Dying cells protect survivors from radiation-induced cell death in Drosophila

Abstract

We report a phenomenon wherein induction of cell death by a variety of means in wing imaginal discs of Drosophila larvae resulted in the activation of an anti-apoptotic microRNA, bantam. Cells in the vicinity of dying cells also become harder to kill by ionizing radiation (IR)-induced apoptosis. Both ban activation and increased protection from IR required receptor tyrosine kinase Tie, which we identified in a genetic screen for modifiers of ban. tie mutants were hypersensitive to radiation, and radiation sensitivity of tie mutants was rescued by increased ban gene dosage. We propose that dying cells activate ban in surviving cells through Tie to make the latter cells harder to kill, thereby preserving tissues and ensuring organism survival. The protective effect we report differs from classical radiation bystander effect in which neighbors of irradiated cells become more prone to death. The protective effect also differs from the previously described effect of dying cells that results in proliferation of nearby cells in Drosophila larval discs. If conserved in mammals, a phenomenon in which dying cells make the rest harder to kill by IR could have implications for treatments that involve the sequential use of cytotoxic agents and radiation therapy.

Figure 1. Activation of ban after clonal induction of cell death in wing discs.

The larvae were heat-shocked (hs) to induce FLP-mediated excision of intervening sequences to allow GAL4 expression from the actin promoter. After heat-shock, actin-GAL4 was kept repressed with tubulin-GAL80^ts to allow clones to form. A shift to 29°C inactivated GAL80 and de-repressed GAL4, which drove the expression of UAS-hid, UAS-rpr and the UAS-RFP clonal marker. (A, C, E) Wing discs from a control RFP only larva. (B, D, F) Wing discs from an RFP, hid, rpr larva. The discs were imaged live for RFP and GFP. (G) shows the timeline followed. GFP images in C and D were acquired and processed using identical parameters to allow comparison. RFP was weaker in Hid/Rpr clones than in control clones; therefore RFP images such as those in B were visualized with increased brightness. The mean GFP signal was quantified for each disc, normalized to the average for controls (RFP only) discs and shown on panels D and F. *p<0.001. ‘RFP’  =  hs-FLP/Y; GFP ban sensor/+;tub-GAL80^ts/Act>>GAL4, UAS-RFP. ‘RFP, Hid/Rpr’ carries the same transgenes and has UAS-hid, UAS-rpr instead of Y. ‘h’  =  hour or hours.

Figure 2. Changes in GFP ban sensor in discs with ptc-GAL4-driven cell death.

(A–D) Larvae carrying one copy each of ptc-GAL4, UAS-ds RNA against dE2F1 (ptc4>dE2F1^RNAi) and the GFP ban sensor were raised at 25°C for 72–96 hours and either maintained at the same temperature (A, B) or shifted to 29°C for 24 h (C, D). Wing imaginal discs were fixed and stained for cleaved Caspase 3 (A, C) or imaged live for GFP ban sensor (B, D). (E) shows the timeline followed. Embryos were collected for 4–6 hours. A =  Anterior; P  =  Posterior. Cell death was minimal at 25°C (A) but prominent after the temperature shift (C). Cells anterior to the ptc domain (black arrowheads) showed lower GFP compared to cells posterior (white arrowheads) in (D) but not in (B). The ratio of GFP fluorescence in A and P compartment for each disc was quantified using Image J, and the averages are shown in Table 1. Only the pouch regions were considered. (F–I) Larvae carrying one copy each of ptc-GAL4, UAS-hid, UAS-rpr, tub-GAL80^ts and the GFP ban sensor were raised at 25°C for 96 hours and shifted to 29°C to inactivate GAL80. (J) shows the timeline followed. Embryos were collected for 4–6 hours. Wing imaginal discs were fixed and stained for cleaved Caspase 3 (F, H) or imaged live for GFP ban sensor (G, I). Cell death was absent in ‘no GAL4’ controls (F) but robust after de-repression of GAL4 (H). ban sensor was bright immediately after de-repression of GAL4 for 12 hours (G, t1 time point), but was reduced throughout such discs 24 hours later (I, t2 time point). The mean GFP signal was quantified for each disc, normalized to the average for discs at t1 and shown on panels G and I. Dying cells in these discs sometimes, but not always, displayed elevated GFP (* in D and G); we do not know the reason.

