Differential metabolic sensitivity of insulin-like-response- and TORC1-dependent overgrowth in Drosophila fat cells

Abstract

Glycolysis and fatty acid (FA) synthesis directs the production of energy-carrying molecules and building blocks necessary to support cell growth, although the absolute requirement of these metabolic pathways must be deeply investigated. Here, we used Drosophila genetics and focus on the TOR (Target of Rapamycin) signaling network that controls cell growth and homeostasis. In mammals, mTOR (mechanistic-TOR) is present in two distinct complexes, mTORC1 and mTORC2; the former directly responds to amino acids and energy levels, whereas the latter sustains insulin-like-peptide (Ilp) response. The TORC1 and Ilp signaling branches can be independently modulated in most Drosophila tissues. We show that TORC1 and Ilp-dependent overgrowth can operate independently in fat cells and that ubiquitous over-activation of TORC1 or Ilp signaling affects basal metabolism, supporting the use of Drosophila as a powerful model to study the link between growth and metabolism. We show that cell-autonomous restriction of glycolysis or FA synthesis in fat cells retrains overgrowth dependent on Ilp signaling but not TORC1 signaling. Additionally, the mutation of FASN (Fatty acid synthase) results in a drop in TORC1 but not Ilp signaling, whereas, at the cell-autonomous level, this mutation affects none of these signals in fat cells. These findings thus reveal differential metabolic sensitivity of TORC1- and Ilp-dependent growth and suggest that cell-autonomous metabolic defects might elicit local compensatory pathways. Conversely, enzyme knockdown in the whole organism results in animal death. Importantly, our study weakens the use of single inhibitors to fight mTOR-related diseases and strengthens the use of drug combination and selective tissue-targeting.

Introduction

The mTOR (mechanistic Target of Rapamycin) regulatory network orchestrates organismal growth in response to growth factor signaling and nutritional status (Laplante and Sabatini 2012; Howell et al. 2013; Lamming and Sabatini 2013; Shimobayashi and Hall 2014; Caron et al. 2015; Saxton and Sabatini 2017; Mossmann et al. 2018). Activation of this network promotes basal cellular functions, thereby providing building blocks to sustain cellular growth. However, despite a plethora of studies on the mTOR signaling network, the requirement of basal metabolism—glycolysis and fatty acid (FA) synthesis—for cell growth has not been systematically investigated. The Drosophila model provides a powerful genetic system to address these issues (Ugur et al. 2016), since both the intermediates of this signaling network and the basal metabolic pathways are conserved in the fruit fly (Montagne et al. 2001; Hay and Sonenberg 2004; Padmanabha and Baker 2014; Antikainen et al. 2017; Wangler et al. 2017; Lehmann 2018).

The mTOR protein kinase is present in two distinct complexes, mTORC1 and mTORC2, which comprise raptor and rictor, respectively (Kim et al. 2002; Sarbassov et al. 2005). ATP and amino acid levels control the recruitment of an mTORC1 homodimer at the lysosomal membrane in the vicinity of the small GTPase Rheb (Ras homolog enriched in brain) (Goberdhan et al. 2009; Ma and Blenis 2009; Dibble and Manning 2013; Groenewoud and Zwartkruis 2013; Montagne 2016). Rheb stimulates mTORC1 activity (Yang et al. 2017), which in turn regulates several downstream targets. S6Kinase1 (S6K1) is one such kinase, sequentially activated through the phosphorylation of its T389 and T229 residues by mTORC1 and by PDK1 (Phosphoinositide-dependent protein kinase 1), respectively (Montagne and Thomas 2004; Magnuson et al. 2012). Rheb activation of mTORC1 is repressed by the tumor suppressor TSC (Tuberous sclerosis complex) that comprises subunits TSC1 and TSC2 (Radimerski et al. 2002a; Garami et al. 2003; Inoki et al. 2003a; Dibble et al. 2012). The integrity of mTORC2 is required to sustain the downstream insulin-signaling response (Sarbassov et al. 2005). Binding of insulin-like-peptides (Ilps) to their cognate receptors results in recruitment of class I PI3K (Phosphoinositide 3-kinase) to the membrane. PI3K phosphorylates inositol lipids to phosphatidylinositol-3,4,5-triphosphate (PIP3) (Engelman et al. 2006; Haeusler et al. 2018), while the tumor suppressor PTEN counteracts this process (Cully et al. 2006; Goberdhan et al. 2009). PIP3 constitutes a membrane docking site for the protein kinase Akt whose activity requires the subsequent phosphorylation of its S473 and T308 residues by mTORC2 and PDK1, respectively (Lien et al. 2017).

Constitutive activation of mTORC1 in mouse embryonic fibroblasts represses the tricarboxylic acid (TCA) cycle and stimulates anaerobic glycolysis and biosynthesis of FAs and cholesterol (Duvel et al. 2010). Conversely, adipose-specific knockout of raptor to impede mTORC1 formation, results in enhanced uncoupling of mitochondrial activity (Polak et al. 2008). The increased lipogenesis observed in mTORC1 stimulated cells depends on the cofactor Lipin 1 and on an SREBP (Sterol responsive element binding-protein) family member (Duvel et al. 2010; Peterson et al. 2011). Congruently, TSC2 mutant cells become addicted to glucose as a result of mTORC1 hyperactivity (Inoki et al. 2003b) and, in the context of mTORC1 inhibition, promote glutamine anaplerosis (Choo et al. 2010; Csibi et al. 2013, 2014). Ilp signaling also impinges on basal metabolism. Intracellular activation of Akt increases ATP levels (Hahn-Windgassen et al. 2005; Robey and Hay 2009) through the stimulation of GLUT4-mediated glucose uptake (Jaldin-Fincati et al. 2017) and the enhancement of the expression and activity of glycolytic enzymes (Gottlob et al. 2001; Houddane et al. 2017). Akt also dampens glucose production by suppressing PEPCK (gluconeogenesis), glucose-6-phosphatase (glycogenolysis), and the glycogen synthesis repressor GSK3 (Nakae et al. 2001; McManus et al. 2005). In addition, Akt promotes mitochondrial metabolism and oxidative phosphorylations (Gottlob et al. 2001; Majewski et al. 2004). Conversely, hepatic knockout of the mTORC2 specific-subunit rictor results in constitutive gluconeogenesis and impaired glycolysis and lipogenesis (Hagiwara et al. 2012; Yuan et al. 2012). Taken together, these studies strongly emphasize the role of mTOR in metabolic-related diseases and in adjusting metabolism to the nutritional and energy status (Mossmann et al. 2018).

