Atg8/LC3 controls systemic nutrient surplus signaling in flies and humans

Summary

Organisms experience constant nutritional flux. Mechanisms at the interface of opposing nutritional states – scarcity and surplus – enable organismal energy homeostasis. Contingent on nutritional stores, adipocytes secrete adipokines, such as the fat hormone leptin, to signal nutrient status to the central brain. Increased leptin secretion underlies metabolic dysregulation during common obesity, but the molecular mechanisms regulating leptin secretion from human adipocytes are poorly understood. Here, we report that Atg8/LC3-family proteins, best known for their role in autophagy during nutrient scarcity, play an evolutionarily conserved role during nutrient surplus by promoting adipokine secretion. We show that in a well-fed state, Atg8/LC3 promotes the secretion of the Drosophila functional leptin ortholog Unpaired2 (Upd2) and leptin from human adipocytes. Proteomic analyses reveal that LC3 directs leptin to a secretory pathway in human cells. We identified LC3-dependent extracellular vesicle loading and secretion (LDELS) as a required step for leptin release, highlighting a unique secretory route adopted by leptin in human adipocytes. In Drosophila, mutations to Upd2’s Atg8 interaction motif (AIM) result in constitutive adipokine retention. Atg8-mediated Upd2 retention alters lipid storage, hunger response, and rewires the bulk organismal transcriptome in a manner conducive to starvation survival. Thus, Atg8/LC3’s bidirectional role in nutrient sensing – conveying nutrient surplus and responding to nutrient deprivation – enables organisms to manage nutrient flux effectively. We posit that decoding how bidirectional molecular switches – such as Atg8/LC3 - operate at the nexus of nutritional scarcity and surplus will inform therapeutic strategies to tackle chronic metabolic disorders.

eTOC blurb

Adipocytes secrete adipokines like leptin to signal nutrient surplus to the brain. Madan et al. uncover that Atg8/LC3 proteins, known for autophagy, also promote adipokine secretion in flies and humans. By pinpointing a specific means to target leptin secretion, this study provides strategies to sensitize leptin signaling in common obesity.

Graphical Abstract

Introduction

Organisms constantly evaluate their nutritional reserves to allocate resources for functions that enhance fitness or conserve energy. Adipokines, secreted by fat cells, provide systemic information on nutrient reserves. Leptin in mammals¹ and its Drosophila ortholog Unpaired2 (Upd2) are primary adipokines released in proportion to fat stores^2–4. They influence brain circuits that control energy expenditure, appetite, and metabolism^5,6. In nutrient-rich states, adipokines signal that energy can be devoted to costly activities like immunity and reproduction. During scarcity, their secretion reduces, signaling an energy deficit.

In humans^2,3 and mice^1,7, starvation reduces leptin levels, crucial for neuroendocrine responses to reduced energy reserves. Injecting leptin into starved mice disrupts neuroendocrine physiology¹. High circulating leptin levels (hyperleptinemia) lead to ineffective brain signaling^8,9. Identifying regulators of leptin secretion may restore adipocyte-brain signaling and improve energy balance. Upd2, a JAK/STAT pathway agonist¹⁰, is a functional ortholog of human leptin in fruit flies⁶. Human leptin can rescue metabolic defects in Upd2-deficient flies, showing evolutionary conservation. We used the fly system to identify cell-intrinsic mechanisms regulating adipokine secretion.

Organisms experience constant nutritional flux, necessitating homeostatic mechanisms to balance scarcity and surplus. We hypothesized that overlapping molecules respond to both states efficiently. Therefore, we explored whether proteins pivotal to nutrient deprivation regulate adipokine secretion. Autophagy, a primary mechanism for adapting to metabolic stress^11,12, controls lipid metabolism and is critical for adipocyte differentiation^13–15. Central players in autophagy are ubiquitin-like proteins, including LC3/GABARAP in mammals and Atg8 in flies and yeast¹⁶. When conjugated to phosphatidylethanolamine (PE)¹⁷, Atg8 regulates autophagosome-lysosome fusion for cargo degradation during nutrient scarcity¹⁸. Beyond degradation, Atg8/LC3 proteins control various cellular processes, such as phagocytosis¹⁹, viral replication²⁰, extracellular vesicle secretion²¹. Atg8 and other autophagy (ATG) genes participate in autophagy-independent trafficking processes²². For example, autophagy genes are key components in the unconventional secretion of yeast Acyl-coA Binding protein 1 (ACB1)²³, mammalian ACB protein²⁴, and interleukins ^25,26.

