Sestrin mediates detection of and adaptation to low-leucine diets in Drosophila

Mechanistic target of rapamycin complex 1 (mTORC1) regulates cell growth and metabolism in response to multiple nutrients, including the essential amino acid leucine1. Recent work in cultured mammalian cells established the Sestrins as leucine-binding proteins that inhibit mTORC1 signalling during leucine deprivation2,3, but their role in the organismal response to dietary leucine remains elusive. Here we find that Sestrin-null flies (Sesn−/−) fail to inhibit mTORC1 or activate autophagy after acute leucine starvation and have impaired development and a shortened lifespan on a low-leucine diet. Knock-in flies expressing a leucine-binding-deficient Sestrin mutant (SesnL431E) have reduced, leucine-insensitive mTORC1 activity. Notably, we find that flies can discriminate between food with or without leucine, and preferentially feed and lay progeny on leucine-containing food. This preference depends on Sestrin and its capacity to bind leucine. Leucine regulates mTORC1 activity in glial cells, and knockdown of Sesn in these cells reduces the ability of flies to detect leucine-free food. Thus, nutrient sensing by mTORC1 is necessary for flies not only to adapt to, but also to detect, a diet deficient in an essential nutrient. Fruitflies require Sestrin to regulate mTORC1 signalling in response to dietary leucine, survive a diet low in leucine, and control leucine-sensitive physiological characteristics, which establishes Sestrin as a physiologically relevant leucine sensor.

Mechanistic target of rapamycin complex 1 (mTORC1) regulates cell growth and metabolism in response to multiple nutrients, including the essential amino acid leucine 1 . Recent work in cultured mammalian cells established the Sestrins as leucine-binding proteins that inhibit mTORC1 signalling during leucine deprivation 2,3 , but their role in the organismal response to dietary leucine remains elusive. Here we find that Sestrin-null flies (Sesn −/− ) fail to inhibit mTORC1 or activate autophagy after acute leucine starvation and have impaired development and a shortened lifespan on a low-leucine diet. Knock-in flies expressing a leucine-binding-deficient Sestrin mutant (Sesn L431E ) have reduced, leucine-insensitive mTORC1 activity. Notably, we find that flies can discriminate between food with or without leucine, and preferentially feed and lay progeny on leucine-containing food. This preference depends on Sestrin and its capacity to bind leucine. Leucine regulates mTORC1 activity in glial cells, and knockdown of Sesn in these cells reduces the ability of flies to detect leucine-free food. Thus, nutrient sensing by mTORC1 is necessary for flies not only to adapt to, but also to detect, a diet deficient in an essential nutrient.
The protein kinase mTORC1 regulates growth and metabolism in response to diverse signals, including growth factors and nutrients such as amino acids 1 . Amino acids activate mTORC1 by promoting its translocation to the lysosomal surface, where its essential activator Rheb resides [4][5][6] .The heterodimeric Rag GTPases, which are under the control of several multi-component protein complexes, including GATOR1 and GATOR2 (ref. 7 ), regulate the lysosomal localization of mTORC1 (refs. 4,5 ). GATOR1 is a GTPase-activating protein for RagA and RagB and is necessary for amino acid deprivation to inhibit mTORC1 signalling 8,9 . By contrast, GATOR2 is required for amino acids to activate mTORC1 and directly interacts with several of the amino acid sensors so far discovered, indicating that it acts as a nutrient-sensing hub despite its still unknown biochemical function 7 .
Among the proteogenic amino acids, leucine is the best-established activator of mTORC1 (refs. [10][11][12][13]. Work in cultured mammalian cells has shown that leucine controls mTORC1 by regulating the interaction of GATOR2 with the Sestrin family of proteins 3,14,15 , which are repressors of mTORC1 signalling 16,17 . Human Sestrin1 and Sestrin2 bind leucine at affinities consistent with the leucine concentration needed to activate mTORC1 and are required for leucine deprivation to inhibit mTORC1 signalling 3 . Moreover, a Sestrin2 mutant that does not bind leucine fails to dissociate from GATOR2 in the presence of leucine, and in cells expressing this mutant, mTORC1 activity remains low even when the cells are cultured in leucine-replete conditions 2,3 . Despite the evidence that Sestrin is a leucine sensor for the mTORC1 pathway in cultured mammalian cells, the roles of Sestrin-mediated leucine sensing in the physiology of an intact organism remain largely unexplored. Although much of the work on leucine sensing has been in mammalian systems, Sestrin and the core nutrient-sensing machinery, including the Rag GTPases, GATOR1 and GATOR2, are conserved in most invertebrates, including the fly Drosophila melanogaster 18 . Unlike in mammals, flies express only one gene for Sestrin (Sesn) 16 , greatly facilitating the in vivo study of leucine sensing by mTORC1. Here we show that Sestrin and its leucine-binding pocket are required for leucine to regulate mTORC1 activity in fly tissues in vivo and for flies to detect and adapt to leucine-deficient diets.