Despite a plethora of studies, the net requirement of these metabolic changes to sustain overgrowth induced by over-activation of mTOR signaling has not been formally addressed. This signaling network is conserved in Drosophila (hereafter called as TOR), providing a powerful genetic system to address these issues. In the present study, we investigated the requirement of glycolysis and FA synthesis for the cellular growth induced by hyperactivation of TORC1 signaling and Ilp response in Drosophila. Consistent with previous studies (Radimerski et al. 2002a, 2002b; Dong and Pan 2004; Montagne et al. 2010; Pallares-Cartes et al. 2012), we show that TORC1- and Ilp-dependent overgrowth can be independently modulated in the Drosophila fat body (FB), the organ that fulfills hepatic and adipose functions (Padmanabha and Baker 2014; Antikainen et al. 2017; Lehmann 2018). We show that ubiquitous over-activation of TORC1 or Ilp signaling affects basal metabolism and that metabolic restriction at the organismal level has dramatic consequences on animal survival, but minor effect at the cell-autonomous level. This suggests that within an organism, alternative strategies may operate to compensate local metabolic defects. Finally, at the cell-autonomous level, metabolic restriction can partially restrain overgrowth dependent on hyperactivation of Ilp signaling but not TORC1 signaling, indicating that the potential compensatory metabolic pathways do not fully operate in the context of Ilp-signaling stimulation.

Materials and methods

Genetics and fly handling

Fly strains: P[w[+mC]=tubP-GAL80]LL10, P[ry[+t7.2]=neoFRT]40A, daughterless(da)-gal4, tub-gal80^ts, UAS-Dcr-2 from BDSC (bdsc.indiana.edu); hs-flp, tub-gal4, UAS-nls-GFP/FM7, FASN^1-2 (Garrido et al. 2015); Tor^ΔP (Zhang et al. 2000); Tor^2L1, Tor^2L19, and PTEN^- (Oldham et al. 2000); EP(UAS)-Rheb (Stocker et al. 2003); inducible interfering RNA (UAS-RNAi) lines from NIG (shigen.nig.ac.jp/fly/nigfly) to PTEN (5671R-2), from VDRC (stockcenter.vdrc.at) to FASN1 (29349), PFK1 (3017), PK (49533) PDH (40410), LDH (31192) (Dietzl et al. 2007). To generate MARCM clones in a Minute background, the FRT40/P(arm-LacZ w⁺) chromosome (Bohni et al. 1999) was recombined with the P[w[+mC]=tubP-GAL80]LL10, P[ry[+t7.2]=neoFRT]40A chromosome (Lee and Luo 2001). Standard media: agar (1 g), polenta (6 g), and yeast (4 g) for 100 ml. Lipid- (beySD) and sugar-complemented media were as described (Garrido et al. 2015). FASN^1-2 mutant larvae were collected at L1 hatching on grape juice plates following the lack of a GFP-labeled balancer and transferred to fresh tubes. Prepupae were collected once a day to evaluate developmental delay and to measure bodyweight and size.

Molecular biology and Biochemistry

RNAi-knockdown efficacy (Supplementary Figure S1): UAS-Dcr-2;da-gal4,tub-gal80^ts females mated with UAS-RNAi males were let to lay eggs overnight. tubes were kept at 19°C for 2 days, then transferred at 29°C; 2 days later, larvae of roughly the same size were collected for quantitative PCR as described (Parvy et al. 2012). For western blotting, protein extracts were prepared as described and antibodies were previously described for dS6K and phospho-S6 (Montagne et al. 2010; Romero-Pozuelo et al. 2017) or commercially provided for phospho-dS6K, Akt, and ERK (Cell signaling 9209, 9272, 4054, 4695, 4370).

For metabolic measurements, tubes containing 24-h egg collection were maintained at 29°C to strengthen the gal4/UAS effect. Larvae were selected prior to L2/L3 transition and transferred on either a standard diet or a 20%-sucrose supplemented diet (20%-SSD). Collection of prepupae and metabolic measurements were performed as described (Garrido et al. 2015). For protein and triacylglycerol (TAG) measurements, prepupae were weighted and crushed in 400 µl PBT (PBS, 0.1% Triton X-100) and centrifuged for 5 mn at 12,000 rpm. For proteins, 10 µl of extract mixed with 1 ml of 5×-diluted reagent (Bio-Rad Protein Assay Kit) were incubated 15 mn in dark at room temperature; OD was measured at 595 ηm. For TAGs, 20 µl of extract mixed with 500 µl of reagent (Kit STANBIO Triglyceride LiquiColor) were incubated 15 mn at 37°C; OD was measured at 500 ηm. For trehalose and glycogen, prepupae were crushed in 350 µl Na₂${\text{CO}}_3^{-}$ (0.25 M) (PBS 0.1% Triton X-100). After 2-h incubation at 65°C, 700 µl of Na Acetate (0.3 M) and 130 µl of Acetic acid (1.7 M) were added and tubes were centrifuged for 10 mn at 12,000 rpm. For each sample, three sets of 20 µl of extract were incubated overnight at 37°C, two of them contained 1 U/ml of either Trehalase or Amyloglucosidase (SIGMA). Next, all samples mixed with 500 µl of glucose oxidase [Thermo Electron, GLUCOSE (GDO-POD)] were incubated 20 mn at 37°C; OD was measured at 510 ηm. The net concentration of trehalose and glycogen was obtained by subtracting the concentration values of the extracts without enzyme.

