Targeted necrosis of Cry1Aa receptor-expressing cells in cultured imaginal discs incubated with Cry1Aa
To test whether Cry1Aa can induce cell death in cells overexpressing Cry1Aa receptors, we prepared the transgenic flies UAS-BmABCC2 and UAS-BtR175. We overexpressed each receptor in the wing pouch region, which eventually develops into adult wings, using the WP-Gal4 driver [20]. These flies had no morphological defects in the wing pouch of third instar larvae or in adult wings, suggesting that the ectopic expression of exogenous Bombyx mori receptors in these cells was not toxic. We dissected and cultured wing discs from third instar larvae and treated them with Cry1Aa ex vivo. We observed propidium iodide (PI)-positive dying cells in the wing pouch region when BmABCC2, but not BtR175, was overexpressed (Fig. 1a–c, Additional file 1: Figure S1). PI is a nucleic acid stain that acts as a marker for cell death, particularly for necrosis, as it enters the cells only when membrane integrity is disrupted. Since BmABCC2 is sufficient for Cry1Aa-induced cell death, we designated this receptor as CryR (Cry1Aa toxin receptor) for simplicity. We defined Cry1Aa-induced osmotic cell lysis as a form of necrotic cell death (or accidental cell death) that was not genetically regulated.
When CryR was driven by dpp-Gal4, which expresses at the midline region of wing discs, strong PI signals were observed in CryR-expressing cells when incubated with Cry1Aa (Fig. 2a, b). Simultaneous expression of both receptors further enhanced susceptibility to Cry1Aa-induced cell death (Fig. 3), similar to the synergistic effect observed in Sf9 cells [19]. Serial dilutions of Cry1Aa in the culture medium (concentrations 6.25–100 nM) revealed that 12.5 nM Cry1Aa induced weak but significant PI staining. This became stronger as the toxin concentration was increased; therefore, dose-dependent control of the strength of ablation might be possible (Additional file 1: Figure S2). These data indicate that the Cry1Aa/CryR ablation system can be used to induce conditional cell necrosis in Drosophila.
Rapid and selective cell necrosis by Cry1Aa in cultured imaginal discs and fat body
To investigate the precise time course of Cry1Aa-induced cell necrosis, we performed live imaging analysis of cultured wing discs from WP > GFP, CryR with 200 nM Cry1Aa (Fig. 4a, Additional files 2, 3 and 4). PI signals were first detected after around 30 min, and increased throughout the 90-minute incubation period. Interestingly, the morphology of the wing pouch region drastically changed just before PI-positive cells were first observed. We observed swelling of GFP-positive cells, which resulted in expansion of the wing pouch region. The WP-Gal4 driver was also expressed in a small population of leg discs, and these cells also died by cell swelling following Cry1Aa treatment.
Imaginal discs are proliferating tissues that seem to be relatively sensitive to cell death stimuli. Therefore, we investigated whether the Cry1Aa/CryR ablation system was applicable to non-proliferating tissues, such as the fat body, by live imaging analysis (Fig. 4b, Additional file 5). The fat body, a counterpart of mammalian liver and white adipose tissue, is composed of large, postmitotic, polyploid cells. We overexpressed CryR randomly in fat body cells using the flip-out clone technique, which simultaneously labels CryR-positive cells with GFP. When incubated with 100 nM Cry1Aa, CryR-expressing cells became PI-positive 20 min after toxin treatment, without any effect on neighboring CryR-negative cells. This suggests that the Cry1Aa/CryR system is a highly selective method for inducing necrosis with single cell resolution.
Cry1Aa induced necrosis independently of apoptosis, c-Jun N-terminal kinase (JNK) activation, or autophagy
One of the limitations of conventional cell ablation systems is their dependency on the cell’s ability to activate programmed cell death. In contrast, Cry1Aa directly forms a pore on the plasma membrane, suggesting that cellular context does not affect toxicity. To test this, we overexpressed p35, a baculoviral inhibitor of apoptosis that inactivates caspases, and tested whether Cry1Aa could still induce cell death (Fig. 5a, b). As expected, PI signals were not attenuated by p35 overexpression. In addition, Cry1Aa-induced cell death was not blocked by treatment with z-VAD-fmk, a pan-caspase inhibitor (Additional file 1: Figure S3A). We further confirmed that knock down of either dronc, an initiator caspase, or of pro-apoptotic RHG genes, did not inhibit cell death (Additional file 1: Figure S3B, C), indicating that apoptosis is not required for the Cry/CryR system.