Fly mTORC1 senses leucine in vivo through Sestrin
In an equilibrium binding assay, Drosophila Sestrin bound leucine with a dissociation constant (K d ) of about 100 µM (Fig. 1a), an affinity several fold lower than those of human Sestrin1 and Sestrin-2 (K d values of about 15-20 µM) 3 . This reduced affinity is probably the result of a difference between the leucine-binding pockets of human and fly Sestrin. Structural studies show that in human Sestrin2 a tryptophan (W444) forms the floor of the pocket, but in the fly protein, the analogous residue is a leucine (L431), a smaller residue that when introduced into human Sestrin2 (W444L) is sufficient to reduce its leucine-binding capacity by several fold 2 . The low leucine affinity of fly Sestrin is consistent with the observation that fly haemolymph has substantially higher amino acid concentrations than human plasma 18,19 , a difference probably reflected Article intracellularly. Like the analogous mutant of human Sestrin2 (W444E), fly Sestrin(L431E) does not bind leucine (Fig. 1b).
To examine whether leucine regulates the interaction of fly Sestrin with GATOR2, we stably expressed in Drosophila S2R+ cells a Flag-tagged control protein (und, the Drosophila orthologue of mammalian metap2, methionyl aminopeptidase) or WDR59, one of the five core components of the GATOR2 complex. Sestrin co-immunoprecipitated with GATOR2, but not und, and removal of leucine from the cell medium strongly enhanced the interaction. The addition of leucine, but not isoleucine, valine or methionine, to the immunoprecipitates was sufficient to release Sestrin from GATOR2 (Fig. 1c). Thus, like the human protein, fly Sestrin binds to GATOR2 in a fashion that is specifically disrupted by leucine.
To extend our work in vivo, we generated flies that ectopically express MYC-tagged WDR24, another core component of GATOR2 (lpp>myc-WDR24 flies), in the fat body, and are either wild type at the Sesn locus or have a knock-in mutation causing the L431E substitution that renders Sestrin unable to bind leucine (Sesn L431E ). For a period of 4.5 h, we fed third instar larvae a chemically defined diet (see Methods and Extended Data Tables 1-4 for details) containing all proteogenic amino acids (amino acid replete) or the same diet lacking just leucine (leucine free) or valine (valine free). Regardless of genotype, larvae eating the leucineor valine-free diets had reduced levels of leucine or valine, respectively (Extended Data Fig. 1a,b). In lysates prepared from isolated fat bodies, endogenous Sestrin co-immunoprecipitated with GATOR2, but not a control protein (GFP-MYC), and deprivation of leucine, but not valine, strongly boosted the interaction. In contrast, Sestrin(L431E) bound equally well to GATOR2 under all dietary conditions, consistent with the mutant being leucine insensitive (Fig. 1d). In cultured cells and in fat bodies, we observed that Sestrin has multiple isoforms (Fig. 1c,d), probably the result of differential splicing 16 .
In wild-type larvae, feeding of the diet free in leucine, but not valine, inhibited mTORC1 in the fat body, as assessed by the phosphorylation of S6K, a canonical mTORC1 substrate. The loss of Sestrin (Sesn −/− ) did not impact mTORC1 activity in larvae eating the amino-acid-replete diet, but completely prevented the inhibition of mTORC1 normally caused by leucine deprivation (Fig. 1e). Sestrin was also required for the leucine-free diet to activate autophagy, a process suppressed by mTORC1, as monitored by the formation of mCherry-Atg8a-positive puncta (Extended Data Fig. 1c). In Sesn L431E larvae, mTORC1 activity was low relative to that in wild-type larvae and also unaffected by leucine deprivation, indicating that the leucine-binding mutant of Sestrin acts as a non-repressible inhibitor of mTORC1 (Fig. 1e). Notably, mTORC1 signalling was inhibited in Sesn −/− larvae deprived of all food to a similar extent as in wild-type larvae (Extended Data Fig. 1d), which is consistent with work in cultured mammalian cells showing that Sestrin has a specific role in transmitting leucine availability to mTORC1 (refs. 3,14 ). Last, in larvae lacking a component of GATOR1 (    the absence of dietary leucine did not impact mTORC1 activity and it remained as hyperactive or suppressed, respectively, as when the larvae were fed the amino-acid-replete diet (Fig. 1e). Consistent with mTORC1 promoting Sesn transcription as part of a feedback loop 16,20 , Nprl2 −/− and Mio −/− flies had increased and decreased Sestrin levels, respectively (Fig. 1e). Collectively, these results show that dietary leucine modulates mTORC1 in vivo and that this regulation requires Sestrin and its leucine-binding pocket as well as the GATOR1 and GATOR2 complexes.

Sestrin mediates adaption to low-leucine diets
We reasoned that Sestrin-mediated suppression of mTORC1 helps animals adapt to and thus survive a diet low in leucine. We first tried to test this idea by feeding larvae food lacking leucine, but all larvae, independently of genotype, died within 2-3 days of starting the diet, consistent with leucine being an essential amino acid required for larval growth. When given food containing one-tenth of the normal leucine content, about 40% of wild-type larvae survived over a period of 16 days (Fig. 2a,b). In contrast, only about 10% of Sesn −/− larvae did so (Fig. 2b). Moreover, the surviving larvae grew to a much smaller size than their wild-type counterparts (Fig. 2c), a defect rescued by the expression of wild-type Sestrin from the ubiquitous Tubulin-Gal4, Tubulin-Gal80 ts promoter (Fig. 2c). When fed the standard laboratory diet, Sesn −/− and wild-type larvae developed indistinguishably (Extended Data Fig. 2a). Consistent with previous work showing that adult flies can live for weeks on a diet lacking any amino acid source 21 , our observations showed that wild-type flies also survived for many weeks on a leucine-free diet (Fig. 2e,h, Extended Data Fig. 2c,f and Supplementary Data 1). As with larvae, adult flies also required Sestrin to adapt to leucine scarcity, as Sesn −/− male and female animals had greatly shortened lifespans on the leucine-free, but not amino-acid-replete, diet (Fig. 2d,e,g,h and Supplementary Data 1). On the other hand, Sesn −/− flies had slightly shorter lifespans than wild-type counterparts only when eating the valine-free food (Fig. 2f,i and Supplementary Data 1), a diet on which the activity of processes controlled by mTORC1, such as protein synthesis and autophagy, would be expected to impact survival. When the Sesn L431E flies were fed the same chemically defined diets, they survived similarly to the wild-type flies (Extended Data Fig. 2b-g and Supplementary Data 1). Consistent with the chronic suppression of mTORC1 signalling, Sesn L431E larvae reared on the standard laboratory diet developed more slowly than wild-type ones (Extended Data Fig. 2h).

Wild type attP2
Sesn -/-attP2  The P values were determined using a two-proportion z-test (two-sided). The bars show the percentage of surviving larvae in each genotype and the error bars represent the 95% Wald confidence interval. c, Sestrin is required for larval growth on a low-leucine diet. Shown are age-synchronized animals of the indicated genotypes raised for 9 days on either an amino-acid-replete diet or a reduced (10%)-leucine diet.

