TORC1 modulation in adipose tissue is required for organismal adaptation to hypoxia in Drosophila

Animals often develop in environments where conditions such as food, oxygen and temperature fluctuate. The ability to adapt their metabolism to these fluctuations is important for normal development and viability. In most animals, low oxygen (hypoxia) is deleterious. However some animals can alter their physiology to tolerate hypoxia. Here we show that TORC1 modulation in adipose tissue is required for organismal adaptation to hypoxia in Drosophila. We find that hypoxia rapidly suppresses TORC1 signaling in Drosophila larvae via TSC-mediated inhibition of Rheb. We show that this hypoxia-mediated inhibition of TORC1 specifically in the larval fat body is essential for viability. Moreover, we find that these effects of TORC1 inhibition on hypoxia tolerance are mediated through remodeling of fat body lipid storage. These studies identify the larval adipose tissue as a key hypoxia-sensing tissue that coordinates whole-body development and survival to changes in environmental oxygen by modulating TORC1 and lipid metabolism.

INTRODUCTION 1 Animals often have to grow and survive in conditions where their environment fluctuates. For example, 2 changes in nutrition, temperature or oxygen availability, or exposure to toxins and stress can all impact 3 development. Animals must therefore adapt their physiology and metabolism in response to these 4 environmental challenges in order to ensure proper growth and survival 1,2 . 5 6 In most animals decreases in oxygen are particularly deleterious. Low oxygen (hypoxia) can lead to 7 rapid tissue damage and lethality, and oxygen deprivation is a hallmark of diseases such as stroke and 8 ischemia 3 . However, some animals have evolved to live in oxygen-deprived conditions and 9 consequently exhibit marked tolerance to hypoxia. For example, birds and aquatic mammals can 10 tolerate extensive periods of low oxygen without incurring any tissue damage 4,5 . Indeed, some animals 11 show quite remarkable levels of tolerance to oxygen deprivation: brine shrimp embryos have been 12 reported to recover from four years of continuous anoxia 6 , while the naked mole rat can survive up to 13 18 minutes of complete oxygen deprivation, a condition that kills laboratory rodents within about one 14 minute 7 . Understanding how these animals adapt their metabolism to low oxygen may shed light on 15 how to protect tissues from hypoxic damage in disease states. alter their physiology and metabolism to slow growth and development, and to promote survival. One 23 main regulator of these nutrient-regulated processes in Drosophila is the conserved TOR kinase 24 signalling pathway 13 . TOR exists in two signalling complexes, TORC1 and TORC2, with TORC1 being 25 the main growth regulatory TOR complex 14 . A conserved signalling network couples nutrient availability 26 to the activation TORC1 to control anabolic process important for cell growth and proliferation 14 . 27 Moreover, studies in Drosophila have been instrumental in revealing non-autonomous effects of TORC1 28 signalling on body growth. For example, nutrient activation of TORC1 in specific larval tissue such as 29 the fat body, muscle and prothoracic gland, can influence whole animal development through the 30 control of endocrine signalling via insulin-like peptides and the steroid hormone, ecdysone 9,10,15 . In 31 addition, TORC1 regulation of autophagy in the larval fat body is important for organismal homeostasis 32 and survival during periods of nutrient deprivation 16,17 . 33 34 Drosophila larvae are also hypoxia tolerant [18][19][20] . In their natural ecology, Drosophila larvae grow on 35 rotting food rich in microorganisms, which probably contribute to a low oxygen local environment. Even 36 in the laboratory, local oxygen levels are low at the food surface of vials containing developing larvae 19 . 37 Drosophila have therefore evolved metabolic and physiological mechanisms to respond to and thrive in 38 hypoxic conditions. However, compared to our understanding of the nutrient regulation of growth and 39 homeostasis, considerably less is known about how Drosophila adapt to low oxygen during 40 development. A handful of studies have shown that larval survival in oxygen requires regulation of gene 41 expression by the transcription factors HIF-1 alpha and ERR alpha, and the repressor, Hairy 21-24 . 