Crosstalk between Dpp and Tor signaling coordinates autophagy-dependent midgut degradation

The majority of developmentally programmed cell death (PCD) is mediated by caspase-dependent apoptosis; however, additional modalities, including autophagy-dependent cell death, have important spatiotemporally restricted functions. Autophagy involves the engulfment of cytoplasmic components in a double membrane vesicle for delivery to the lysosome. An established model for autophagy-dependent PCD is Drosophila larval midgut removal during metamorphosis. Our previous work demonstrated that growth arrest is required to initiate autophagy-dependent midgut degradation and Target of rapamycin (Tor) limits autophagy induction. In further studies, we uncovered a role for Decapentaplegic (Dpp) in coordinating midgut degradation. Here, we provide new data to show that Dpp interacts with Tor during midgut degradation. Inhibiting Tor rescued the block in midgut degradation due to Dpp signaling. We propose that Dpp is upstream of Tor and down-regulation promotes growth arrest and autophagy-dependent midgut degradation. These findings underscore a relationship between Dpp and Tor signaling in the regulation of cell growth and tissue removal.


Introduction
Programmed cell death (PCD) is essential for animal life and in most contexts is mediated by caspasedependent apoptosis 1 . Multiple additional modes of PCD have recently come to light 2 . One such modality is autophagy-dependent cell death, which plays important spatiotemporal restricted functions [3][4][5] . Several contextspecific examples of autophagy in cell death have been identified in Drosophila including the removal of obsolete larval tissues, including the midgut, during metamorphosis 3,5,6 . The larval midgut is a large tissue with anterior appendages called gastric caeca and in response to a pulse of the steroid hormone ecdysone at the larval-pupal transition, midgut PCD initiates with the initial contraction of the gastric caeca and the condensation of the gut 7,8 . Degradation of the larval midgut does not require caspase-dependent apoptosis but occurs by an autophagy-dependent cell death mechanism 6 . Autophagy is important for normal development, cellular homeostasis, metabolism, cell growth, and cell death 9 . Basal levels of autophagy are required under growth conditions to maintain cellular homeostasis, and in response to various stress and extracellular cues high levels of autophagy are induced. The induction of autophagy occurs in response to upstream signaling pathways that converge on Target of rapamycin (TOR) kinase, as part of a multi-protein complex TORC1 10 . In the presence of nutrients and growth signals TORC1 activity negatively regulate autophagy phosphorylating and inhibiting Atg1/Unc51-like kinase 1 (Ulk1) complex activity 11 . Under growth-limiting conditions such as starvation, TORC1 is no longer active enabling autophagy induction by Atg1 activation promoting the initiation of autophagosome formation 12 .
Degradation of the Drosophila larval midgut is triggered by an increase in the steroid hormone ecdysone. In addition to the hormonal cue, down-regulation of growth signaling and TORC1 activity precedes autophagydependent midgut degradation 13,14 . Similar to conditions of nutrient limitation where TORC1 inactivation promotes autophagy induction, ablation of Tor and raptor (but not TORC2 component rictor) promotes premature autophagy-dependent midgut degradation 14,15 . While TORC1 is critical for autophagy regulation, the interplay with the signals upstream of Tor in the regulation of autophagy-dependent cell death remains poorly understood. In recent studies we identified Decapentaplegic (Dpp), the Drosophila bone morphogenetic protein/ transforming growth factor β ligand, in the regulation of autophagy-dependent midgut degradation 16 . To understand the crosstalk between these pathways in regulating autophagy-dependent midgut PCD, in this report we have investigated epistasis between the Dpp pathway and Tor.

Results and discussion
Dpp expression prevents autophagy and midgut degradation A complex interplay between hormonal cues and growth signaling pathways is important for the initiation of autophagy-dependent cell death. To dissect out the regulatory mechanisms we identified dpp as a novel regulator of autophagy-dependent PCD 16 . Expression of Dpp in the midgut using the NP1-GAL4 driver resulted in enlarged midguts that do not contract like the control midguts (Fig. 1a). These animals fail to undergo metamorphosis and die as late third instar larvae 16 . The Thickveins (Tkv) receptor is required for Dpp signaling and ligand independent signaling can be achieved by expression of a constitutively active receptor Tkv Q253D (Tkv ACT ) 17 . Similar to Dpp, expression of Tkv ACT using NP1-GAL4 resulted in enlarged midguts and larval lethality (Fig. 1a).
Induction of autophagy results in association of Atg8a with autophagosomal membranes that can be observed as puncta. To examine autophagy flux in whole midguts we examined GFP-mCherry-Atg8a puncta formation. This revealed that similar to Atg1 knockdown, Dpp expression completely blocked induction of autophagy (Fig. 1b) 16 . While the control midguts showed strong induction of autophagy as indicated by the presence of both red and yellow puncta, thus quenching of the GFP signal in the autolysosome, both Atg1 knockdown and Dpp overexpression lacked any puncta (Fig. 1b). These results confirm that sustained Dpp signaling prevents autophagy and midgut size contraction, suppressing developmental PCD 16 .