Figure 3. Cells in discs with prior cell death are resistant to IR-induced cell death.

Wing imaginal discs were extirpated from third instar larvae 4(-IR) or 4000 R (+IR) of X-rays, fixed and stained for cleaved Caspase 3. Only the pouch regions are shown. A =  Anterior. P =  Posterior. (A–B) control discs without GAL4. (B) IR-induced caspase staining was similar in the A and P halves of the pouch in y¹w¹¹¹⁸ discs. Cells along the D/V boundary where ban sensor is high, reflecting low ban activity, were more sensitive to IR-induced death (yellow brackets). (C–F) discs from larvae with one copy of ptc-GAL4 and one copy of UAS-ds RNA against ATM (C,D) or dE2F1 (E, F). Areas with reduced IR-induced caspase activity (arrows) were found anterior to the domain of cell death induced by ptc-GAL4 (between lines). Cells along the D/V boundary were refractory to the protection (yellow brackets in F). (G–J) discs from larvae with one copy of ptc-GAL4 and one copy of UAS-rpr (G, H) or UAS-hid, UAS-rpr (I, J). These larvae also carried one copy of tub-GAL80^ts. The protected areas (arrows) included most of the A compartment and parts of the P compartment in (H). The protected area was even greater in (J). We note that death in the ptc domain in Hid/Rpr discs was wider in –IR discs than in +IR discs. These cells seemed to protrude from the disc after irradiation, which could explain the narrowing of the domain. (K) shows the timeline followed. Embryos were collected for 4–6 hours. The duration of incubation at 29°C was 24 hours for ATM or dE2F1 RNAi larvae and 12 hours for Rpr or Hid/Rpr larvae.

Figure 4. Cell death and protection in discs with clonal Hid/Rpr expression.

The larvae were generated as described for Figure 1 and irradiated with 4000R of X-rays after de-repression of GAL4. Wing imaginal discs were extirpated 4 h after irradiation and fixed and stained for DNA (A, E) and cleaved active Caspase 3 (B, F). RFP clonal marker (C, G) was used to mark the clonal boundaries in the images. (D, H) show merged images. (I) shows the timeline followed. * shows areas outside the clone that display robust caspase activation in RFP only discs (top row) but not in RFP, Hid/Rpr discs (bottom row).

Figure 5. The effect of ptc>dE2F1^RNAi is modified in different genetic backgrounds.

Wing imaginal discs from larvae carrying one copy each of ptc4-GAL4 and UAS-dsRNA against dE2F1 in different mutant backgrounds were fixed and stained for cleaved Caspase 3 (as indicated) or processed for TUNEL (F) at 4 h after irradiation with 4000R of X-rays. The discs were also stained for DNA. DNA stained images were used to locate the pouch (within the dashed line), the A/P boundary (solid vertical line) and the ptc domain (between vertical lines). UAS-GFP (green) in (D) confirmed the location of the ptc domain. Yellow brackets indicate apoptosis along the presumed D/V boundary in some panels. The experimental timeline in Fig. 3 K was followed, using a 24 h incubation at 29°C to de-repress GAL4. Embryos were collected for 4–6 hours. The caspase or TUNEL signal from A and P compartments of the pouch, exclusive of the ptc domain, were quantified from images such as these and shown in Table 1. (A–B) y¹w¹¹¹⁸ discs reproduced from Fig. 3B, for comparison. (C-P) All larvae carried one copy each of ptc-GAL4 and UAS-dsRNA against dE2F1. The larva in (C,D) also carried one copy each of tub-GAL80^ts and UAS-GFP. Additional genotypes are indicated next to the panels. banA =  UAS-banA.

Figure 6. Protection by Hid/Rpr-induced cell death is sensitive to ban gene dosage.