42 Developmental hypoxia sensing and signalling has also been shown to be mediated through a nitric 43 oxide/cGMP/PKG signalling pathway 25,26 . 44 45 Here we report a role for modulation of the TOR kinase signalling pathway as a regulator of hypoxia 46 tolerance during Drosophila development. In particular, we find that suppression of TORC1 specifically 47 in the larval fat body is required for animals to reset their growth and developmental rate in hypoxia, and 48 to allow viable development to the adult stage. We further show that these effects of TORC1 inhibition 49 require remodelling of lipid droplet and lipid storage. Our findings implicate the larval fat body as a key 50 hypoxia-sensing tissue that coordinates whole animal development and survival in response to 51 changing oxygen levels. 52 53 RESULTS 54 1 Exposure of larvae to hypoxia slows growth and delays development 2 We began by examining the effect of exposing larvae to hypoxia on their growth and development. We 3 used 5% oxygen as our hypoxia conditions for all experiments in this paper. We allowed embryos to 4 develop in normoxia and then, upon hatching, larvae were either maintained on food in normoxia or 5 transferred to food vials in hypoxia chambers that were perfused with a constant supply of 6 5%oxygen/95% nitrogen. We found that hypoxia led to reduced larval growth rate and larvae took 7 approximately an extra two days to develop to the pupal stage (Fig 1a). We also found that the hypoxia-8 exposed animals had a reduced wandering third instar larval weight (Fig 1b) and reduced final pupal 9 size (Fig 1c). We found that exposure of larvae to hypoxia did not alter their feeding behaviour (Suppl 10 Fig 1), suggesting that the decreased growth rate was not simply due to a general reduction in nutrient 11 intake. Together, these data indicate that Drosophila larvae adapt to low oxygen levels by reducing their 12 growth and slowing their development. These data are consistent with previous reports showing that 13 moderate levels of hypoxia (10% oxygen) can also affect final body size 20 .
14 15 Hypoxia suppresses TORC1 signalling via TSC1/2. 16 The conserved TORC1 kinase signalling pathway is one of the main regulators of tissue and body 17 growth in Drosophila. TORC1 can be activated by dietary nutrients and growth factors such as insulin. 18 Mammalian cell culture experiments have also shown that hypoxia can suppress TORC1 activity 27-30 . 19 We therefore examined whether changes in TORC1 signalling play a role in adaption to hypoxia in 20 Drosophila larvae. We transferred third instar larvae from normoxia to hypoxia and then measured 21 TORC1 activity by western blotting using an antibody that recognizes the phosphorylated form of S6 22 kinase (pS6K), a direct TORC1 kinase target. We found that hypoxia led to a rapid suppression of 23 whole body TORC1 activity that was apparent within 10-20 minutes of hypoxia exposure (Fig 2a). This 24 suppression persisted when larvae were maintained in hypoxia for longer periods (48 hours, Suppl Fig   25   2a). We also examined how different levels of oxygen affected TORC1 activity. Third instar larvae were 26 transferred from normoxia to different levels of hypoxia (from 20-1% oxygen) for one hour and then 27 TORC1 activity measured by western blotting for phosphorylated S6K. We found that suppression of 28 TORC1 occurred at 5 and 3% oxygen but remained unchanged at higher (20 and 10%) or lower (1%) 29 levels (Fig 2b). We examined this further by exposing larvae to several different concentrations of 30 oxygen between 1 and 10%, and found that the range within which TORC1 was inhibited was between 31 2 and 6 % oxygen (suppl Fig 2C). These data indicate that larvae rapidly respond to hypoxia by 32 suppressing TORC1 signalling, and that this response occurs within a specific range of low 33 environmental oxygen rather than simply being triggered below a threshold level of low oxygen.