We examined whether Atg8/LC3 proteins control the secretion of Drosophila Upd2 and human leptin from adipocytes. Our results show that LC3 family proteins promote leptin secretion. The Atg8/LC3 interaction motif (AIM/LIR) in both Upd2 and leptin is critical for their interaction and secretion. Comparative proteomics revealed that leptin is secreted via LC3-mediated extracellular vesicle secretion. Factors involved in autophagy initiation are not necessary for leptin secretion, but those involved in Atg8’s conjugation to membranes are required. We uncover a noncanonical role for Atg8/LC3 in secreting nutrient surplus signals from adipocytes to the brain, suggesting a bidirectional role in nutrient response.

Results

Atg8 regulates Upd2 secretion in Drosophila cells.

We investigated whether Upd2 release from cells is coupled to nutrient state using Drosophila S2R+ cells. Previous functional genomics studies with the Drosophila S2R+ system have provided insights into secretion pathways²⁷, lipid droplet formation²⁸, and nutrient sensing sensing^29,30. These cells respond robustly to amino-acid (AA) deprivation^29,30, allowing us to test the effect of acute nutrient deprivation on Upd2 retention versus secretion. Hence, we set out to assay the effect of acute nutrient-deprivation on Upd2 localization. GFP-tagged Upd2 is functional and can activate the JAK/STAT pathway^6,10,31. Therefore, we transiently transfected cells with Upd2-WT::GFP and assessed GFP localization in cells cultured in complete media (fed) or AA-depleted media (starved). Within 8 hours of AA deprivation, we noted increased Upd2 accumulation, with a maximum observed within 4–12 hours. Refeeding reversed the intracellular accumulation, highlighting the nutrient state’s impact on Upd2’s intracellular retention (Figure 1A, A’, S1).

Figure 1. Nutrient state and autophagy govern adipokine retention vs. secretion in Drosophila S2R+ cells.

A) S2R+ cells transfected with Upd2-WT::GFP (green; anti-GFP), stained with Lamin (magenta) under different nutrient conditions. The ratio of Upd2::GFP nuclear/whole cell intensity is plotted. See Figure S1. B-D) GFP sandwich ELISA on conditioned media to assess Upd2 secretion under various nutrient conditions, drugs, or dsRNA treatments. E) S2R+ cells transfected with Upd2-WT::GFP, stained with Lamin, and co-transfected with indicated dsRNA. F) S2R+ cells transfected with Upd2-WT::GFP (green) and mCherry::Atg8 (magenta). Live cell dyes mark the nucleus. See also Figure S2. Data is shown as mean ± SEM. Statistical significance: Mann Whitney U test (A’, E’) or unpaired t-test (B-D). Scale bars: 5μM (A, E), 2μM (F).

To determine if Upd2’s intracellular accumulation correlates with reduced secretion, as we did in our prior work³², we performed a quantitative ELISA for GFP-tagged Upd2. AA deprivation for 8 hours significantly reduced Upd2 secretion (85%; p=0.0017), and refeeding for 6 hours significantly ameliorated this effect (p=0.0002) (Figure 1B). These results suggest that Drosophila cells possess mechanisms that couple Upd2 secretion to nutrient state.

Considering that autophagy is a well-documented cellular response to starvation³³, we explored its role in restricting adipokine secretion. By inhibiting autophagy with 3-Methyladenine (3-MA)³⁴, we observed a dose-sensitive increase in Upd2 secretion (Figure 1C; 57% increase at 5mM; p=0.08). Conversely, activating autophagy with Torin1³⁵ reduced Upd2 secretion (Figure 1C; 30% reduction at 750nM; p=0.04). These results suggest that autophagy induction antagonizes Upd2 secretion.