Article
We monitored mTORC1 activity in whole-fly lysates of female and male adult flies that had been fasted overnight and then refed for 90 min with the chemically defined diets used above. The loss of Sestrin prevented the inhibition of mTORC1 caused by the leucine-free diet in male and female flies (Extended Data Fig. 3a,b).
We further focused on oogenesis, a physiological trait that is known to be regulated by diet 22 . Moreover, diet is known to regulate ovarian function through the GATOR1-GATOR2 complexes 21,[23][24][25] , and Mio, the gene for one of the components of GATOR2, was so named because mutations in it result in a missing oocyte phenotype 26 . We found that mTORC1 activity was strongly increased in the ovaries of Sesn −/− flies eating the standard laboratory diet, and as in larval fat bodies (Fig. 1e), it was suppressed in the ovaries of Sesn L431E flies (Extended Data Fig. 3c).
When fed the amino-acid-replete or valine-free diet, Sesn −/− and wild-type flies had ovaries of similar sizes, but the loss of Sestrin greatly reduced ovarian size in flies under conditions of acute leucine deprivation (Extended Data Fig. 3d,e), again pointing to a specific role for Sestrin in adapting to leucine scarcity. The ovaries of the Sesn L431E flies were equally small on all of the diets (Extended Data Fig. 3d,e), consistent with a role for mTORC1 in the control of gonad development. Sesn L431E flies also had reduced fecundity as they laid fewer eggs than wild-type flies (Extended Data Fig. 3f). Eggs from wild-type, Sesn L431E and Sesn −/− flies had comparable hatching rates, suggesting that Sestrin does not impact fertility (Extended Data Fig. 3g). Collectively, these data reveal that in larvae and adult flies Sestrin promotes survival on a low-leucine diet and has a particularly important role in controlling ovarian size and function.

Sestrin regulates feeding behaviour
Having established that Sestrin is important for flies to adapt to and survive on diets low in leucine, we examined whether flies also require Sestrin to detect and thus avoid food that is poor in leucine. To do so, we developed an assay to test whether adult flies prefer eating leucine-rich over leucine-poor food. The experimental set-up consisted of 15 female and 5 male flies in a bottle containing 2 apple pieces, the first painted with a solution of one or more amino acids and the second with an appropriate control (Fig. 3a). Each also contained a trace amount of a unique DNA oligonucleotide, which served as a barcode for measuring the food consumption, an approach previously described 27 and that we validated (Extended Data Fig. 4a-c and Methods). We chose apple as the base food because it is carbohydrate rich and protein poor 28 , allowing us to set up food choices that have different amino acid compositions but the same content of sugars. Apples are reported to contain very little leucine and valine 29-31 .
We found that wild-type female flies prefer to eat apples coated with leucine rather than water. This preference emerges after the flies have been eating the food for about 6 h and increases to 5-6-fold by 24 h, the time point we used in subsequent experiments (Fig. 3b). The preference for leucine is concentration dependent (Extended Data Fig. 4d) and not every amino acid elicits a preference, as flies do not distinguish between apples coated with valine or water (Extended Data Fig. 4e). Given a choice between equal amounts of leucine and valine, flies still prefer leucine, suggesting that the leucine preference is not simply the result of a nitrogen imbalance (Extended Data Fig. 4e). Moreover, the leucine preference requires differential mTORC1 activity, as when flies were fed the mTORC1 inhibitor rapamycin, they no longer showed a preference (Fig. 3c). Rapamycin treatment also lowered the total amount of food consumed by the flies (Extended Data Fig. 4f), consistent with previous reports 32, 33 .
Remarkably, neither Sesn −/− nor Sesn L431E female flies-both of which have leucine-insensitive mTORC1 signalling-had a preference for leucine as they ate similar amounts of leucine-rich and leucine-poor foods (Fig. 3d,e and Extended Data Fig. 4g). However, the two Sesn mutants probably differ in the total amount of food each ate. The amount of food (leucine-rich or leucine-poor) that Sesn −/− female flies ate was similar to the amount of leucine-rich food consumed by wild-type (w 1118 ) flies (Extended Data Fig. 4h). The opposite was true for Sesn L431E female flies. These flies ate an amount of food (leucine-rich or leucine-poor) similar to the amount of leucine-poor food consumed by the wild-type (OreR) flies (Extended Data Fig. 4i). That Sesn L431E files, which have low mTORC1 signalling, eat less food than wild-type controls is consistent with rapamycin causing a reduction in food consumption (Extended Data Fig. 4f). Whole-body re-expression in the Sesn −/− female flies of Sestrin driven by Tub>Gal4 partially restored the leucine preference of the animals (Extended Data Fig. 4j).
We also examined whether flies can distinguish between foods with a more subtle difference in amino acid composition: an apple coated with the 20 proteogenic amino acids versus just 19 of them (that is, lacking only leucine). Indeed, this was the case and this preference was also absent in the Sesn −/− and Sesn L431E flies (Fig. 3f). Valine again served as a control: when removed from the 20-amino-acid cocktail, neither wild-type nor Sesn mutant flies showed preference for the valine-containing food (Extended Data Fig. 4k).
To obtain temporal control of Sestrin suppression, we generated a conditional knockdown system using a short hairpin RNA (shRNA) targeting Sesn. Ubiquitous expression of the shRNA reduced Sestrin protein levels (Fig. 3g), and as expected, the preference of the flies for the leucine-containing food (Fig. 3h). Using a temperature-sensitive shRNA driver, we suppressed Sestrin specifically during adulthood ( Fig. 3i,j). This too reduced their leucine preference (Fig. 3k), indicating that the acute loss of Sestrin in adult flies is sufficient to impact the leucine preference. Notably, the temperature shift to 29 °C increased Sestrin levels (Fig. 3j), consistent with previous work showing that multiple stresses induce its transcription 17,34 . Thus, female flies can readily detect food lacking leucine even if it contains sugars and other amino acids. This ability requires Sestrin and its capacity to bind leucine.
To further analyse the physiological relevance of leucine sensing through the Sestrin-mTORC1 axis, we tested the impact of both leucine and Sestrin on the choice between low-and high-protein diets: apple coated with a low or high amount of yeast extract, which is a complex type of food and the major protein source for laboratory-raised flies. Wild-type flies had a strong preference for the apple with a higher protein content. The addition of leucine to the protein-poor food reduced the preference of wild-type female flies for the protein-rich food, but only minimally impacted the preference of the Sesn L431E mutants (Extended Data Fig. 5a). Sesn −/− mutants showed a similar trend (Extended Data Fig. 5b), but it was not statistically significant. Together, these data suggest that flies use leucine sensing through the Sestrin-mTORC1 axis as a proxy for the food protein content.