Wing imaginal discs from larvae carrying one copy each of ptc4-GAL4, UAS-hid, UAS-rpr and tub-GAL80^ts were fixed and stained for DNA (A, C, E) and cleaved Caspase 3 (B, D, F) at 4 h after irradiation with 0 (-IR) or 4000R (+IR) of X-rays. Additional genotypes were as indicated. (G) shows the timeline followed. In otherwise wild type background (D), areas outside the ptc domain showed reduced caspase staining reflecting protection in response to ptc4>Hid/Rpr (arrows). In contrast, the corresponding areas in ban/+ discs showed caspase activity (F, arrows).

Figure 7. tie interacts with ban to promote radiation survival.

Larvae were irradiated at 96±2 h after egg deposition with 4000R of X-rays. Percent eclosion was determined by counting full and empty pupal cases 10 days after irradiation. Error bar  =  1 SEM. (A) Homozygous tie mutants were more radiation sensitive than w¹¹¹⁸ controls (p<0.05). N = 106 to 494 pupae per genotype in at least three independent experiments. tie alleles were Df(3L)Exel9028 and transponson insertion alleles in Fig. S5: tie^e03394 or ‘03394’, tie^c098 or ‘co98’ and tie^e02680 or ‘e02680’. (B) tie and ban showed genetic interaction in radiation survival. The observed eclosion of double heterozygotes (tie/ban¹¹⁷⁰) is shown for each allele of tie. The % eclosion expected from an additive effect between ban¹¹⁷⁰/+ and tie/+ was computed from observed % eclosion for each allele and shown as solid horizontal bars above each tie allele. In all cases, the observed eclosion was lower than the expected. The difference between expected and observed were significantly different for the three tie alleles shown (X² test, p<0.05). N = 94–536 pupae in at least three independent experiments per genotype. (C) A UAS-ban transgene ‘banA’ rescued the radiation sensitivity of Df(3L)Exel9028 (‘Df’) and tie^e03394 heterozygotes. N = 359–634 pupae in at least six experiments per genotype.

Figure 8. tie is needed for IR-induced changes in GFP ban sensor.

(A–I) Third instar larvae were irradiated at 96±2 hr after egg deposition with 0R (-IR) or 4000R (+IR) of X-rays and wing discs imaged live 24 h later. Images were acquired and treated identically. The mean GFP signal was quantified for each disc using Image J, and the averages are shown for each genotype/treatment. Age-matched sensor only controls (i.e. wild type for tie) were included in each experiment (images not shown here but see Fig. S1). (A–C) Wing discs from homozygotes for Df(3L)Exel2098 that removes only tie (‘tie ’). N = 31, 28, 31 and 31 discs (left to right) in three independent experiments. (D–F) Wing imaginal discs from homozygotes for tie^e03394. N = 26, 27, 26 and 25 discs (left to right) in three independent experiments. (G–I) Wing imaginal discs from trans-heterozygotes tie^e03394/Df(3L)Exel2098. N = 29, 29, 30 and 30 discs (left to right) in three independent experiments. (J–L) Wing imaginal discs from Df(3L)Exel2098 homozygotes that carry one copy of a UAS-tie transgene. The data are from 5 discs per sample in two independent experiments. The larvae carried two copies of the GFP sensor in A-I and one copy in J-L. Error bar =  ±1SEM. p-values reflect a comparison of two samples at the ends of each bracket. (M) tie is needed for IR-induced changes in ban levels. Quantitative RT-PCR was used to quantify mature ban miRNA. The values were normalized to an internal a-tubulin control, and expressed as fold change from un-irradiated w¹¹¹⁸ controls (‘w2-’ set at 1). ‘-‘  =  no IR; ‘+’  =  4000R. ‘w’  =  w¹¹¹⁸; Df’  =  homozygotes of Df(3L)Exel2098 that removes only tie; ‘3394’  =  tie^e03394 homozygotes. *  =  difference compared to w2-. ** and *** =  significance compared to w2+. *  =  p <0.01; **  =  p <0.001; ***  =  p < 10⁻⁶. Student's t-test was used to determine significance.

Figure 9. A model for the Mahakali effect.

Known genetic determinants are shown.