of PO was measured as the relative change in OD. Experiments were repeated at least three times and sets of experiments at least twice. Data were compared using a two-tailed Student's t-test (see figure legends for further details).

Apoptosis assays

Apoptosis was detected by TUNEL and annexin V binding assays. In TUNEL, DNA fragmentation was detected in hemocytes in a hanging drop preparation of hemolymph (see above) using the Roche Diagnostics in situ cell death detection kit, with fluorescein, following the manufacturer's instruction. Annexin V binding was performed using the CLONTECH ApoAlert Annexin V-FITC Apoptosis kit. Hemolymph from five third instar larvae was bled into 100 μl drop of Drosophila Ringer's solution on a well of a ten-well multitest slide. After incubation for 7 minutes, attached hemocytes were rinsed three times with annexin V binding buffer and continued as recommended by the supplier with the slight modification that 1.5 times more annexin V and four times less propidium iodide were used.

β-Galactosidase activity

Three puc^A251-lacZ larvae were bled into 100 μl anticoagulant Ringer's buffer and incubated at room temperature for 90 minutes in a humid chamber followed by a 10-minute fixation in 4% paraformaldehyde. Hemocytes were then permeabilized in 0.5% Triton-X100 for 5 minutes and stained with 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) for 18 hours.