Sestrin regulates egg-laying behaviour
We found that female flies prefer to lay eggs on the leucine-coated apples. To explore this further, we put 15 female and 5 male flies in the assay bottle and 24 h later counted the number of eggs on each piece of apple (Extended Data Fig. 6a). In an initial test, we found that flies laid many more eggs on an apple piece painted with a yeast suspension instead of water, consistent with yeast being a food rich in nutrients and the olfactory cues that attract flies 35-38 (Extended Data Fig. 6b).
Wild-type flies that had been deprived of protein overnight deposited 5-6-fold more eggs on an apple piece coated with the 20 proteogenic amino acids instead of water (Extended Data Fig. 6c,d,f). Flies had a similar, albeit smaller (threefold), preference for leucine-coated apples, and this preference was more profound when the flies had been starved for protein. Importantly, flies did not distinguish between apple pieces painted with the same substance (Extended Data Fig. 6d,f).
We found that Sesn L431E mutant flies lacked a strong preference for laying eggs on the apple coated with leucine and had a reduced preference for the apple with the 20 amino acids (Extended Data Fig. 6e), although the total number of eggs Sesn L431E mutant flies laid was about 25% reduced compared to that for the wild-type flies (Extended Data Fig. 3f). This altered egg-laying behaviour was also observed in the Sesn −/− flies, which laid a similar number of eggs to the wild-type animals (Extended Data Fig. 6g). Furthermore, the wild-type flies mildly preferred to deposit eggs on an apple piece painted with the 20 proteogenic amino acids instead of 19 (that is, lacking leucine), a much more complex choice, and this ability was reduced in the Sesn L431E flies (Extended Data Fig. 6h). When facing the same complex choice, Sesn −/− flies did not show a statistically significant different behaviour compared to the wild-type flies (Extended Data Fig. 6h), which might reflect the subtleness and noise of this complex choice set-up. Consistent with the leucine preference we observed in the food choice assay, we found that female flies also laid fewer eggs on food lacking leucine, and this capacity requires the intact leucine-binding pocket of Sestrin. This finding might reflect an active choice for egg deposition or the amount of time that flies spend on each apple owing to their preference for eating leucine-containing food.

Glial Sestrin regulates leucine preference
To determine in which tissue(s) Sestrin is required for flies to distinguish between food with or without leucine, we suppressed Sestrin with the Sesn shRNA under the control of a variety of cell-type-specific   The data show the fold difference in relative food intake for the leucine-coated compared to water-coated apples. n ≥ 11 per time point. c, Rapamycin prevents flies from developing a preference for the leucine-coated apple. n ≥ 5 per condition. d-f, Sesn L431E and Sesn −/− animals fail to develop a preference for the leucine-containing apple. In d,e, n ≥ 4 per condition; in f, n ≥ 6 per condition. g, Immunoblotting for Sestrin following knockdown of Sesn in adult flies. Akt serves as a loading control. h, Ubiquitous knockdown of Sesn reduces the preference of adult female flies for leucine. The data show the fold difference in food intake for the leucine-coated apple relative to the water-coated apple. n ≥ 5 per condition. i, The approach used to achieve temporal control of Sesn knockdown in j,k. j, Sesn immunoblot showing Gal80 ts -mediated depletion of Sestrin in adult, but not developing, animals. Extracts were prepared from flies raised at the indicated temperatures. S6K serves as a loading control. Note that heat shock induces Sestrin protein levels in control flies. k, Knockdown of Sesn during adulthood is sufficient to decrease the preference of female flies for leucine-containing apples. n ≥ 13 per condition. a,i, Created with BioRender. com. In b-f,h,k the values are mean ± s.d. of biological replicates from a representative experiment. Each experiment was repeated three (d-k) or two (b,c) times with similar results. Statistical analyses were carried out using one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparisons test (b), two-way ANOVA followed by Šídák's multiple comparisons test (c-e), one-way ANOVA followed by Šídák's multiple comparisons test (f) and two-tailed unpaired t-test (h,k).