Dpp signaling interacts with Tor signaling
Autophagy is maintained at basal level under growth conditions through upstream signaling pathways that converge on TOR kinase 12 . A key first step in autophagy induction is activation of a multi-protein complex containing Atg1, which is inhibited by active TOR. In previous studies we have shown that depletion of Atg1 and Atg18 in the midgut blocks autophagy and severely delays PCD 6 . Conversely, knockdown of Tor results in premature autophagy induction and midgut PCD 14 . To understand the mechanism(s) by which Dpp signaling regulates midgut removal we examined if there is an interaction between Tor and Dpp.
Initially, we examined the consequence of simultaneous expression of a dominant-negative Tor (Tor TED ) with Dpp and Tkv ACT expression. Interestingly expression of Tor TED was sufficient to significantly suppress both the Dpp and Tkv ACT midgut phenotypes (Fig. 1c, d). The block in gastric caeca contraction due to Dpp or Tkv ACT expression was rescued by expression of Tor TED to a size similar to the control. We then examined the consequence of simultaneous ablation of Tor with Dpp and Tkv ACT expression. We have previously shown the level of knockdown for two independent Tor RNAi lines, with one line (Tor RNAi ) providing greater knockdown than a second line (Tor RNAi#2 ) 14 . Consistent with Tor TED expression, the knockdown of Tor (Tor RNAi#2 ) significantly suppressed both the Dpp and Tkv ACT midgut phenotypes (Fig. 2a, b). This finding was further supported by the use of the independent RNAi knockdown line that provides stronger knockdown, which also showed that Tor knockdown is a suppressor of the Dpp and Tkv ACT midgut phenotypes (Fig. 2c, d). The expression of Dpp and Tkv ACT in the midgut causes a developmental arrest prior to the onset of metamorphosis; however, the knockdown of Tor rescued the block in midgut degradation by Dpp and Tkv ACT and promoted animal survival to a later stage of development (Fig. 2e). Strikingly, the Tkv ACT animals survived until +12 h Relative to puparium formation (RPF) and examination of the midgut at this later stage revealed the tissue had undergone contraction similar to the control (Fig. 2f). This suggests that the suppression of Tkv ACT by Tor knockdown was maintained. These data indicate an interaction between Tor and Dpp signaling pathways and are consistent with the down-regulation of Dpp signaling required for autophagy-dependent midgut removal 16 .
Given that the midgut phenotype due to Dpp expression could be suppressed by ablation of Tor, we examined if growth signaling and Tor levels were altered in response to Dpp signaling. Under growth conditions PI3K activates Tor thus inhibiting autophagy and down-regulation of PI3K leads to Tor inactivation promoting autophagy 12 . Growth signaling can be monitored by PI3K activity through the localization of phosphorylated Akt to the cell cortex, owing to its interaction with PIP3, and its subsequent phosphorylation that is required for downstream signal transduction 13 . To further investigate if Dpp expression perturbs growth signaling in the midgut we examined the localization of phosphorylated Akt, compared with the cell cortex marker Dlg. In both control and Dpp-expressing midguts during late larval stages (−4 h RPF), phosphorylated Akt was detected at the cell cortex (Fig. 3a). Furthermore, there was no significant change in the Tor transcript levels either in the presence (Dpp and Tkv ACT expression) (Fig. 3b) or absence (expression of inhibitory Smad, Dad) of Dpp activity (Fig. 3c). Together the data suggest that in the presence of Dpp activity growth signaling is similar to the control and is not promoting increased Tor expression.