35
We next examined how hypoxia suppresses TORC1 activity. One of the main ways by which TORC1 is 36 activated is through a TSC1/2-Rheb signalling pathway 14 . Rheb is a small G-protein that binds to and 37 activates TOR kinase at lysosomes. TSC2 is a GTPase activating protein, and when bound to its 38 partner TSC1, it inhibits Rheb by converting it from its active GTP-bound state to an inactive GDP-39 bound state. Several diverse stimuli including nutrients, growth factors and hypoxia have been shown to 40 regulate TSC1/2 function and to control TORC1 activity in mammalian cell culture 14 . We therefore 41 explored a role for TSC1/2 and Rheb in the suppression of TORC1 kinase signalling during larval 42 hypoxia. We found that ubiquitous overexpression of a UAS-Rheb transgene (using daughterless-gal4, 43 da-gal4) was sufficient to prevent the hypoxia-mediated suppression of TORC1 signalling in larvae (Fig   44   2c). We also found that tsc1 null mutant (tsc1 Q87X ) larvae also were unable to suppress TORC1 45 signalling when exposed to hypoxia (Fig 2d). Together these data indicate that hypoxic exposure in 46 larvae inhibits TORC1 by TSC1/2-mediated suppression of Rheb.

48
Studies in mammalian cells have described how hypoxia can induce TSC-mediated TORC inhibition via 49 the classic HIF-1 alpha transcription factor. In this mechanism, HIF-1 alpha leads to upregulation of 50 REDD1, an activator of TSC1/2 31 . In Drosophila, the homolog of REDD1, Scylla, and its partner protein, 51 Charybdis, have been shown to inhibit TOR and suppress growth 32 . We therefore examined a role for 52 Sima (the Drosophila HIF-1 alpha homolog) and Scylla/Charybdis in larval hypoxia. However, we found 53 that sima mutants still showed a suppression of TORC1 signalling when exposed to hypoxia (Fig 2e). 54 Similarly, both scylla and charybdis mutants also showed a similar suppression of TORC1 signalling as 1 control larvae in hypoxia (Fig2f, Suppl Fig 2c). We also explored a potential role for AMPK in hypoxia-2 mediated TOR regulation. AMPK is activated under hypoxia in mammalian cell culture and can 3 suppress TORC1 signalling, in part by phosphorylating and inhibiting TSC2 28, 33,34 . However, when we 4 suppressed AMPK by expression of a Gal4-dependent AMPK inverted repeat transgene (UAS-AMPK 5 IR), we still saw that hypoxia exposure lead to an inhibition of TORC1 (Suppl Fig 2d). Together, our 6 data suggest that the rapid suppression of TORC1 signalling upon hypoxia exposure in larvae requires 7 TSC1/2 function but is independent of both HIF-1 alpha mediated transcription and AMPK activation. Suppression of TORC1 signalling in the fat body is required for adaptation to hypoxia 10 We next examined whether the suppression of overall TORC1 activity we observed was important for 11 animal adaptation and tolerance to hypoxia during Drosophila development. Our approach was to 12 genetically maintain TORC1 signalling in larvae exposed to hypoxia and then to examine the effects of 13 this manipulation on animal growth, development and survival. To do this, we used the ubiquitous 14 expression of UAS-Rheb with da-Gal4 since we found this condition led to larvae maintaining TORC1 15 activity under hypoxia (Fig 2c). We compared development in control (da>+) vs. Rheb overexpressing 16 (da>Rheb) animals that were grown throughout their larval period from hatching in either normoxia or 17 hypoxia. We first found that larval Rheb overexpression had no effect on overall survival to the pupal 18 stage in either normoxia or hypoxia (Fig 3a). We next examined developmental rate by measuring the 19 time to pupation. In normoxic conditions, we found that Rheb overexpression (da>Rheb) lead to a slight 20 increase in developmental rate compared to control animals (da>+, Fig 3b). When raised in hypoxia, the 21 da>+ animals had an approximately two-day delay to pupation, and this developmental delay was even 22 further exacerbated in da>Rheb animals. (Fig 3b). We also measured effects on overall body size at the 23 end of larval development. We found that da>Rheb animals exhibited an increase in both wandering 24 third instar larval weight (Fig 3c) and pupal volume (Fig 3d). These results are consistent with increased 25 growth caused by modest elevation of TORC1 signalling. However, we found that when raised in 26 hypoxia, the increase in size in da>Rheb animals was abolished (Fig 3c, d). Given that the da>Rheb 27 pupae required an additional ~2 days of larval development to reach the same size as da >+, this 28 indicates that the Rheb overexpressing animals actually had a reduced growth rate in hypoxia.