JNK activation can also induce programmed cell death [21]. For example, Eiger, a tumor necrosis factor superfamily protein in Drosophila, kills cells in a JNK-dependent manner, at least partially by a different mechanism from apoptosis. We therefore overexpressed puckered (puc), which is a negative regulator of JNK. Similar to p35, puc expression did not affect Cry1Aa-induced cell necrosis (Fig. 5c).
Autophagic cell death is another type of programmed cell death [22, 23]. For example, during metamorphosis in Drosophila, degeneration of larval salivary gland or midgut requires autophagic components [24, 25]. When autophagy was inhibited by knock down of atg1 alone, or together with z-VAD-fmk, we still observed PI-positive dying cells induced by Cry1Aa (Additional file 1: Figure S3D-F). These data demonstrated that Cry1Aa-induced target cell ablation does not require genetic components for cell death.
Cry1Aa injection induced selective cell necrosis in vivo
To investigate whether the Cry1Aa/CryR system can be used for conditional cell ablation in vivo, Cry1Aa was fed to either developing larvae or adult flies through a Cry1Aa-containing diet. However, this failed to kill the animals even though CryR was overexpressed ubiquitously (da >CryR) or in gut enterocytes (NP1 >CryR), probably due to the instability of Cry toxin in the Drosophila medium and/or in the digestive tract. Therefore, we injected concentrated Cry1Aa directly into the body cavity of wandering third instar larvae (Fig. 6a). Injection of Cry1Aa into control flies had no apparent effect on viability or development, suggesting that there is little “off-target effect” on flies without exogenous CryR. In contrast, when Cry1Aa was injected into WP >GFP, CryR larvae, we observed selective cell necrosis in the wing pouch region (Fig. 6b, c). If cells in the wing pouch undergo successful ablation, injected larvae should become adult flies without wings. Strikingly, just a single Cry1Aa injection in third instar larvae resulted in adults without wings (Fig. 6d). This phenotype was only observed in flies expressing CryR and exposed to Cry1Aa, indicating conditional cell ablation by the Cry1Aa/CryR system in vivo.
We also tested whether our system is applicable to the developmental study of sensory organs by injecting Cry1Aa toxin into Neur >CryR third-instar larvae. Neur-Gal4 is expressed in sensory organ precursors (SOPs) in wing discs that become adult bristles. Cry1Aa injection during late-third instar larvae (6–12 h before pupal formation) resulted in a loss of bristles from epithelia (Fig. 7a). Although not all bristles are lost, the loss of bristles is probably due to differences in developmental timing as some SOPs such as the anterior scuteller bristle, which arise during later stages of development (0–6 h before pupal formation). Indeed, ablated macrocheates such as posterior scuteller bristle, or posterior drosocentral bristle are generated in early to middle-third larval stage (12–30 h before pupal formation) [26]. This suggested that Cry1Aa kills cells rapidly upon injection into larvae, and that Cry1Aa can be inactivated or removed from the hemolymph within a relatively short period after injection. Therefore, the Cry1Aa/CryR system could be useful for spatiotemporal cell ablation. Furthermore, a lack of melanization in adult epithelia at the original SOP positions implied that dying SOPs are eventually eliminated from the tissue rather than remaining in the epithelial sheet.
We observed a lethal effect of injected Cry1Aa on adult male flies with ubiquitous expression of CryR (Fig. 7b). Almost all CryR-expressing flies died within 18 h of the injection, while all control flies survived. In addition, Cry1Aa injection into Elav > CryR where CryR was overexpressed in neurons, resulted in an unsteady gait that eventually led to organismal death as early as 6 h post-injection (Additional file 6, Fig. 7b), suggesting that CryR-expressing neurons were damaged by Cry1Aa. To test whether Cry1Aa could permeate across the blood–brain barrier (BBB) and induce cell death in CNS, we performed PI staining upon Cry1Aa injection into larval hemolymphs. The injected brain from Elav > CryR larvae were positive for PI staining, but only at the surface of the brain tissue (Fig. 7c). BBB in larval brain is established by surface glia, which are distinct from cortex glia [27, 28]. To further validate that Cry1Aa cross the BBB, we overexpressed CryR by cortex glia driver, Nrv2-Gal4, and then investigated whether injected Cry1Aa into the hemolymph could induce cell death in these glia. Compared to the control (without Cry1Aa), we observed a large portion of PI-positive cortex glia, although PI-positive cells observed were not limited to GFP-positive cells (Additional file 1: Figure S4). Therefore, we believe that Cry1Aa penetrated the BBB, just as DT did in mice [8], although cells deep inside the tissue were not affected, suggesting that Cry1Aa/CryR system might be applicable for CNS.