|Home | About | Journals | Submit | Contact Us | Français|
Most developmentally programmed cell death in metazoans is mediated by caspases. During Drosophila metamorphosis obsolete tissues, including the midgut and salivary glands, are removed by programmed cell death . The initiator caspase Dronc and its activator Ark are required for the death of salivary glands, but not for midgut removal [2, 3]. In addition to caspases, complete removal of salivary glands requires autophagy . However, the contribution of autophagy to midgut cell death has not been explored. Examination of combined mutants of the main initiator and effector caspases revealed that the canonical apoptotic pathway is not required for midgut cell death. Further analyses revealed that the caspase Decay is responsible for most of the caspase activity in dying midguts, yet inhibition of this activity has no effect on midgut removal. By contrast, midgut degradation was severely delayed by inhibition of autophagy, and this occurred without a decrease in caspase activity. Surprisingly, the combined inhibition of caspases and autophagy did not result in an additional delay in midgut removal. Together, our results indicate that autophagy, not caspases, is essential for midgut programmed cell death, providing the first in vivo evidence of caspase-independent programmed cell death that requires autophagy, despite the presence of high caspase activity.
A conserved pathway of cell death involving Ark (the Drosophila Apaf-1 homologue) mediated activation of the caspase Dronc is required for most developmentally programmed cell death (PCD) in Drosophila, including removal of larval salivary glands [2, 3, 5-9]. Like salivary glands, larval midguts undergo steroid-triggered PCD, and larval midgut structures, including gastric caeca and the proventriculus, are rapidly destroyed . Midgut destruction is accompanied by DNA fragmentation, autophagy, and the induction of caspases and caspase regulators [2, 5, 10] suggesting that midgut and salivary gland PCD are regulated by similar mechanisms. However, previous analyses have revealed that larval midgut PCD occurs normally in both dronc and ark mutants [2, 3]. dronc and ark mutants also contain high levels of caspase activity in the midgut, suggesting that the canonical Ark/Dronc pathway is not essential for caspase activation in this tissue [2, 3]. To analyze this further, we generated ark; dronc double mutants, and found that these animals still showed normal larval midgut PCD, as analyzed by TUNEL and morphology to determine the contraction of gastric caeca (Figure S1). Given that dronc and ark mutants show inhibition of PCD of the salivary glands and other larval tissues, maternal contribution of these genes is unlikely to play a major role. Our data suggest that PCD can occur in the apparent absence of both Ark and Dronc, and that an Ark/Dronc-independent mechanism of effector caspase activation may be present in this tissue.
During apoptosis active Dronc subsequently activates the downstream effector caspases Drice and Dcp-1, resulting in the execution of cell death . Examination of drice and dcp-1 mutant midguts revealed that cell death occurs normally in absence of these main effector caspases (Figure S2). As Dcp-1 can function redundantly with Drice, dcp-1; drice double mutants were analyzed and shown to have normal midgut cell death (Figure S2). In addition, dcp-1; dronc drice triple mutants also undergo cell death as detected by TUNEL, with morphology similar to the midgut from wild type animals (Figure 1A, B). Surprisingly, the dcp-1; dronc drice mutant maintained wild type levels of caspase activity (Figure S3). Thus midgut cell death and caspase activation can occur in the absence of the main initiator and effector caspases, indicating that an alternative caspase may be required.
Of the remaining caspases in Drosophila, there are two putative long prodomain caspases, Dredd and Strica . Dredd plays a key role in regulation of the innate immune response and is not involved in PCD . Strica has been shown to function redundantly with Dronc to regulate PCD during oogenesis . strica null mutants are viable and underwent normal midgut cell death (Figure S2). To determine if Strica plays a role in regulating midgut PCD redundantly with dronc, strica; dronc double mutants were analyzed, but found to undergo normal midgut removal (data not shown). strica; dronc drice triple mutant larval midguts contained high levels of caspase activity with TUNEL staining and morphology indicating that they are undergoing cell death (Figure 1C; Figure S3). On the other hand, salivary glands from the dcp-1; dronc drice and the strica; dronc drice triple mutants lacked any significant TUNEL at the normal time of histolysis, consistent with the role of Dronc in salivary gland PCD (Figure S2). Our data indicate that caspase activation and midgut PCD can occur in the absence of the main initiator and effector caspases, and that another caspase is active in the midgut.