Clonal analysis

MARCM clones (Lee and Luo 2001) were generated from 6- to 7-h egg collections heat-shocked for 65 mn at 38°C. FB of late L3 feeding larvae were dissected, fixed, and labeled with Phalloidin (membranes) and DAPI (nuclei) as described (Garrido et al. 2015). Image was acquired on a Leica SP8 confocal laser-scanning microscope. Cell size calculation has been performed as described (Garrido et al. 2015) from at least eight larvae per genotype, containing a few MARCM clones surrounded by control cells. Because of an extremely low ratio of larvae that fully develop and harbor mitotic clones, the mean values were calculated from several experiments that could not be performed at the same time, but over a long period. Therefore, for the graphs of cell size measurement (Supplementary Figures S2, A and C, Figures 5M and 7M), values are reused when they correspond to the same genotype and conditions, allowing direct comparison between experiments.

Figure 5.

Cell-autonomous requirement of glycolysis for Ilp- but not TORC1-dependent overgrowth. (A–G) MARCM clones labeled by GFP (green) in the FB of L3 larvae. Nuclei were labeled with DAPI (silver) and membranes with phalloidin (red). Genotypes of MARCM clones are: PFK1-RNAi (A), PK-RNAi (B), LDH-RNAi (C), PDH-RNAi (D), Rheb⁺,PFK1-RNAi (E), Rheb⁺,PK-RNAi (F), Rheb⁺,LDH-RNAi (G), Rheb⁺,PDH-RNAi (H), PTEN^−/−;PFK1-RNAi (I), PTEN^−/−;PK-RNAi (J), PTEN^−/−;PDH-RNAi (K), and PTEN^−/−;PDH-RNAi (L). Scale bars: 50 μm. (M) Relative size of clonal cells corresponding to the clones shown in (A–L), and in Figure 1A for control (Co).

Figure 7.

Cell-autonomous requirement of FASN activity for Ilp- but not TORC1-dependent overgrowth. (A–L) MARCM clones labeled by GFP (green) in the FB of L3 larvae fed either a standard (A–C, G–I) or a 20%-SSD (D–F, J–L). Nuclei were labeled with DAPI (silver) and membranes with phalloidin (red). Genotypes of MARCM clones are: Rheb⁺ (A, D), PTEN^−/− (B, E) PTEN^−/−,Rheb⁺ (C, F), FASN^1-2;Rheb⁺ (G, J) FASN^1-2, PTEN^−/− (H, K), and the FASN^1-2, PTEN^−/−;Rheb⁺ (I, L). Scale bars: 50 μm. (M) Relative size of clonal cells corresponding to the clones shown in (A–L) and in Supplementary Figure S5, A–D for FASN^1-2 and PTEN-RNAi clones and in Figure 1A for control (Co).

Data availability

Fly stocks and reagents are available upon request. Supplementary materials include Supplementary Figures S1–S5, statistical analysis (methods and Supplementary Tables S1–S7) and supdata/script files.

Supplementary material is available at figshare DOI: https://doi.org/10.25386/genetics.13238312.

Results

TORC1 and Ilp signaling independency in the FB

To investigate the relationship between TORC1- and Ilp-dependent overgrowth in the FB, we generated somatic clones either over-expressing Rheb (Rheb⁺) (Stocker et al. 2003) or PTEN homozygous mutants (PTEN^−/−) (Oldham et al. 2000). The precursors of FB cells divide in the embryo; during larval life, the differentiated cells do not divide but endoreplicate their DNA content to reach a giant size (Edgar and Orr-Weaver 2001). Therefore, to precisely evaluate the effect on cell growth, somatic recombinations were induced during embryogenesis at the stage of proliferation of the FB cell precursors and the resulting MARCM clones were analyzed in the FB of late feeding L3 larvae, prior to the wandering stage that precedes metamorphosis entry. In accordance with previous studies manipulating Ilp and TOR signaling in FB cells (Britton et al. 2002; Scott et al. 2004), both PTEN^−/− and Rheb⁺ clonal cells were bigger than the surrounding control cells and this cell size effect was dramatically increased in PTEN^−/−; Rheb⁺ combined clones (Figure 1, A–D and Supplementary Figure S2A). We next analyzed this growth increase in the context of previously reported Tor mutations, the Tor^ΔP deletion (Zhang et al. 2006), and the Tor^2L1 and Tor^2L19 alleles that result in a kinase inactive and in a short truncated protein, respectively (Oldham et al. 2000). Given that we could not find mutant clones, which were likely eliminated by cell competition, we generated somatic clones in a Minute background to slow down the growth of the surrounding control cells. In these conditions, any of the Tor mutant cells exhibited a dramatic size reduction (Figure 1, G, J, and M, Supplementary Figure S2C) and this phenotype was dominant in Rheb⁺ combined clonal cells (Supplementary Figure S2C and compare Figure 1, E to H, K and N). In contrast, Tor^ΔP and Tor^2L19 but not Tor^2L1 exhibited a clear dominant phenotype over the PTEN^−/− mutation; the size of Tor^ΔP, PTEN^−/− and Tor^2L19, PTEN^−/− clonal cells was dramatically reduced (Supplementary Figure S2C and compare Figure 1, F to L and O), whereas Tor^2L1, PTEN^−/− clonal cells were bigger than the surrounding cells (Figure 1I) and tend to be relatively bigger than control clonal cells (Supplementary Figure S2, B and C). These findings indicate that the Tor^2L1 mutation restrains TORC1- but not Ilp-dependent overgrowth, whereas Tor^ΔP and Tor^2L19 affects both.