LC3, a key autophagy player, is also implicated in LC3-dependent extracellular vesicle loading and secretion (LDELS)²¹. Hence, we hypothesized that LC3/Atg8 controls Upd2 secretion in well-fed cells. Supporting this hypothesis, we observed reduced secretion of Upd2-WT::GFP upon dsRNA-mediated knockdown of Atg8, measured by a quantitative GFP sandwich ELISA (p=0.0002) (Figure 1D). A corresponding increase in Upd2 intracellular accumulation was also observed (Figure 1E, E’). Cytosolic Upd2 puncta co-localize with GM130, a cis-Golgi secretory compartment marker³⁶ (Figure S2A). If Atg8 promotes Upd2 secretion, we would expect Atg8 and Upd2 to co-localize with cytosolic Upd2 puncta in well-fed states. We used live imaging in S2R+ cells co-transfected with mCherry-tagged Atg8 and GFP-tagged Upd2, observing that a subset of Upd2 cytoplasmic puncta exhibits transient co-localization with Atg8 (Figure 1F, S2B). These observations suggest that Upd2 secretion in well-fed Drosophila S2R+ cells depends on the nutrient state and that Atg8 promotes Upd2 secretion.

Upd2’s Atg8-interaction motif (AIM) is required for its secretion from Drosophila cells.

During autophagy, Atg8 recruits specific proteins to the autophagosome via a short linear protein domain called the Atg8-interacting motif (AIM) in Drosophila or the LC3-interacting region (LIR) in humans, with a consensus sequence W/F/Y-X-X-L/I/V³⁷. AIM/LIR sequences are required for LC3 loading of cargos like RNA-binding proteins (RBPs) into EVs during LDELS-based secretion. Point mutations in the RBP’s LIR sequence disrupt LDELS-mediated secretion²¹. We hypothesized that Drosophila Upd2 might be a direct cargo for Atg8. Protein sequence analysis showed that Upd2 has putative AIM sequences (Figure 2A). In vitro pulldown assays with GST-tagged Upd2-WT or Upd2-AIM (single point mutations in the AIM motif) proteins and 35S-labeled Atg8 revealed that Upd2-WT successfully pulled down Atg8, whereas Upd2-AIM’s ability was reduced (Figure 2A). Transfecting S2R+ cells with Upd2-WT::GFP or Upd2-AIM::GFP showed that Upd2-AIM::GFP cells had increased intracellular accumulation (Figure 2B, B’) and reduced secretion compared to Upd2-WT::GFP cells (Figure 2C). These results suggest that Atg8 promotes Upd2 secretion via its AIM sequence, in a nutrient-sensitive manner.

Figure 2. Upd2’s interaction with Atg8 governs its secretion in Drosophila S2R+ cells.

A) GST pulldown assays with IVT Atg8 and GST-tagged Upd2-WT/Upd2-AIM. B) S2R+ cells transfected with Upd2-WT::GFP or Upd2-AIM::GFP, stained with Lamin. Ratio of Upd2::GFP nuclear/whole cell intensity plotted. C) GFP sandwich ELISA to quantify Upd2 secretion with AIM mutation. D) S2R+ cells transfected with Upd2-WT::GFP/Upd2-AIM::GFP and co-transfected with VHH nanobody-tagged constructs. E) GFP sandwich ELISA on conditioned media with various constructs to assess Upd2 secretion. Data shown as mean ± SEM. Statistical significance: Mann Whitney U test (B’, D’) or unpaired t-test. Scale bar: 5μM.