Similar to other effector caspases, the Drosophila caspases Decay and Damm lack a long prodomain, and display caspase activity on both DEVD and VDVAD substrates [14, 15]. To ascertain which effector caspases may be contributing to caspase activity in the midgut of late third instar larvae and white prepupae, expression of all effector caspases was examined by qPCR. As expected , drice transcript levels increased in dying midgut whereas the levels of dcp-1 and damm, which were very low, remained unchanged (Figure 2A). Interestingly, decay expression in the midgut was relatively high (Figure 2A), indicating that drice and decay are most abundant effector caspases in midgut undergoing PCD. Given that drice mutants do not show an effect of midgut PCD role of decay was further investigated.
We used RNA interference (RNAi) to knockdown decay expression in the midgut using the midgut driver NP1-GAL4 and a significant decrease in decay transcripts, when compared to the wild-type levels, was clearly evident (Figure 2B). Since high levels of caspase activity were seen in all other caspase mutants, the effect of decay knockdown (decay-IR) in midgut caspase activity was examined. Interestingly both decay-IR alone and decay-IR; dronc drice had similarly large reductions of caspase activity (Figure 2C). Inhibition of caspase activity in the midgut by expression of the baculovirus p35, that inhibits most caspases except Dronc, resulted in a dramatic reduction in caspase activity (Figure 2C). Furthermore the combined decay knockdown and p35 expression abolished any detectable caspase activity in the midgut when assayed on DEVD (Figure 2C), or other commonly used caspases substrates, including VDVAD, IETD and LEHD (data not shown). These results suggest that Decay is the primary contributor of caspase activity in dying midguts.
To determine if reduced caspase activity has an effect on the timing of midgut histolysis, the morphology of decay-IR, decay-IR; dronc drice, p35 and decay-IR p35 midgut was examined. At +4 hr relative to puparium formation (RPF) midgut histolysis has initiated with contraction of the gastric caeca, and sections from pupae at +12 hr RPF revealed condensed midgut similar to wild type (Figure 2D-H). These data indicate that, despite Decay being responsible for the majority of the high levels of caspase activity present in dying midgut, reducing or inhibiting this activity had no significant effect on the PCD in this tissue.
Given the transcriptional upregulation of many pro-apoptotic genes, including dronc and ark, in response to ecdysone during midgut histolysis [17, 18] (Figure S4), it was surprising that caspases are not required for cell death. It is also intriguing that the main caspase in midgut is Decay, which differs from the salivary gland which requires Dronc and Drice activity, and unlike dronc, decay is not induced in response to ecdysone . Despite the high levels of caspase activity, inhibition of this activity by decay-IR and p35 had no significant affect on midgut cell death raising questions about the role of caspase activity in midgut PCD. The presence of active Decay in the absence of dronc and ark suggests that Decay may be activated by a yet unknown mechanism independent of Dronc.
The function of autophagy as a survival response to nutrient deprivation is highly conserved and in Drosophila, autophagy is induced in response to starvation during oogenesis and in the larval fat body [19, 20]. Contrary to its role in survival, autophagy is induced in dying larval salivary glands, and inhibition of autophagy results in persistent salivary glands . The complete removal of the salivary glands requires both caspases, including the activation of pro-apoptotic genes that regulate caspase activity, as well as autophagy [4, 21]. The combined inhibition of caspases and autophagy results in a greater inhibition of tissue degradation than either pathway alone . This suggests that in the salivary glands, caspases and autophagy are both required for PCD. Expression and proteomic screens have revealed that many of the autophagy (Atg) genes are up-regulated during salivary gland PCD [22, 23]. Our data show that several Atg genes are also up-regulated in dying midgut (Figure S5). Given the role of both autophagy and apoptosis for complete histolysis of salivary gland, the expression of Atg genes in the midgut and the finding that caspase activity is not required in dying midgut, we explored the contribution of autophagy to midgut PCD.