30
Finally, we examined how maintaining TORC1 activity during larval development in hypoxia affects 31 subsequent survival to adulthood. For these experiments, we maintained animals in either normoxia or 32 hypoxia throughout their larval period and then switched them to normoxia and monitored their 33 development. We first saw that animals carrying either the da>Gal4 (da>+) or UAS-Rheb (+>Rheb) 34 transgenes alone had no effect on viability in either normoxia or hypoxia (Suppl Fig 3a). We found that 35 both da>+ and da>Rheb animals grown in normoxia as larvae showed normal development to the 36 pharate adult stage. Similarly, da>+ animals grown in hypoxia as larvae also showed no significant 37 change in development to pharate adults. In contrast, da>Rheb animals that were maintained in 38 hypoxia during their larval period showed a marked lethality at the pupal stage with few animals 39 developing to pharate adults (Fig 3d). When we further examined adult eclosion, we again saw that 40 da>Rheb animals that were maintained in larval hypoxia showed almost complete lethality, but in this 41 case the da>Rheb animals raised in normoxia also showed a reduction in eclosion, albeit to a much 42 lesser extent than their hypoxia-raised counterparts. We repeated our Rheb overexpression 43 experiments with a second independent UAS-Rheb transgene and we observed similar, but slightly 44 weaker effects, where da>Rheb animals grown in hypoxia as larvae showed a significant decrease in 45 survival to adult stage compared to da>+ animals (Suppl Fig 3b). 46 47 Taken together, these experiments using ubiquitous expression of Rheb to maintain TORC1 signaling 48 indicate that suppression of TORC1 is required for larvae to reset their development and growth rate in 49 hypoxic environments, and for subsequent viable development to the adult stage.

51
The adaptation to hypoxia may reflect a cell-autonomous requirement for each cell to sense low oxygen 52 and inhibit TORC1 to promote overall development and survival. Alternatively, hypoxia may modulate 53 TORC1 in one particular tissue to control overall body growth and development. A precedent for this is 54 the nutrient regulation of larval physiology and growth. For example, nutrient-dependent changes in 1 TORC1 signalling in specific tissues such as the fat body or prothoracic gland can control whole animal 2 growth and development through non-autonomous effects on endocrine signaling. In this manner, one 3 tissue functions as a sensor of environmental stimuli to coordinate whole body responses. To examine 4 a potentially similar role in hypoxia sensing, we examined whether TORC1 suppression in a specific 5 tissue was required for hypoxia tolerance in developing Drosophila. To do this we again took the 6 approach of expressing a UAS-Rheb transgene to maintain TORC1 signaling under hypoxia, but this 7 time we restricted Rheb expression to specific larval tissues. We chose to examine effects on hypoxia 8 tolerance by maintaining animals in either normoxia or hypoxia during their larval period and then 9 measuring survival to eclosion. We tested Gal4 drivers that express in the fat body (r4-Gal4), neurons 10 (elav-Gal4), the intestine (MyoIA-Gal4), the prothoracic gland (P0206-Gal4) and the muscle (dmef2-11 Gal4). We found the most dramatic effects were seen with fat-specific expression of Rheb: r4>Rheb 12 animals grown in hypoxia during their larval stage showed a significant decrease in adult survival 13 compared to r4>+ control animals (Fig 4a). However, in contrast to ubiquitous expression of Rheb, we 14 found that fat body restricted expression did not delay larval development in hypoxia -r4>Rheb animals 15 developed slightly faster to the pupal stage in both normoxia and hypoxia compared to control (r4>+) 16 animals. Also, r4>Rheb animals showed no significant change in final pupal size compared to r4>+) 17 animals. When we performed similar experiments with expression of Rheb in either neurons, intestine 18 or prothoracic gland we saw no effect on viability (Fig 4b-d). Animals expressing Rheb in muscle 19 (dmef2>Rheb) did show reduced adult survival when grown in hypoxia as larvae, however they also 20 showed reduced survival in normoxia, making the effects on hypoxia tolerance difficult to interpret (Fig   21   4e).