Article
Gal4 drivers. Notably, Sesn knockdown specifically in glial cells (repo-Gal4) was sufficient to reduce the preference of flies for the leucine-containing food to a similar extent as when it was expressed ubiquitously (da-Gal4; Fig. 4a). In contrast, Sesn knockdown in many other tissues, including the fat body and muscle, did not impact the leucine preference. It is important to note that the intrinsic capacity of each Gal4 driver line to distinguish between food with or without leucine varied considerably (Extended Data Fig. 7a), probably owing to their different genetic backgrounds. Thus, although we are confident that the preference of flies for leucine-containing food requires Sestrin in glial cells, we are cautious in ruling out contributions from other tissues, particularly those examined with driver lines with intrinsically lower leucine preferences, such as the pan (elav-Gal4) and dopaminergic and cholinergic (ddc-Gal4) neuronal lines (Extended Data Fig. 7a).
Consistent with an important role for glial Sestrin in regulating the leucine preference, expression of wild-type Sestrin just in glial cells in Sestrin-null flies partially rescued the defect in detecting leucine-poor food (Extended Data Fig. 7b). In wild-type flies, expression in the glial cells of either wild-type Sestrin or Sestrin(L431E) decreased the leucine preference, consistent with the inhibition of mTORC1 caused by Sestrin overexpression (Extended Data Fig. 7b). Indeed, overexpression under the control of repo-Gal4 of TSC1 and TSC2-well-established inhibitors of mTORC1 signalling-was also sufficient to decrease the leucine preference (Extended Data Fig. 7c).
Analyses of a single-cell RNA-sequencing dataset indicated that Sestrin is expressed in most glial subtypes 39 (Extended Data Fig. 7d). Expression of the Sesn shRNA under the control of Gal4 driver lines that target subtypes of glial cells revealed that none caused as strong a suppression of the leucine preference as with the pan-glial driver The images in b,c were taken with 10× and 40× objectives, respectively. Scale bars, 50 µm (b) and 10 µm (c). d, In wild-type flies, but not Sesn L431E or Sesn −/− flies, leucine starvation increases the number of GFP-positive peri-oesophageal glial cells. Each point represents the ratio of the number of GFP-to Repo-positive cells in the oesophageal area of one fly brain. n ≥ 3 per condition. e, Proposed role of the Sestrin-mTORC1 pathway in regulating the preference of flies for leucinecontaining food. In a,d, the values are mean ± s.d. of biological replicates from a representative experiment. The data are representative of three independent experiments with similar results. Statistical analysis was performed using two-tailed unpaired t-test (a), and two-way ANOVA followed by Šídák's multiple comparisons test (d).
repo-Gal4 (Extended Data Fig. 7e), although Wrapper-Gal4-driven Sesn knockdown led to a partial reduction of the leucine preference. Thus, multiple glial subtypes probably participate in mediating the leucine preference.
Given the importance of glial Sestrin in mediating the leucine preference, we examined mTORC1 signalling in glial cells in the brains of adult female flies. To do so, we used a line expressing a GFP-based reporter for the MITF transcription factor 40 , which is the Drosophila orthologue of mammalian TFEB (ref. 41 ). mTORC1 suppresses MITF so that after mTORC1 inhibition, MITF activity increases 41 and drives GFP expression. In wild-type flies, starvation for total protein activated, as indicated by elevated GFP expression, MITF in Repo-positive glial cells, particularly in those surrounding the oesophagus (Extended Data Fig. 7f). Remarkably, starvation for just leucine also increased the number of peri-oesophageal GFP-positive glial cells (Fig. 4b-d and Extended Data Fig. 8a,b). In contrast, in Sesn −/− flies, leucine starvation did not increase the number of peri-oesophageal GFP-positive glial cells, which were few in number irrespective of the diet (Fig. 4c,d and Extended Data Fig. 8a,b). In Sesn L431E flies, there were many peri-oesophageal GFP-positive glial cells, and, like in Sesn −/− flies, leucine starvation did not increase their numbers (Fig. 4c,d and Extended Data Fig. 8a,b). Notably, quantification of GFP-positive cells in the mushroom body and optic lobe areas showed that, unlike in peri-oesophageal glial cells, the mTORC1 activity in these cells did not significantly respond to acute dietary treatments (Extended Data Fig. 8b-e). Thus, dietary leucine regulates mTORC1 signalling in a subset of glial cells in a fashion that depends on Sestrin and its capacity to bind leucine, and this regulation correlates with the ability of flies to distinguish between food that is rich or poor in leucine.