To test the hypothesis that Upd2’s direct interaction with Atg8 promotes its secretion, we used a nanobody-based approach³⁸. Nanobodies are small antibodies (27kDa), used as tags, to study how interactions between GFP-nanobody (V_HH)-tagged proteins and corresponding GFP-tagged proteins alter localization and behavior^39–41. We engineered interactions between secretion incompetent Upd2-AIM::GFP and Atg8 (See Schematic in Figure 2D). Co-expression with V_HH::Myc-Atg8-WT reduced Upd2-AIM accumulation, localizing it to cytosolic puncta, unlike the control nanobody (Figure 2D, bottom panel, D’). This indicates that restoring Upd2-AIM’s interaction with Atg8 reduces its intracellular accumulation. Upd2-AIM secretion, compromised with the control nanobody, improved significantly when reconstituted with Atg8 (Figure 2E). Overall, Atg8 interacts with Upd2 via its AIM sequence, promoting Upd2 secretion in fed Drosophila cells.

Atg8 controls Upd2 intracellular accumulation in adult Drosophila adipocytes.

To validate our in vitro findings in vivo, we generated transgenic flies with fat-tissue-specific expression of GFP-tagged Upd2-WT (LppGAL4>UAS-Upd2-WT::GFP) and Upd2-AIM (LppGAL4>UAS-Upd2-AIM::GFP), and performed image-based quantitative analysis of Upd2 accumulation in fly fat. Upd2-WT is detected at low levels in adult Drosophila adipocytes despite overexpression. In contrast, Upd2-AIM is readily detectable with increased intracellular localization (Figure 3A, A’; p=0.0400). Co-expression of V_HH-tagged Atg8 with Upd2-AIM reduces its intracellular accumulation to Upd2-WT levels (Figure 3A, bottom panel, A’; p=0.0061). Thus, Upd2-AIM mutation leads to intracellular retention, and nanobody-based reconstitution with Atg8 reduces this accumulation.

Figure 3. In vivo Drosophila model to study Atg8’s role in regulating Upd2’s intracellular retention.

A) Transgenic flies co-expressing UAS-Upd2::GFP and UAS-VHH under Lpp-GAL4 driver. B) CRISPR-tagged Upd2-WT::HA abdominal fat under fed/starved conditions. C) CRISPR-tagged Upd2-AIM::HA abdominal fat in fed/starved conditions. D) Clonal adult fat tissue with Atg8-RNAi expressed in a GFP positive -cells (top-panel demarcated by a yellow dashed line), and bottom panel shows a higher magnification of the inset. The HA-signal measures Upd2 levels. n = 7 clones for D’. See also Figure S3. E) Acute Atg8 knockdown in adult fly fat. Data shown as mean ± SEM. Statistical significance: unpaired t-test or Mann Whitney U test. Scale bars: 10μM (A-C, E), 20μM (D). Each dot in the corresponding quantitation (A’-E’) represents either a cell (A’-C’, E’) or an area demarcated by a GFP filter (D’). Adjusted mean intensity of GFP-LUT signal (A’, E’) or HA-LUT signal (B’, C’, D’) was calculated.

Studying Upd2 secretion in intact adult Drosophila has been challenging due to the lack of robust quantification methods. Even sensitive mass spectrometry cannot detect Upd2 in fly hemolymph. Therefore, as in prior work on circulating insulin in flies^42,43, we use Upd2 accumulation within adult adipocytes as a proxy for secretion.

To investigate acute starvation’s effect on Upd2 retention in vivo, we used CRISPR-Cas9 to generate a tagged knock-in, with the endogenous Upd2 gene tagged with HA at the N-terminus. Subjecting these Upd2-WT::HA flies to 4 hours of acute starvation revealed a significant increase in Upd2 levels within adult abdominal adipose tissue (Figure 3B, B’; p<0.0001), suggesting that Upd2 accumulation acutely increases on starvation. We generated point mutations in Upd2’s AIM sequence in the endogenous HA-tagged Upd2 locus, resulting in viable and fertile homozygous Upd2-AIM::HA flies. Confocal microscopy showed a significant increase in intracellular accumulation of Upd2-AIM (Figure 3C, C’; p=0.0170), corroborating our in vitro findings.