The occurrence of autophagy during midgut PCD was examined using the pGFP-Atg8a autophagosome marker, which results in expression of Atg8a fused to GFP from the endogenous promoter. A characteristic of autophagy is the formation of the autophagosome, and this can be monitored by the association of GFP-Atg8a with autophagosomal membranes observed as GFP puncta. Prior to midgut cell death in early third instar larvae, very few GFP-Atg8a puncta were detected in the gastric caeca and midgut (Figure 3A). Following puparium formation a dramatic increase in GFP-Atg8a puncta was observed in the midgut and gastric caeca (Figure 3A). Given that autophagy is induced as the midgut undergoes cell death, the effect of inhibition of autophagy on midgut cell death was examined using Atg mutants as well as knockdown of Atg genes previously shown to be important for autophagy . Morphological analyses of Atg2 and Atg18 mutant midguts at +4 hr RPF revealed a delay in midgut cell death, with the persistence of the gastric caeca (Figure 3C, E). Similar morphological defects were observed in midguts following knockdown of Atg1 and Atg18 (Figure 4 B, E). Histological analysis of +12 hr RPF pupal sections from Atg2 and Atg18 mutants and Atg1 and Atg18 knockdown showed a dramatic delay in midgut condensation, including persistent gastric caeca and proventriculus structures (Figure 3C, E and Figure 4B, E). Significantly, the localisation of GFP-Atg8a revealed that autophagy is greatly reduced in Atg2 and Atg18 mutants (Figure 3C, E and Figure S6). Thus inhibition of autophagy leads to a delay in midgut removal indicating a key role for autophagy in midgut PCD.
We examined if autophagy and caspases may play an additive role in midgut PCD. Morphological analyses of Atg1-IR and Atg18-IR revealed a delay in midgut histolysis +4 hr RPF, with extended gastric caeca being present (Figure 4B, E). However, these Atg1-IR and Atg18-IR phenotypes were not enhanced with the addition of decay-IR (Figure 4C, F), even though Atg1, Atg18, and decay all exhibited significant decrease in transcript levels compared to the wild-type levels by qPCR (Figure S7). Furthermore, histological analysis of +12 hr RPF pupal sections from knockdown of Atg1 and Atg18, showed a dramatic delay in midgut histolysis that was not affected by the knockdown of decay (Figure 4B, C, E, F). This was surprising, as complete histolysis of salivary glands requires both caspases and autophagy, indicating that midgut histolysis is regulated by a mechanism distinct from salivary glands. Autophagy was inhibited by knockdown of Atg1 and Atg18 as observed by markers for autophagy including, microtubule-associated protein light chain 3 (LC3)/Atg8 fused to GFP, (UAS-GFP-LC3), and Atg5 fused to GFP (UAS-GFP-Atg5) [24, 25] (Figure 4 and S8). Furthermore, induction of autophagy occurred normally when caspase activity was inhibited by either decay RNAi, p35 expression (Figure 4A, D, G) and the combined decay-IR p35 (data not shown), indicating that in this tissue, caspase activity functions independently of autophagy and is not required for cell death. Significantly, the inhibition of autophagy and caspase activity had no greater affect on midgut histolysis than inhibition of autophagy alone.
Our data show that despite the high levels of caspase activity in the Drosophila larval midgut during PCD, caspases do not have a significant function in midgut degradation. Instead, autophagy is essential for the proper removal of the midgut. Consistent with the role of autophagy in midgut removal, previous studies indicate that Atg7 mutants exhibit a slight delay in midgut histolysis . While it is difficult to understand why Atg7 loss-of-function mutants have a milder phenotype than Atg1, Atg2 and Atg18 in the midgut, it is worth noting Atg7 mutants are viable while these other Atg mutants are all lethal during development. These results suggest at least two possibilities; either a second gene acts redundantly with Atg7, or that Atg1, Atg2 and Atg18 may be pleiotropic and influence processes in addition to autophagy. Future studies should resolve these important differences in the function of different Atg genes.
Autophagy appears to have tissue specific roles depending on cell and tissue context, and can promote either cell survival or cell death. There has been considerable debate over the role of autophagy in cell death and recent studies have proposed that autophagy can occur in parallel to apoptosis, that autophagosome formation is downstream of caspase activation, or that autophagy can trigger apoptosis [27, 28]. Given the controversy over the role of autophagy in cell death, our study provides the first evidence that during midgut histolysis, autophagy is a key regulator of cell death.
Detailed Materials and Methods are provided as Supplemental Information. Briefly, all fly protocols, TUNEL, caspase assays and qPCR were essentially as described [29, 30]. For histological examination pupae were staged +12h RPF, fixed in FAAG (85% ethanol, 4% formaldehyde, 5% acetic acid and 1% glutaraldehyde) prior to paraffin embedding, sectioning, and staining [2, 4].
This work was supported by the National Health and Medical Research Council of Australia and the NIH. We thank Andreas Bergmann, Bruce Hay, Kim McCall, Tom Neufeld, the Australian Drosophila Research Support Facility, Vienna Drosophila RNAi Center, and Bloomington Drosophila Stock Center for Drosophila stocks and DNA constructs, Katherine Adriaanse and Tina Fortier for maintaining Drosophila stocks, and Christopher McPhee for technical support.