Discussion
We show that D. melanogaster requires Sestrin to regulate mTORC1 signalling in response to dietary leucine, survive a leucine-poor diet, and control leucine-sensitive physiological measures such as food choice and ovarian size. Flies with a point mutation that eliminates the leucine-binding capacity of Sestrin(L431E) have suppressed, leucine-insensitive mTORC1 signalling. Moreover, whereas wild-type flies can live on leucine-free diets for weeks, flies lacking Sestrin die much faster. In all, our results establish Sestrin as a physiologically relevant leucine sensor in vivo. Recently, Lu et al. reported complementary findings of an amino acid-sensing role of Sestrin upstream of mTORC1 in the control of Drosophila development, fecundity and longevity 42 .
We find that Sestrin and its leucine-binding pocket are required for the preference of adult female flies for consuming, as well as laying eggs on, leucine-rich instead of leucine-poor food even when it contains sugars and other amino acids. To our knowledge, the ability of flies to choose food that is rich in leucine over food that lacks leucine but still retains a complex set of other nutrients has not been previously documented, although such behaviour has been reported in mice 43 . When given a starker choice than we provided-a pure sugar, such as sucrose or glucose, versus an individual amino acid-flies prefer to eat a variety of essential amino acids in sex-and developmental stage-dependent fashions [44][45][46] .
There has been a long-standing interest in understanding the mechanisms that enable animals, including flies and rodents 43,47 , to prefer diets rich in protein. A variety of mechanisms in flies have been implicated, including amino acid transporters 44 , taste receptors 45,48,49 , GCN2 (ref. 50 ), serotonin 51 and dopamine signalling 50,52 , sex peptide receptor 53 , microbiome 54 , and mTOR and S6K (refs. 51,53 ). How these mechanisms coordinate together to impact organismal protein detection in the diet remains unclear.
Our work raises several questions for future study. One such question concerns whether there is crosstalk between the food preference behaviour controlled by glial cells and acute changes in ovarian size caused by nutritional stress. Another question is whether female flies actively choose to lay more eggs on the leucine-containing food because it has the nutrients needed for larval growth, or whether the apparent preference simply reflects the amount of time they spend on it owing to their dietary preference. As it takes flies many hours to distinguish between leucine-containing and leucine-free food (Fig. 3b), it seems unlikely that the alterations in Sestrin eliminate the preference for leucine by substantially interfering with the capacity of flies to taste leucine. Rather, we favour the idea that leucine, through Sestrin-mTORC1, turns on a neuronal reward circuit that drives food consumption (see potential model in Fig. 4e). Previous work has identified a set of dopaminergic neurons that controls protein hunger 52 , and it will be interesting to examine whether Sestrin-mediated leucine-sensitive mTORC1 signalling can impact these cells. In this regard, it is intriguing that the preference for leucine requires the expression of Sestrin in glia as there is increasing evidence that glial cells can be key intermediates between an environmental signal and its modulation of a neuronal circuit [55][56][57] . Last, it will be interesting to investigate why mTORC1 activity in a set of peri-oesophageal glial cells is particularly sensitive to Sestrin-dependent regulation by dietary leucine.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-022-04960-2.  Table 2 of a recent authentication 58 ). Suspension FreeStyle 293F cells were obtained from Thermo Fisher and cultured in FreeStyle 293 expression medium (Thermo Fisher (12338018)), supplemented with 100 IU ml −1 penicillin and 100 µg ml −1 streptomycin, at a shaking speed of 125 r.p.m. at 37 °C and 8% CO 2 , 80% humidity. No mycoplasma contamination was detected using PCR.
Dissected tissues and whole flies were crushed physically using a bead beater in 1% Triton lysis buffer (same as above). The resulting lysates were cleared by centrifugation in a microcentrifuge (15,000 r.p.m. for 10 min at 4 °C) and analysed as above. For anti-Flag immunoprecipitations, the anti-Flag M2 affinity gel (Sigma number A2220) was washed with lysis buffer three times and then resuspended to a ratio of 50:50 affinity gel to lysis buffer. A 25 µl volume of a well-mixed slurry was added to cleared lysates and incubated at 4 °C in a shaker for 90-120 min. For anti-MYC immunoprecipitations, magnetic anti-MYC beads (Pierce) were washed three times with lysis buffer. A 30 µl volume of resuspended beads in lysis buffer was added to cleared lysates and incubated at 4 °C in a shaker for 90-120 min. Immunoprecipitates were washed three times; once with lysis buffer and twice with lysis buffer with 500 mM NaCl. Immunoprecipitated proteins were denatured by addition of 50 µl of SDS-containing sample buffer (0.121 M Tris, 5% SDS, 12.5% glycerol, 0.25 M dithiothreitol and bromophenol blue) and heated in boiling water for 5 min. Denatured samples were resolved by 8-12% SDS-PAGE, and analysed by immunoblotting.
Leucine-binding assay and K d calculation. For radiolabelled leucine-binding assays using Flag-tagged Drosophila Sestrin, suspension HEK293F cells were seeded at 2.5 million cells ml −1 , and transfected with the pRK5-Flag-Sestrin cDNA using polyethylenimine. At 72 h after transfection, cells were rinsed once in cold PBS and lysed in 1% Triton lysis buffer (1% Triton, 40 mM Hepes pH 7.4, 2.5 mM MgCl 2 and 1 tablet of EDTA-free protease inhibitor (Roche) per 25 ml buffer). Following an anti-Flag immunoprecipitation, the beads were washed four times with lysis buffer containing 500 mM NaCl and then incubated for 1 h on ice in cytosolic buffer (0.1% Triton, 40 mM HEPES pH 7.4, 10 mM NaCl, 150 mM KCl, 2.5 mM MgCl 2 ) with the indicated amount of [ 3 H] leucine and unlabelled leucine. After 1 h, the beads were aspirated dry and rapidly washed four times with binding wash buffer (0.1% Triton, 40 mM HEPES pH 7.4, 300 mM NaCl, 2.5 mM MgCl 2 ). The beads were aspirated dry again and resuspended in 80 µl of cytosolic buffer. Each sample was mixed well, and then 15 µl aliquots were separately quantified using a TriCarb scintillation counter (Perkin Elmer). This process was repeated in pairs for each sample, to ensure similar incubation and wash times for all samples analysed across different experiments.
The affinity for leucine of Drosophila Flag-Sestrin was determined by first normalizing the bound Liquid chromatography-mass spectrometry-based metabolomics and quantification of metabolite abundances. Liquid chromatography-mass spectrometry (LC-MS)-based metabolomics was performed and data were analysed as previously described 59,60 using 500 nM isotope-labelled internal standards. Briefly, an 80% methanol extraction buffer with 500 nM isotope-labelled internal standards was used for whole-fly metabolite extraction. Samples were dried by vacuum centrifugation, and stored at −80 °C until analysed. On the day of analysis, samples were resuspended in 100 µl of LC-MS-grade water, and insoluble material was cleared by centrifugation at 15,000 r.p.m. The supernatant was then analysed as previously described by LC-MS (refs. 59,60 ).