23
These results suggest that the larval fat body is an important hypoxia sensing tissue that responds to 24 low oxygen by suppressing TORC1 activity to ensure subsequent viable development. We therefore 25 focused our attention on understanding how reduced TORC1 signaling in the fat body contributes to 26 hypoxia tolerance.

28
Suppression of TORC1 signalling in the fat body leads to increases in lipid droplet size 29 and lipid storage. 30 We next examined how reduction of TORC1 signaling in the larval fat body contributes to normal 31 organismal development and survival in hypoxia. The role of the fat body as a coordinator of overall 32 body physiology and development has been best studied in the context of altered dietary nutrients. In 33 particular, when larvae are starved of nutrients the fat body mobilizes stored sugars and lipids in order 34 to maintain circulating levels of these nutrients and support tissue homeostasis 9,10 . Upon starvation, fat 35 body cells also rapidly engage autophagy to promote organismal survival 17 . We therefore examined 36 whether these changes are associated with exposure to low oxygen. We first examined autophagy 37 since this is a well-studied conserved process known to be induced by TORC1 inhibition. We subjected 38 early third instar larvae to hypoxia for six hours and then stained fat bodies with LysoTracker Red to 39 visualize lysosomes and late stage autophagosomes as an indicator of autophagy. We also stained fat 40 bodies from larvae maintained in normoxia and from larvae subjected to six hours of nutrient starvation, 41 a condition known to induce autophagy. We found that fat bodies from normoxic animals showed little 42 staining with LysoTracker Red, while starved fat bodies showed a marked increase in LysoTracker Red 43 punctae, consistent with induction of autophagy ( Fig 5). In contrast, we saw little or no LysoTracker Red 44 punctae in fat bodies from larvae exposed to hypoxia for six hours (Fig 5). Even longer hypoxia 45 exposure (24 hours) also did not induced autophagy.

47
We then explored effects on lipid metabolism. In the fat body, triacylglycerol (TAGs) are stored within 48 large lipid droplets. These lipid stores then can be mobilized under starvation conditions to supply a 49 source of free fatty acid for beta-oxidation and other metabolic processes required for homeostasis 35 . 50 We observed that when larvae were raised in hypoxia they showed a noticeable change in fat body 51 morphology, which became less opaque in appearance as has been reported previously 36 . When we 52 examined the fat bodies under light microscopy we saw an increase in cytoplasmic lipid droplet size 53 (Fig 6a). We examined this phenotype in more detail by using Nile Red to stain the neutral lipids that 54 compose these cytoplasmic lipid droplets. When we transferred second instar larvae to hypoxia for two 1 days we observed a significant increase lipid droplet diameter compared to larvae maintained in 2 normoxia for the same period (Fig 6b, c). This effect on lipid droplets was opposite to that seen in larvae 3 that were starved of all nutrients for two days (PBS only), which exhibited a marked decrease in lipid 4 droplet size (Fig 6b). Instead, the hypoxia phenotype was similar to animals that were transferred to a 5 sugar-only diet for two days. These results indicate that the effects of hypoxia on lipid droplet size are 6 opposite to those seen in nutrient-deprivation and suggest that under hypoxia larvae may increase TAG 7 levels through increase synthesis from dietary sugars. To measure TAG levels more quantitatively, we 8 raised larvae from hatching in either normoxia or hypoxia and then measured whole-body TAG levels 9 using a colorimetric assay. We found that hypoxic animals exhibited approximately a two-fold increase 10 on total TAG levels when corrected for total larval weight (Fig 6d). We additionally used a previously 11 described sucrose solution buoyancy assay to estimate larval lipid content 37,38 . In this assay groups of 12 isolated wandering third instar larvae are mixed with increasing concentrations of a sucrose solution 13 and the percentage of larvae floating at each concentration is measured. Using this approach, we found 14 that hypoxic larvae were more buoyant than larvae growth in normoxia, consistent with an increase in 15 lipids as a proportion of total body mass (Fig 6e). Altogether, these results indicate that hypoxia induces 16 a remodelling of lipid droplet and an increase in total lipid storage.