Figure 1.

TORC1- and Ilp-dependent growth in FB cells. MARCM clones labeled by GFP (green) in the FB of L3 larvae. Nuclei were labeled with DAPI (silver) and membranes with phalloidin (red). Control (A), Rheb⁺ (B), PTEN^−/− (C) and PTEN^−/−;Rheb⁺ (D) clones were generated in a wild-type background. Rheb⁺ (E), PTEN^−/− (F), Tor^2L1 (G) Tor^2L1, Rheb⁺ (H) Tor^2L1, PTEN^−/− (I), Tor^ΔP (J), Tor^ΔP, Rheb⁺ (K), Tor^ΔP, PTEN^−/− (L) Tor^2L19 (M) Tor^2L19, Rheb⁺ (N), and Tor^2L19, PTEN^−/− (O), clones were generated in a Minute (M) background. Scale bars: 50 μm.

Next, we used phospho-specific antibodies in immunostaining assays to analyze the phosphorylation of Akt (P-Akt) and of the dS6K target, ribosomal protein rpS6 (P-S6). The P-Akt intracellular signal was increased in PTEN^−/− clones, absent in Tor^ΔP and Tor^2L19 clones, but not affected in Tor^2L1 clones (Figure 2, A–D). Staining with the rpS6 phospho-specific antibody revealed a patchy signal, with only a subset of cells expressing the P-S6 signal in the FB (Figure 2, E–H), a pattern previously described in the wing imaginal disc (Romero-Pozuelo et al. 2017). Therefore, to evaluate TORC1 activity, we measured the ratio of P-S6 positive cells among the population of GFP⁺ clonal cells. About half of the control clones were P-S6 positive (Figure 2, E and M), whereas most of the Tor^2L1, Tor^2L19, and Tor^ΔP clones were P-S6 negative (Figure 2, F–H and M). Importantly, almost all the Rheb⁺ clones were P-S6 positive (Figure 2, I and M), whereas the ratio of P-S6 positive cells was slightly but not significantly increased in the PTEN^−/− clones (Figure 2, J and M). Taken together, these findings reveal that the Tor^2L1 mutation affects only TORC1, whereas the Tor^ΔP and Tor^2L19 mutations affect both signaling branches and support the notion that Ilp- and TORC1-dependent overgrowth can be independently modulated in FB cells.

Figure 2.

TORC1 and Ilp signaling activity in FB cells. (A–J) MARCM clones labeled by GFP (green) in the FB of L3 larvae. Clones were generated in a wild type (A, E, I, J, K, L) or a Minute (B, C, D, F, G, H) background and nuclei were labeled with DAPI (silver). FB tissues with PTEN^−/− (A), Tor^2L1 (B), Tor^ΔP (C), Tor^2L19 (D), and FASN^1-2 (K) clones were stained with a phospho-AKT antibody. FB tissues with control (E), Tor^2L1 (F), Tor^ΔP (G), Tor^2L19 (H) Rheb⁺ (I), PTEN^−/− (J), and FASN^1-2 (L) clones were stained with a phospho-S6 antibody. Scale bars: 50 μm. (M) Percentage of P-S6 positive clones with respect to the total number of MARCM clones for control, FASN^1-2, PTEN^−/−, Rheb⁺, Tor^2L1, Tor^ΔP, and Tor^2L19 genotypes.

Activating TORC1 or Ilp signaling impacts basal metabolism

To evaluate how TORC1 and Ilp affect basal metabolism in Drosophila, we analyzed various metabolites in whole animals that express the ubiquitous da-gal4 driver to direct Rheb overexpression (Rheb⁺⁺) or PTEN knockdown (PTEN-RNAi). Larvae were fed either a standard or a 20%-SSD and 0–5 h prepupae were collected, as this is a convenient phase to stage the animals after the feeding period. When fed a standard diet, a high rate of lethality was observed for Rheb⁺⁺ and PTEN-RNAi larvae, although a sufficient number of prepupae could be recovered. Given that none of the Rheb⁺⁺ and PTEN-RNAi larvae fed a 20%-SSD reached the prepupal stage, larvae were first fed a standard diet and transferred onto a 20%-SSD at the L2/L3 molting transition. In this way, we could recover a few prepupae for metabolic measurements. For both males and females fed a standard diet, the bodyweight of Rheb⁺⁺ and PTEN-RNAi prepupae was roughly similar to that of controls (Figure 3, A and B). Conversely, providing a 20%-SSD resulted in a drop of the prepupal weight of control animals that was significantly compensated in Rheb⁺⁺ and PTEN-RNAi prepupae (Figure 3, A and B).

Figure 3.