Synthetic fly food formulation and preparation.
Drosophila diet formulations were derived from previous recipes 66,67 with the following modifications: the type of agar (Micropropagation Agar-Type II; Caisson Laboratories number A037); the final percentage of Agar (1%); the amount of sucrose (25 g per litre of food); and the amino acids that were added to stock solutions before or after autoclaving 68 whose order is described below. The amino acid composition of the diet including the concentrations of leucine, isoleucine and valine were based on the exome-matched (that is, the concentrations used for a given amino acid correspond with the prevalence of exons for that amino acid in the Drosophila genome) and Drosophila diet formulation developed in a previous study 67 that was found to be optimal for growth and fecundity without compromising lifespan. The rationale for which amino acids were part of the autoclaving process was based on solubility considerations 68 .
The complete procedure, formula and stock solutions for food production are as follows: prepare mixture 1 (Extended Data Tables 1, 3 and 4); stir using stir bar; autoclave mixture 1 for 15 min; prepare mixture 2 (Extended Data Tables 2-4) and set aside; remove mixture 1 from the autoclave, combine it with mixture 2 and stir, making sure to mix well; quickly pipette the food into Drosophila vials (5-10 ml food per vial); allow the food to solidify/cool for roughly an hour, and then cover the vials (either with cotton plugs or with plastic wrap) and store food at 4 °C. The food is good for about 3 weeks at 4 °C (it will shrink and pull away from the sides of the vials owing to evaporation). (Note, after autoclaving, mixture 1 containing agar can start solidifying (both before and after the two mixtures are combined, but combining the two mixtures will cause food to cool down and solidify fast). Quickly combine and pour the food while the autoclaved mixture is still hot to avoid this. Adding water to the autoclave tray and keeping mixture 1 in this hot water until ready to combine and pour helps prevent premature solidification.) The catalogue numbers for the reagents not listed in Extended Data Tables 1-4 are as follows: sucrose (Sigma, S7903), agar (Caisson, A037), propionic acid (Sigma, P5561). Stocks can be stored at 4 °C for several months unless otherwise specified.
Generation of clones expressing the Sesn shRNA. Clones were generated by crossing yw,hs-flp; mCherry-Atg8a; Act>CD2>GAL4, UAS-nlsGFP/TM6B with the indicated UAS lines. Progeny of the relevant genotype was reared at 25 °C and spontaneous clones were generated in the fat body owing to the leakiness of the heat-shock flipase (hs-flp).
For the assay, the surface of fresh Gala apples was sprayed and cleaned using 70% ethanol. Fresh Gala apple pieces (about 1 g) containing both a piece of peel and pulp were cut on a clean field using a knife (both the knife and the field were precleaned by 70% ethanol). Two apple pieces with similar shape and weight were placed in the opposite corners of a 6 oz (177 ml; 57 length × 57 width × 103 height (in mm)) clean Drosophila bottle. Solutions of 100 µl in volume that contained one DNA oligomer (final concentration 3.5 ng µl −1 ) and substances (that is, sterile water, amino acid solutions and so on) were placed evenly on top of the apple pieces and allowed to soak in for 1.5-2 h. Age-synchronized adult flies (15 female and 5 male animals) were flipped into these assay bottles and allowed to feed ad libitum on the apples for the indicated times in the time course experiments ( Fig. 3b and Extended Data Fig. 4g) and for 24 h in the other food preference experiments. CO 2 -anaesthetized flies were collected using a tweezer. From each bottle, two tubes of female flies were collected with five flies per tube. Five flies were homogenized for each qPCR sample. Homogenization was performed using a beads beater in the cold after adding 250 µl of squishing buffer (10 mM Tris-HCl pH 8.2, 1 mM EDTA, 1 mM NaCl) and 0.5 µl of 20 mg ml −1 proteinase K (Thermo Fisher number AM2546). The whole-fly lysates were digested at 37 °C for 30-40 min after homogenization followed by proteinase K inactivation at 95 °C for 5 min. The samples were centrifuged for 10 min at 15,000 r.p.m. at room temperature and 2 µl of the supernatant was loaded in each qPCR reaction in a 96-well qPCR plate. We used the SYBR green qPCR master mix from Bio-Rad and a CFX96 Touch Real-Time PCR Detection System with a melting temperature of 60 °C and 40 cycles per run.
Genomic Cyp1 qPCR Ct values were used to control for extraction efficiency. For every batch of samples, an average of Cyp1 qPCR Ct values was taken and all samples beyond ±0.5 Ct away from the average were discarded. Standard curves for DNA oligomers 1 and 2 were generated, and the amount of DNA oligomer from each tube of flies was calculated by fitting their Ct values to the standard curves. The preference index was generated by dividing the calculated amount of DNA oligomer 1 by that of DNA oligomer 2.
To remove external oligomer that may stick to the outside of the flies, we used a four-step protocol described previously 27 : a 10-min wash with 10% Contrex AP Powdered labware detergent (catalogue number 5204, Decon Laboratories); a 5-min wash in double-distilled H 2 O; a 2-min wash in 30% bleach; and a 5-min wash in double-distilled H 2 O. All washes were performed in a 1,500 µl microfuge tube with continuous rocking at room temperature.
For Fig. 3c and Extended Data Fig. 4f, we fed the flies with food containing either 25 µM rapamycin or 25 µM ethanol for 2 days before either protein starvation overnight or not (including 25 µM Rapamycin or 25 µM ethanol). Then for the final choice assay, 25 µM of rapamycin or 25 µM ethanol was added to both apple pieces in the container.
Immunofluorescence assays. Fat bodies from aged larvae (96 h after egg laying) were dissected in PBS at room temperature, fixed for 25-30 min in 4% formaldehyde, washed twice for 10 min in PBS 0.3% Triton (PBST), blocked for 30 min (PBST, 5% BSA, 2% FBS, 0.02% NaN 3 ), incubated with primary antibodies in the blocking buffer overnight, and washed four times for 15 min. Secondary antibodies diluted 1:500 in PBST were added for 1 h and tissues were washed four times before mounting in Vectashield (Vector Laboratories) containing 4′,6-diamidino-2-phenylindole (DAPI). Brains from 5-10-day-old adult female flies were dissected and processed as in a previous study 69 .
Images for Fig. 2c and Extended Data Fig. 3d were acquired on a Zeiss Axio Zoom v16. Images for Fig. 4b,c and Extended Data Figs. 1c, 7f and 8b,c were acquired on a Zeiss AxioVert200M microscope with a 63× or 40× oil-immersion objective or a 10× objective and a Yokogawa CSU-22 spinning-disc confocal head with a Borealis modification (Spectral Applied Research/Andor) and a Hamamatsu ORCA-ER CCD camera. The MetaMorph software package (Molecular Devices) was used to control the hardware and image acquisition. The excitation lasers used to capture the images were 405 nm, 488 nm and 561 nm. Images for Extended Data Fig. 6b,c were acquired on an iPhone XR camera through a binocular microscope.
Egg-laying preference assay. The set-up for the egg-laying preference assay was identical to that for the food preference assay. Instead of collecting female flies for qPCR analyses, the two apple pieces were removed from the bottle and examined under a binocular microscope. The number of eggs on each apple piece was determined.
Ovary size quantification. Ovaries were dissected in PBS and bright-field images were acquired using a Zeiss Axio Zoom v16 scope. The size of the ovaries was quantified using the average area of individual ovaries on ImageJ.
Developmental timing. Three-day-old crosses were used for 3-4-h periods of egg collection on standard laboratory food. Newly hatched L1 larvae were collected 24 h later for synchronized growth using the indicated diets at a density of 30 animals per vial. The time to develop was monitored by counting the number of animals that underwent pupariation, every 2 h in fed conditions, or once/twice a day in starved conditions. The time at which half the animals had undergone pupariation is reported. For larva developmental timing experiments, 10%-leucine chemically defined diet was used because complete leucine starvation quickly caused lethality before any size comparison across genotypes could be efficiently and meaningfully performed.