18
We next examined whether these changes in lipid metabolism occurred as a consequence of reduced 19 TORC1 activity. To test this, we generated GFP-marked fat body tsc1 mutant cell clones. As we 20 previously described, loss of TSC1 completely reversed the hypoxia-mediated suppression of TORC1 21 signaling. Hence, we examined these tsc1 mutant fat body cells to see if they still showed the hypoxia-22 mediated changes in lipid droplets. We induced clones during mitosis in the embryo and then when the 23 animals hatched we transferred them to hypoxia for their entire larval development. When we dissected 24 and examined the fat bodies from third instar larvae using DIC microscopy, we observed the hypoxia 25 increase in lipid droplet size in all non-GFP cells (Fig 7). However, the tsc1 mutant cells showed no 26 increase in lipid droplet size. Instead they maintained the small lipid droplet morphology typical of 27 normoxic animals at the same stage even though the animals had been grown in hypoxia for several 28 days (Fig 7). These data indicate suppression of TORC1 signalling is required of the hypoxia-mediate 29 remodelling of lipid storage.

31
Reorganization of lipid metabolism is required for hypoxia tolerance. 32 We next examined whether the changes in lipid storage caused by the hypoxia-mediated suppression 33 of fat body TORC1 signaling was important for development and survival. To do this we used genetic 34 knockdown of Lsd2, a Drosophila perilipin homolog 39-41 . Lsd2 is a protein associated with the surface of 35 lipid droplets that is necessary for normal lipid droplet formation. We used expression of an inverted 36 repeat (IR) to Lsd2 (UAS-lsd2 IR) to specifically knockdown Lsd2 in the fat body using the r4-Gal4 37 driver. When we did this and then transferred animals to hypoxia for two days, we found that the large 38 lipid droplet phenotype seen in control (r4>+) animals was blocked when Lsd2 levels were reduced 39 (r4>lsd2 IR; Fig 8a). We then explored how this inhibition of lipid droplet size affected tolerance to 40 hypoxia. We maintained r4>+ and r4>lsd2 IR larvae in hypoxia from larval hatching to pupation, and 41 then switched them back to normoxia and monitored viability to adult stage. We found that the r4>lsd2 42 IR showed a significant reduction in survival compared to r4>+ control animals (Fig 8b). To confirm this 43 effect, we also examined a previously reported lsd2 mutant allele (lsd2 KG00149 ). These lsd2 mutants are 44 viable and show normal development when grown on normal laboratory food in normoxia. However, 45 when we maintained these lsd2 mutants in hypoxia throughout their larval period, they showed a 46 marked reduction in survival to adult stage compared to control (w 1118 ) animals (Fig 8c). These results 47 indicate that the increase in lipid droplet size caused by reduced TORC1 is required for organismal 48 adaptation to hypoxia. In this paper, we explored how Drosophila are able to tolerate hypoxia. A central finding of our work is 53 that when larvae are exposed to low oxygen, the fat body serves as a key hypoxia sensor that mediates 54 changes in physiology to ensure viable organismal development. This hypoxia sensor role is mediated 1 through inhibition of TORC1 signaling and reorganization of lipid storage. This function of the fat body 2 as a hypoxia sensor is reminiscent of the role of the fat body is coordinating whole body physiology 3 responses to changes in dietary nutrients 9,10,17,42,43 . As we find in hypoxia, these nutrient effects are 4 also dependent on modulation of TORC1 activity and they can exert both metabolic and endocrine 5 effects to control growth and development. These studies and our findings in hypoxia, emphasize how 6 the fat body functions a sentinel tissue to detect changes in environmental conditions and to buffer the 7 internal milieu from these changes. Moreover, while most work on hypoxia has focused on studying 8 cells in culture 27,44 , our findings emphasize the importance of non-cell autonomous mechanisms in 9 controlling how animals adapt to low oxygen. 