Enhanced TORC1 or Ilp signaling affects larval metabolism. (A, B) Bodyweight of female (A) and male (B) prepupae formed from larvae fed either a standard (0%) or a 20%-SSD (20%) as from the L2/L3 transition. (C–F) Measurement of total protein (C), TAG (D), glycogen (E), and trehalose (F) levels in prepupae fed either a standard or a 20%-SSD. Prepupae used in these measurements were the F1 progeny from da-gal4;UAS-Dcr-2 females mated to either control (Co), EP(UAS)-Rheb (Rheb⁺⁺), or UAS-PTEN-RNAi (PTEN-Ri) males.

Next, we measured the total amounts of protein, TAG, glycogen, and trehalose—the most abundant circulating sugar in Drosophila. Although protein levels did not vary in PTEN-RNAi and Rheb⁺⁺ animals fed a standard diet, a mild but significant increase was observed in PTEN-RNAi animals fed a 20%-SSD (Figure 3C). TAG levels were strongly increased in animals fed a 20%-SSD (Figure 3D). When fed a standard diet, TAG levels did not vary in PTEN-RNAi and Rheb⁺⁺ animals, whereas, when fed a 20%-SSD, they were strongly increased in PTEN-RNAi animals (Figure 3D). Feeding larvae a 20%-SSD resulted in a moderate increase in glycogen and trehalose levels in control prepupae (Figure 3, E and F). In Rheb⁺⁺ and, in lower extent, in PTEN-RNAi prepupae, glycogen levels were significantly lower than those measured in controls (Figure 3E). Finally, trehalose levels were strongly decreased in both Rheb⁺⁺ and PTEN-RNAi prepupae fed either a standard or a 20%-SSD as compared to control (Figure 3F). Taken together, these findings suggest that a ubiquitous increased activity of either TORC1 or Ilp signaling provokes an apparent increase in sugar consumption. This metabolic rate is correlated with a relative increase in bodyweight for larvae fed a 20%-SSD, but not for those fed a standard diet. We previously observed that increasing dietary sucrose induced a reduction in food intake (Garrido et al. 2015) that may account for the bodyweight reduction of control animals. Potentially, food intake could be less affected in Rheb⁺⁺ and PTEN-RNAi animals, thereby leading to a compensatory effect on bodyweight. Measuring food intake in Rheb⁺⁺ or PTEN-RNAi larvae was not applicable since most of them die during larval stage and thus, terminate feeding earlier. In sum, our data indicate that basal metabolism is altered in the few Rheb⁺⁺ or PTEN-RNAi larvae that survive and further suggests that in most cases stronger metabolic disruption may happen, resulting in lethal homeostatic defects.

Knocking-down glycolysis at the whole body level

To determine whether the basal energy metabolism affected TORC1 and/or Ilp signaling, we ubiquitously expressed RNAis against phosphofructokinase1 (PFK1-RNAi), pyruvate kinase (PK-RNAi) pyruvate dehydrogenase (PDH-RNAi), and lactate dehydrogenase (LDH-RNAi). PFK1 catalyzes the third glycolytic reaction to form fructose 1,6-bisphophate; PK catalyzes the final glycolytic reaction to form pyruvate; PDH directs the mitochondrial fate of pyruvate, whereas LDH directs its anaerobic fate (Figure 4A). When directed with the ubiquitous da-gal4 driver, PK-RNAi provoked early larval lethality, PFK1-RNAi and PDH-RNAi provoked larval lethality at L2 or L3 stages, whereas LDH-RNAi induced a semi-lethal phenotype at larval or pupal stages (Figure 4B).

Figure 4.

Glycolysis knockdown in whole organisms. (A) Scheme of basal metabolism. Glucose and trehalose enter glycolysis as glucose-6P, whereas fructose follows a distinct pathway to triose-P. Enzymes investigated in the present study are marked in red. (B) Phenotype of ubiquitous RNAi-knockdown of PFK1, PK, LDH, and PDH. Flies were left to lay eggs overnight either at 29°C (column 0 h) or at 19°C and transferred to 29°C the day after (column 24 h); then development proceeded at 29°C (i.e. the temperature that inactivates Gal80). (C) Western blot analysis of total (top) or phosphorylated (mid) dS6K (left) or Akt (right) proteins; tubulin (bottom) was used as a loading control. Protein extracts were prepared with late L2 larvae either control (Co), or expressing RNAi against the indicated metabolic enzymes. (D) Survival at 29°C of male (top) and female (bottom) control flies or flies expressing RNAi against the indicated metabolic enzymes as from adult eclosion.

Next, we monitored the phosphorylation of the Drosophila S6Kinase (dS6K), Akt, and ERK. To circumvent the early lethality, the activity of the da-gal4 driver was blocked until early L1 stage using the thermo-sensitive Gal4 inhibitor, Gal80^ts, then, RNAi expression was induced by a temperature shift at 29°C; protein extracts were prepared 2 days later using late L2 larvae. At this stage, the larvae were still viable, although those expressing PK-RNAi did not undergo L2/L3 transition and eventually died (Figure 4B). Western blotting revealed that RNAi-knockdown of PFK1 or LDH did not affect phosphorylation of dS6K, Akt, or ERK (Figure 4C, Supplementary Figure S3, A–F). In contrast, knockdown of PK or PDH decreased dS6K phosphorylation, whereas PK- but not PDH-knockdown mildly decreased Akt phosphorylation (Figure 4C, Supplementary Figure S3, A–D). Unexpectedly, knockdown of PK and LDH, and in lower extent knockdown of PFK and PDH, increased the amounts of total ERK protein with minor effects on its phosphorylation rate (Figure 4C, Supplementary Figure S3, E  and F). These results indicate that ubiquitous knockdown of aerobic glycolysis specifically affects TORC1 signaling.