Tor knockdown restore autophagy in Dpp expressing midguts
Dpp expression prevents induction of autophagy and Tor ablation induces premature autophagy. To determine if the phenotypic rescue was due to induction of autophagy, we examined autophagy using Atg8a and Lyso-Tracker staining. While the Dpp and Tkv ACT midguts show very little Atg8a staining, when combined with reduced Tor levels Atg8a puncta can be observed (Fig. 4a).
Although not a direct marker of autophagy, LysoTracker staining has also been used to detect autophagyassociated lysosomal activity in the fat body and midgut 14,15 . Such staining showed that reduction of Tor levels in the Dpp and Tkv ACT midguts was sufficient to restore LysoTracker-positive vesicles indicating increased autophagy flux (Fig. 4b). Together, these findings indicate that reducing Tor levels is sufficient to restore autophagy flux in the Dpp and Tkv ACT expressing midguts.
To further investigate the effects of Dpp and Tor on autophagy we used transmission electron microscopy (TEM). The data showed that Dpp and Tkv ACT midgut cells have very few autophagosomes or autolysosomal structures (Fig. 5a, b). A large number of autophagosomes and autolysosomes could be identified when Tor was knockdown (Fig. 5a, b). This analysis also revealed that midgut cells with combined Tor knockdown and Dpp or Tkv ACT expression contained more autophagic vesicles compared to cells expressing Dpp or Tkv ACT alone (Fig. 5a, b). Quantitation of the number of autophagic vesicles and lysosomes is consistent with the rescue of midgut degradation and increase in autophagy markers (Fig. 5b). Together, these data indicate an interaction between Tor and Dpp pathways, whereby downregulation of Dpp signaling is required for autophagydependent midgut removal.
In addition to hormonal cues, the down-regulation of growth signaling is important for the induction of autophagy-dependent cell death in the midgut 13 . During larval development, down-regulation of Tor signaling in the midgut promotes autophagy and midgut removal 14 . Similarly, down-regulation of PI3K and Ras signaling are required for proper midgut removal 13 . In addition, Dpp plays an important role in autophagy-dependent midgut degradation, whereas other morphogens including Hh and Wg are not required. 18 Our data presented here indicate that induction of autophagy by Tor depletion rescues the effect of Dpp signaling. This supports our recent findings that Dpp signaling prevents midgut removal by blocking autophagy induction 16 . Together these findings establish new connections between Dpp and Tor signaling pathways during autophagy-dependent midgut cell death. It will be important to understand how Dpp and Tor signals are integrated with other growth

Larval staging and midgut morphology analysis
To stage larvae 0.05% bromophenol blue was added to food and wandering third instar larvae were transferred to a Petri dish lined with moist Whatmann paper to monitor for clearance of blue food in the gut 19 . The morphology of the midgut was examined from a minimum of 10 appropriately staged animals by dissection in phosphatebuffered saline (PBS), then fixed in 4% formaldehyde/ PBS, and imaged using a stereozoom microscope (Olympus, Tokyo, Japan). The size of the gastric caeca was measured from these images in Photoshop (Adobe, San Jose, CA, USA) using the magnetic lasso tool and histogram function to determine pixels in the area as described previously 13 .

Live imaging
To image fluorescently tagged GPF-mCherry-Atg8a midguts were dissected in PBS with Hoechst 33342 (Sigma-Aldrich), mounted in PBS, and imaged immediately without fixation using a Zeiss LSM 700 or 800 confocal microscope (Detmold Imaging Core Facility, SA Pathology, Adelaide, SA, Australia). For LysoTracker staining, midguts were dissected in PBS and transferred to staining solution containing 100 nM LysoTracker Red DND-99 and 1 μg/ml Hoechst 33342 in PBS and then incubate in the dark for 2-5 min at room temperature. Samples were then washed with PBS for 5 min, mounted in PBS, and imaged immediately without fixation. Quantitation of images was achieved using ImageJ to count the number of puncta per cell (with a size larger than 2 pixels).

Immunohistochemistry
Midguts were dissected in PBS, fixed in 4% paraformaldehyde in PBS for 20 min at room temperature, and blocked with 5% normal goat serum as described. Primary antibodies used were rabbit anti-GABARAP1 (referred to as

Confocal imaging
Confocal images were obtained at room temperature using a Carl Zeiss LSM 700 or Zeiss LSM 800 inverted confocal microscope (Zeiss Laboratories) with 405 nm (5 mW), 488 nm (10 mW), and 555 (10 mW) lasers and C Apo ×40/1.2 W DICII objective. The dual labeled samples were imaged with two separate channels (PMT tubes) in a sequential setting. On the LSM 700, Zen gray was used to capture the images and on LSM 800 images were captured and Airyscan processed using Zen blue.
All images were then processed using Photoshop (Adobe).

Statistical analysis of data
The statistical analysis performed on the quantitation data was an ordinary one-way analysis of variance with Tukey's multiple comparisons test using Prism (GraphPad Software) and data are expressed as mean ± SD. Any images where the cell number could not be accurately determined were excluded from quantitation.