Lifespan experiments.
To generate age-synchronized adult flies, larvae were raised on laboratory food at low density, transferred to fresh food after emerging as adults and allowed to mate for 48 h. Animals were anaesthetized with low levels of CO 2 and sorted at a density of 25 flies per vial. Each condition examined used 8-10 vials of flies. Flies were transferred to fresh vials three times per week at which point deaths were also scored. For adult flies, leucine-free diet or valine-free diet was used.
Statistical analyses. For non-survival experiments, two-tailed unpaired t-tests, multiple t-tests, one-way or two-way ANOVA analyses followed my post hoc tests were used for comparison between two groups in GraphPad Prism (GraphPad Software v9). All comparisons were two-sided unless specified otherwise. All analysed P values are indicated for each comparison made within all figure panels. P values of less than 0.05 were considered to indicate statistical significance.
For survival comparisons in Fig. 2a,b, two-proportion z-tests were performed. Pupariation percentage (Extended Data Fig. 2a,h) data were compared using permutation tests, in which the test statistic was the difference in mean pupariation times of the two genotypes. The distribution of the test statistic under the null hypothesis was estimated by simulating 100 million rearrangements of the data. Permutation tests were performed in R (script available in Supplementary  Data 2). Results for all statistical analyses were summarized in source data files corresponding to each figure.
Analysis of survival data. All data were complete and uncensored. Kaplan-Meier estimates of the survival function were plotted and used to compute median survival times. Log-rank tests were used to compare survival distributions, and univariate Cox proportional hazard analysis (with ties handled by Efron approximation) was used to compute hazard ratios between Sestrin-mutant versus wild-type flies within individual dietary conditions. To examine the interaction between genotype and diet (specifically using the alternative hypothesis that the lifespan defect of Sestrin-mutant versus wild-type flies is exacerbated on a leucine-free compared to a valine-free diet), one-tailed Wald tests were conducted on the interaction coefficients generated by two-factor Cox proportional hazard models with interaction terms (with ties handled by Efron approximation). All statistical analyses on survival data were performed in R (script available in Supplementary Data 3).

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this paper.

Data availability
The data that support the findings of this study are available from the corresponding authors and the Whitehead Institute (sabadmin@ wi.mit.edu) upon reasonable request. Source data are provided with this paper. Values are mean ± SD of biological replicates from a representative experiment. n = 4 independent biological samples. Two samples from wild type (OreR) leucine-free and valine-free, respectively, failed to yield decent peaks for leucine levels, thus discarded. Multiple unpaired t tests, Holm-Šídák multiple comparison method. c, Sesn knockdown prevents autophagy induction upon leucine deprivation. Fat body cells in mid-third instar larvae expressing mCherry-Atg8a were fed the indicated diets for 4.5 h.
The Sesn RNAi was expressed in clones of cells (GFP, outlined) with a FLP-out system 70   Wild-type (OreR) animals were given indicated food choices and the preference fold-difference was shown. n (leucine vs water) = 8, n (valine vs water) = 10, n (leucine vs valine) = 7. f, Rapamycin treatment reduces fly food consumption. Vehicle or Rapamycin pre-treated animals were given a choice between leucineor water-coated apples. For the Rapamycin group during the choice assay, animals were fed on apples painted with Rapamycin in addition to either leucine or water. Data show the normalized values of food consumption. n = 5 for both conditions. g, Sesn L431E animals do not have a preference for valineover water-painted apples. Animals were given a choice between valine-or water-coated apples and food preference was measured at the indicated time points. Data show the fold-difference in relative food intake for the valinecoated apple compared to the water-coated apple. n = 10 (2 hrs), 12 (4 hrs), 12 (6 hrs), 9 (9 hrs), and 9 (24 hrs). h,i, Sesn L431E animals have decreased food intake regardless of the leucine content of the food (h), and Sesn −/− animals have increased food intake regardless of the leucine content of the food (i). n = 4 for all conditions. j, Whole-body re-expression of wild-type Sestrin driven by Tub>Gal4 is sufficient to partially restore the preference for leucine-containing food of Sesn −/− adult female flies. Animals with indicated genotypes were given the choice between leucine-or water-coated apples. Data show the preference of fold-difference. n (attP2) = 10, n (Sestrin WT) = 6. k, Adult female flies do not develop a preference for valine-containing apple regardless of their genotype. Animals with indicated genotypes were given the choice between leucine-or water-coated apples.  Fig. 5 | Leucine-sensing via the Sestrin-mTORC1 axis contributes to the detection of the protein content of food. a, Wild-type (OreR) flies prefer food containing a high amount of yeast extract and this preference is reduced by the addition of leucine to food containing a low amount of yeast extract. Sesn L431E flies have a reduced preference for the food containing a high amount of the yeast extract and the addition of leucine has minimal impact on the preference. How the food preference index was calculated is described in the methods. n (Wild type OreR, no leucine)=5, n (Wild type OreR, with leucine)=7, n (Sesn L431E , no leucine)=6, n (Sesn L431E , with leucine)= 9. b, As in (a) a choice experiment for wild type w 1118 and Sesn −/− flies. n (Wild type w 1118 , no leucine)=9, n (Wild type w 1118 , no leucine)=8, n (Sesn −/− , no leucine)=9, n (Sesn −/− , with leucine)= 12. Values are mean ± SD of biological replicates from a representative experiment. Data are representative of three independent experiments with similar results. Statistical analysis was performed using two-tailed unpaired t test, Holm-Šídák method. Extended Data Fig. 6 | Flies prefer to lay eggs on leucine-containing food in a fashion that requires the leucine-binding capacity of Sestrin. a, Schematic of the setup used in the egg-laying preference assay. Two identical apple pieces were painted with solutions containing different substances and placed on opposite sides of a container. Animals were allowed to feed ad libitum over the course of the assay and the number of eggs deposited on each apple was counted after 24 h. b, c, Wild-type flies prefer to lay eggs on yeast-or amino acid-painted apples over water-painted apples. Scale bars, 1 mm. d-h, Sesn L431E and Sesn −/− animals do not prefer to lay eggs on the leucine-containing apple.
(a) created with BioRender.com. Values are mean ± SD of three biological replicates from a representative experiment. Data are representative of two independent experiments with similar results. Statistical analysis was performed using one-way ANOVA followed by Tukey's multiple comparisons test (d-g), and Šídák's multiple comparisons test (h).

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability Extended Data Fig. 10d: the single cell RNAseq dataset analyzed is Aerts_Fly_AdultBrain_Filtered_57k, which is available here: scope.aerslab.org. All codes required to run the CPH and permutation statistical analyses are provided as source data.