To evaluate the requirement of glycolysis for adult survival, RNAi-knockdown was induced by temperature shift to 29°C in newly emerged flies and lethality was counted every second day. In both males and females, PK and PFK1 knockdown provoked lethality between 10 and 14 days after temperature shift (Figure 4D). Knockdown of PDH and LDH also induced adult lethality, although not as soon as PK and PFK1 knockdown (Figure 4D). As a comparison, to evaluate the consequence of disrupting FA synthesis, we knocked-down FASN1 (Figure 4A) in adults; about a quarter of FASN1-RNAi flies died between 10 and 14 days, while the others survived nearly as well as control flies (FASN1i in Figure 4D). Taken together, these data indicate that glycolysis is essential for both larval development and adult survival.

Cell-autonomous requirement of glycolysis for Ilp- but not TORC1-dependent overgrowth

To investigate the requirement of glycolysis to sustain cell-autonomous overgrowth dependent on Ilp- and TORC1-signaling, PFK1-RNAi, PK-RNAi, PDH-RNAi and LDH-RNAi were induced in PTEN^−/− or Rheb⁺ clones. Except a moderate effect of PK-RNAi, clones expressing RNAi against these metabolic enzymes did not significantly affect the growth of FB cells (Figure 5, A–D and M). In combined clones, none of the RNAi affected the growth of Rheb⁺ clones (Figure 5, E–H and M). In contrast, the size of PTEN^−/− clones was significantly decreased when co-expressing RNAi against any of these metabolic enzymes (Figure 5, I–M). These findings indicate that both aerobic and anaerobic glycolysis are required to sustain cell-autonomous overgrowth dependent on Ilp signaling. In contrast, reducing glycolysis does not counteract cell-autonomous overgrowth dependent on TORC1 signaling.

Linking FA synthesis to TORC1 and Ilp signaling

Since glycolysis and FA synthesis are tightly connected metabolic pathways (Garrido et al. 2015), we investigated whether the deficiency of the latter affects Ilp or TORC1 signaling. The Drosophila genome encodes three FASN genes, FASN1 is ubiquitously expressed but not FASN2 or FASN3 (Parvy et al. 2012; Chung et al. 2014; Wicker-Thomas et al. 2015). The deletion of the FASN1 and FASN2 tandem (FASN^Δ24-23 deletion, hereafter called FASN^1-2) results in a lethal phenotype that can be rescued by feeding larvae a lipid-complemented diet (beySD) (Garrido et al. 2015; Wicker-Thomas et al. 2015). However, beySD-rescued FASN^1-2 mutant larvae exhibited a developmental delay to enter metamorphosis (Figure 6A). Furthermore, when beySD-rescued FASN^1-2 mutant larvae were transferred at the L2/L3 larval transition onto a 10% sucrose-supplemented-beySD, only a few of them completed the third larval stage and, after an extreme developmental delay, entered metamorphosis (Figure 6A). Delay in development can be due to a default in ecdysone production that results in giant pupae (Parvy et al. 2014) or to impaired TOR signaling that results in reduced body growth (Montagne et al. 1999; Oldham et al. 2000). Measurements of prepupal weight revealed that FASN^1-2 mutant prepupae exhibited a severe reduction in bodyweight and body size, whether or not they were supplemented with sucrose (Figure 6B, Supplementary Figure S4), suggesting a defect in TOR signaling. Therefore, we analyzed the phosphorylation of the Drosophila S6Kinase (dS6K) and Akt in protein extracts of late feeding L3 larvae. Western blotting revealed that the dS6K protein resolved in several bands in FASN^1-2 extracts, whereas Akt protein was unchanged (Figure 6C, Supplementary Figure S3, G and I). These results suggest that dS6K but not Akt might be degraded in the FASN^1-2 mutant background. In addition, dS6K phosphorylation decreased in FASN^1-2 extracts and became barely detectable when FASN^1-2 larvae were fed a sucrose-supplemented-beySD (Figure 6C, Supplementary Figure S3H). Conversely, the phosphorylation of Akt was unaffected in larvae fed a beySD, although it was slightly decreased in larvae fed a sucrose-supplemented-beySD (Figure 6C, Supplementary Figure S3J). This finding contrasts with our previous observation showing that FB explants of FASN^1-2 mutant larvae were hypersensitive to insulin (Garrido et al. 2015). However, FASN^1-2 mutants also exhibited a decrease in food intake (Garrido et al. 2015), which might induce a systemic suppression of dS6K phosphorylation, while FB explant was cultured in nutrient media supplemented with insulin. Therefore, to determine whether FASN mutation affects TOR signaling at the cell-autonomous level, we analyzed P-S6 and P-Akt in FASN^1-2 mutant clones in the FB. As for control clones, about half of the FASN^1-2 clonal cells were P-S6 positive (Figure 2, L and M). Furthermore, no effect on P-Akt was observed in FASN^1-2 clonal cells (Figure 2K). In summary, these findings reveal that disrupting FA synthesis does not significantly affect TORC1 and Ilp signaling at the cell-autonomous level, although it seems to impinge on TORC1 signaling when inhibited in the whole animal whether directly or indirectly.

Figure 6.

FASN^1-2 mutation affects developmental growth and TORC1 signaling. (A) Developmental duration from egg laying to metamorphosis onset of w¹¹¹⁸ control (Co) and FASN^1-2 (FASN) larvae fed either a beySD (0%) or a 10% sucrose-supplemented-beySD as from the L2/L3 transition (10%); n: total number of larvae collected for each condition. Each condition and genotype significantly differ from each other, either by final frequency or dynamic of pupariation (Supplementary Table S6). (B) Prepupal weight of females (left) and males (right) as listed in 6A; the numbers of weighted prepupae are indicated above each sample. (C) Western blot analysis of (from top to bottom) total dS6K, phosphorylated dS6K, total Akt, phosphorylated Akt and total tubulin as a loading control. Protein extracts were prepared from feeding L3 larvae prior to the wandering stage as listed in 6A. For each condition, at least 30 larvae were used to prepare protein extracts.