10 11 Inhibition of TORC1 in larvae exposed to hypoxia occurred rapidly and, interestingly, only in response to 12 a specific range of low oxygen (~2-6%). At <2% oxygen and lower, the response to hypoxia is very 13 different compared to exposure to 5% oxygen that was used in this study -larvae crawl away from the 14 food and eventually undergo complete movement arrest, which can be reversed within minutes of return 15 to normoxia. Larvae can only tolerate this level of low oxygen (<2%) for a few hours before dying. Since 16 this low oxygen hypoxic response is different to the behaviour of larvae at 5% oxygen (which maintain 17 their feeding and growth) it may also rely on qualitatively different changes in hypoxia sensing and 18 signaling that do not involve suppression of TORC1. The hypoxia-mediated inhibition of TORC1 that we 19 found required TSC1/2 but was independent of two main mechanisms defined in mammalian cell 20 culture experiments -induction of REDD1 by the well-studied HIF-1 alpha transcription factor or by 21 activation of AMPK. Although the Drosophila homolog of REDD1, Scylla, was previously shown to be 22 sufficient to inhibit TORC1 32 , we found that it was not necessary. Indeed, analysis of the REDD1 23 mutant mouse also showed that in certain tissues, hypoxia-mediated repression of TORC1 was also 24 REDD1-independent 34 . A previous report in cell culture showed that upon different stresses including 25 hypoxia, TSC2 could translocate to the lysosome and inhibit Rheb activation of TORC1 45 . Therefore, 26 upon hypoxia exposure in larvae, the TSC1/2 complex may rapidly re-localize to inhibit TORC1 27 function. The mechanism that could drive this (or any other potential mechanism of TORC1 inhibition) 28 must be triggered rapidly in response to hypoxia in larvae. Given the importance of oxygen as an 29 electron acceptor in the electron transport chain in the mitochondria, it is plausible that the rapid One result we found interesting was that the suppression of fat body TORC1 and altered larval lipid 53 storage was not necessary for viable larval development under hypoxia, but was required for 54 subsequent development in the pupal stage to produce viable adults. During the pupal stage, tissues 1 undergo metamorphosis to establish the adult body. Since this is also a non-feeding stage of the life 2 cycle, the energy required to fuel these extensive tissue rearrangements in pupae must therefore come 3 from stored nutrients. It has been calculated that the lipid stores provide 90% of this energy 52 . Our 4 findings suggest that pupae may be more dependent on these lipid stores after a period of prior larval 5 hypoxia. Hence failure to maintain these stores, either by preventing TORC1 inhibition (Rheb 6 overexpression) or genetic disruption of lipid droplet formation (Lsd2 knockdown), lead to reduced 7 viability in hypoxia, while having no effect on normal development in normoxia. It is also possible that 8 the requirement for altered lipid stores may reflect a role for lipid droplets beyond simply providing a 9 usable energy source 53 . A pertinent example is a report describing how increases in glial lipid droplets 10 in larvae were important for maintaining neuroblast cell proliferation in larvae exposed to hypoxia or 11 oxidative stress 54 . In this case, the lipid droplets were required to play an antioxidant role to buffer 12 neurons from ROS-induced damage. Mammalian cancer cells in culture have also been shown to 13 accumulate lipid droplets in low oxygen, an effect that is important to promote their survival and 14 tumorigenic phenotypes in mouse models 55 . Cancer cells with high levels of TORC1 activity have also 15 been shown be dependent on exogenous fatty acids for their survival in hypoxic conditions 56,57 . Hence, 16 the lipid droplet phenotypes we observed may be important for ensuring cell and tissue viability in pupal 17 stages independent of any role in energy production.

19
In conclusion, our studies presented here pinpoint the Drosophila fat body as a key hypoxia sensing 20 tissue that ensures viable animal development in low oxygen. We suggest that, given the importance of looking at TORC1 activity, either total eIF2 alpha, actin or total S6K levels were used as loading 10 controls because the levels of these proteins were unaffected by hypoxia.