Cell-autonomous requirement of FA synthesis for Ilp- but not TORC1-dependent overgrowth

To determine, whether FA synthesis is required at the cell-autonomous level to sustain TORC1- and/or Ilp-dependent growth, we analyzed FASN^1-2 clones while enhancing either TORC1 or Ilp signaling in FB cells. We previously reported (Garrido et al. 2015) that FASN^1-2 clonal cells in the FB were slightly reduced in size and that this effect was dramatically increased in larvae fed a 20%-SSD (Supplementary Figure S5, A and B, Figure 7M). Therefore, we generated PTEN^−/− and Rheb⁺ clones combined or not with the FASN^1-2 mutation and analyzed them in the FB of larvae fed either a standard diet or a 20%-SSD. As compared to the standard diet, feeding larvae a 20%-SSD had no effect on the size of Rheb⁺ clonal cells, but significantly reduced the size of PTEN^−/− and of PTEN^−/−;Rheb⁺ clonal cells (Figure 7, A–F and M). Furthermore, when combined with the FASN^1-2 mutation, PTEN^−/− but not Rheb⁺ clones were significantly reduced in size (Figure 7, G, H and M). The FASN^1-2 mutation also provoked a severe size reduction of PTEN^−/−,Rheb⁺ clones (Figure 7, I and M). Moreover, as compared to the standard diet, feeding larvae a 20%-SSD induced a significant size reduction of FASN^1-2;Rheb⁺, FASN^1-2,PTEN^−/− and FASN^1-2,PTEN^−/−;Rheb⁺ clonal cells (Figure 7, J–L and M). Of note, except for the FASN^1-2;Rheb⁺ clonal cells in larvae fed a 20%-SSD that exhibited a size roughly identical to that of the surrounding control cells (Figure 7J), the cell size was always bigger than the controls (Figure 7M). These findings indicate that, in larvae fed a standard diet, FA synthesis is at least in part required to sustain over-growth induced by Ilp, but not TORC1. Nonetheless, the metabolic dependency of Ilp-induced overgrowth might be a consequence of the stronger phenotype of PTEN^−/− clones, which could be more sensitive to metabolic restriction than the one of Rheb⁺ clones. To challenge such an unspecific effect, we made use of PTEN-RNAi that diminish but do not abolish its expression. In this setting, the size of PTEN-RNAi clonal cells was significantly higher than the one of control clones, but not as much as the one of Rheb⁺ clonal cells (Figure 7M, Supplementary Figure S5C). Importantly, the size of PTEN-RNAi clonal cells was significantly diminished when combine to the FASN deficiency (Figure 7M, Supplementary Figure S5D), which excludes a strength-phenotype-dependent effect. In conclusion, overgrown cells generated by Ilp over-activation have limited homeostatic abilities to adjust to metabolic restriction, whereas TORC1 activated cells at least in part maintain these abilities.

Discussion

In this study, we have investigated the functional links between the glycolysis/FA synthesis axis and TORC1- or Ilp-dependent growth in Drosophila. Consistent with previous studies (Radimerski et al. 2002a, 2002b; Dong and Pan 2004; Montagne et al. 2010; Pallares-Cartes et al. 2012), we show that TORC1- and Ilp-dependent overgrowth can be independently modulated in the Drosophila FB. TORC1, once recruited at the lysosomal membrane, is activated by Rheb to phosphorylate downstream targets (Montagne 2016). TORC2, which comprises rictor, resides in the Ilp signaling cascade to phosphorylate the S505 residue in the hydrophobic motif (HM) of Akt. However, a study on a Drosophila rictor mutant reported that S505 phosphorylation by TORC2 was not required to sustain Akt-dependent growth and proposed that TORC2 plays as a rheostat for this process (Hietakangas and Cohen 2007). Here, we provide evidence that the previously described Tor^2L1 mutation, which likely results in a kinase-inactive protein (Oldham et al. 2000), affects TORC1 but not Ilp-dependent overgrowth, whereas both overgrowth processes are suppressed by the Tor^ΔP or Tor^2L19 mutations that likely represent null mutants. Surprisingly, we observe that the S505 residue in the Akt HM is still phosphorylated in Tor^2L1, but not Tor^ΔP and Tor^2L19 mutant clones. Taken together, these findings reveal that the TOR protein, but not its kinase activity, is required for Akt HM phosphorylation and for Ilp-dependent overgrowth through a molecular mechanism that remains to be elucidated.

Intriguingly, our study reveals apparent contradictory effects between perturbations at the whole body and cell-autonomous levels. At the organismal level, knockdowns of glycolytic enzymes to mimic drug treatment with a systemic inhibitor result in animal death. Knockdown of PDH suppresses dS6K but not Akt phosphorylation suggesting that aerobic glycolysis is crucial for TORC1 signaling. These observations contrast with mammalian studies showing that mTORC1 stimulation promotes anaerobic glycolysis (Duvel et al. 2010), whereas Akt activation promotes aerobic glycolysis (Gottlob et al. 2001). Furthermore, at the cell-autonomous level, repression of either aerobic or anaerobic glycolysis partly restrains Ilp- but not TORC1-dependent overgrowth of FB cells. Taken together, these findings confirm the existence of tight connections between TOR signaling and basal metabolism, although the cell-autonomous effects suggest that this connection integrates interactions operating between neighboring cells or organs. In addition, the growth defect of beySD-rescued FASN null mutants might result from the decrease in TORC1 activity, suggesting that TORC1 but not Ilp signaling relies on FA synthesis. However, at the cell-autonomous level, the deficiency of FASN restrains Ilp- but not TORC1-dependent overgrowth in FB cells. These apparent contradictory findings suggest that the growth defect and the reduction of TORC1 activity in FASN mutants are not due to the addition of cell-autonomous effects but potentially to a systemic regulation, as for instance, Ilp secretion by neurosecretory cells (Rulifson et al. 2002). Alternatively, considering that TORC1 directly responds to nutrients (Oldham et al. 2000), the drop of TORC1 activity may be a consequence of feeding, consistent with the decrease in food uptake previously reported in FASN^1-2 mutants (Garrido et al. 2015). Congruently, a study on the transcription factor Mondo, which stimulates glycolysis and FA synthesis in response to dietary sugar (Mattila et al. 2015; Richards et al. 2017), suggests the existence of a FASN-dependent control of food intake (Sassu et al. 2012). Finally, the drop of TORC1 activity observed in FASN^1-2 mutants may be a consequence of malonyl-CoA accumulation, since mTOR malonylation has been reported to inhibit mTORC1- but not Ilp/mTORC2-dependent activity (Bruning et al. 2018). Whether TOR malonylation and the subsequent decrease in TORC1 activity may also occur in Drosophila should be investigated in the future.

Our study reveals that over-activation of mTORC1 and of Ilp signaling, results in a decrease in glycogen stores and in circulating trehalose, suggesting that activation of either signaling branch enhances glucose consumption to sustain cell growth. However, the activation of neither mTORC1 nor Ilp signaling induces an increase in bodyweight. Ubiquitous activation of TORC1 or Ilp signaling is likely to promote the growth of most cells, but might concurrently perturb endocrine signals dampening overall growth. For instance, previous studies reported that activation of Ilp signaling within the ecdysone-secreting gland, results in a systemic decrease in body growth (Caldwell et al. 2005; Colombani et al. 2005; Mirth 2005). Of note, we observed that larvae fed a 20%-SSD result in pupae with reduced bodyweight, an effect that is partially suppressed when either TORC1 or Ilp signaling is over-activated. The fact that the overall bodyweight of these animals is maintained within a range likely compatible with organismal survival contrasts with the observed high rate of lethality. The decrease in glycogen and trehalose suggests that in these animals, each cell tends to increase its basal metabolism evoking an egoist behavior that might perturb the equilibrium between cell-autonomous and systemic regulation. Thus, in a stressful situation, as when animals are fed a 20%-SSD, the need of a tight adjustment to an unbalanced diet might enhance the distortion between cell-autonomous effects and systemic regulation, resulting in animal death.

A plethora of studies in mammalian cells indicate that mTOR activation directs metabolism toward glucose consumption, storage, and anabolism (Gottlob et al. 2001; Inoki et al. 2003b; Hahn-Windgassen et al. 2005; Duvel et al. 2010; Peterson et al. 2011; Houddane et al. 2017; Jaldin-Fincati et al. 2017; Wipperman et al. 2019). Our study suggests that in the Drosophila larvae, TOR promotes glucose consumption, but an increase in TAG levels is observed only for larvae fed a 20%-SSD when Ilp signaling is ubiquitously over-activated. However, at the cell-autonomous level, we observe that inhibition of glycolysis or FA synthesis restrains neither larval FB cell growth nor overgrowth induced by TORC1 stimulation in these cells. These findings counteract the idea that mTORC1 potentiates a glycolysis/FA synthesis axis (Duvel et al. 2010) to sustain cell growth. To overcome the lack of glycolytic products and of membrane lipids, these cells may benefit of a transfer from neighboring cells and/or might favor alternative metabolic pathways, including glutamine catabolism to feed TCA anaplerosis, which has been shown to be a crucial pathway in mTORC1-stimulated mammalian cells (Choo et al. 2010; Csibi et al. 2013, 2014). Nonetheless, such compensatory processes do not fully operate to sustain Ilp-dependent overgrowth, potentially because these cells might be more sensitive to homeostasis perturbations or because the modulation of Ilp signaling would be required for these compensatory processes.

As a coordinator of growth and metabolism, mTOR plays a central role in tumor development (Dowling et al. 2010; Harachi et al. 2018; Mossmann et al. 2018; Tian et al. 2019). PTEN, the tumor suppressor that counteracts PI3K activity downstream of the Ilp receptor, is deficient in several human cancers (Cully et al. 2006). Mutation of TSC1 or TSC2, which results in mTORC1 hyperactivation, is associated with benign tumors but also with brain, kidney, and lung destructive diseases (Henske et al. 2016). To investigate the role of mTOR regarding tumor development, a recent study reported the generation of liver-specific double knockout mice for TSC1 and PTEN (Guri et al. 2017). These mice develop hepatic steatosis that eventually progresses to hepatocellular carcinoma. Both processes are suppressed in mice fed the mTORC1/2 inhibitors INK128, but not the mTORC1 inhibitor rapamycin, supporting an Ilp/mTORC2 specific effect. The combination of inhibitors against mTOR and metabolism is currently under clinical investigation to fight cancers (Mossmann et al. 2018). Importantly, our study reveals that ubiquitous inhibition of basal metabolism produces dramatic effects during development, while at the cell-autonomous level, it only moderates growth induced by over-activation of Ilp/TORC2 signaling. Therefore, the use of drug therapy to fight cancer must be taken with caution, in particular if organismal development is not complete and most efforts should be made to selectively target sick tissues.