Next, we tested whether genetic knockdown of Atg8 phenocopies Upd2-AIM mutations. Using the FLP-out GAL4 technique⁴⁴, we expressed Atg8-RNAi in a subset of adipocytes (GFP-positive) surrounded by wild-type tissue (GFP-negative). HA-tagged endogenous Upd2 significantly accumulated only in Atg8-knockdown patches (Figure 3D, D’; p=0.0009). Control Luciferase-RNAi expression did not increase Upd2 accumulation (Figure S3; p=0.3699). Hence, Atg8 knockdown in a subset of cells is sufficient to increase Upd2 retention. We also acutely knocked down Atg8 in adult fat using the tubGAL80^ts system for a week, observing increased Upd2 accumulation compared to control RNAi (Figure 3E, E’; p=0.0041). Our experiments establish that Atg8 knockdown and Upd2’s AIM mutations result in constitutive Upd2 retention in well-fed adipocytes, phenocopying starvation.

Upd2 intracellular accumulation promotes organismal starvation resilience.

Starvation-induced reduction in Leptin/Upd2 levels is crucial for an organism’s neuroendocrine response to reduced energy reserves; this has been well-documented in humans, mice, and flies^1,6,7,45. For instance, artificially increasing leptin during fasting in mice dysregulates neuroendocrine physiology and decreases survival capacity¹. Similarly, genetic mutants of both upd2⁶ and Drosophila insulin-like peptides (Dilps) display starvation resilience⁴⁶. Overall, this suggests that a reduction in nutrient surplus signaling allows organisms to better adapt to prolonged starvation. The Upd2-AIM mutation increases intracellular Upd2 accumulation, at the expense of secretion. Hence, it allows us to examine the effect constitutive adipokine retention on adult fly physiology during starvation.

First, we sought to examine whether Upd2-AIM mutations confer a starvation survival advantage. To this end, we performed a fat-tissue-specific over-expression of the Upd2-AIM transgene in an upd2Δ mutant background (upd2Δ; LppGAL4> UAS-Upd2-AIM) and compared it with overexpression of a control empty transgene (UAS-Luciferase) and Upd2-wildtype transgene (UAS-Upd2-WT). We found that flies expressing the Upd2-AIM transgene, in adipocytes, live longer on a 1% sucrose-agar-starvation diet than Upd2-WT or control transgene over-expressing flies (Figure 4A; p=0.0216), suggesting that adipokine retention confers a survival advantage during starvation. Similarly, flies harboring a CRISPR-engineered AIM mutation at the endogenous Upd2 locus (Upd2-AIM) also displayed starvation resilience relative to a wild-type control (Figure 4B; p=0.0005). Taken together, these results suggest that adipokine retention provides a survival advantage to an organism during nutrient-poor conditions.

Figure 4. Upd2’s intracellular retention confers a survival advantage upon starvation.

A, B) Survival curves of genotypes on 1% sucrose agar starvation media. C) Differentially expressed genes in Fed vs. Stv states for Upd2-WT::HA and Upd2-AIM::HA. See also Figure S4 and Data S1. D) LipidTOX staining to assess lipid content in Upd2-WT::GFP vs. Upd2-AIM::GFP flies. E) Triacylglycerol (TAG) levels in Upd2-WT::HA vs. Upd2-AIM::HA flies in fed vs. starved states. F) Feeding events measured for Upd2-WT::HA vs. Upd2-AIM::HA flies using FLIC assay. Data shown as mean ± SEM. Statistical significance: Kruskal-Wallis test with Dunn’s multiple comparisons.

To get a handle on potential pathways that provide starvation resilience during constitutive adipokine retention, we surveyed the transcriptomes of Upd2-AIM and Upd2-WT flies by performing bulk RNAseq. We performed principal component analysis (PCA) ⁴⁷ of the RNAseq data, Gene Ontology (GO) analysis ^48,49, and GO enrichment⁵⁰ of differentially expressed (DE) genes (See Data S1). Upd2-AIM flies, even on a baseline diet, showed significant bulk transcriptome changes compared to Upd2-WT (Figure S4A, S4B), suggesting that constitutive adipokine retention results in an altered gene expression profile.

On examining DE genes on overnight starvation, we observed 1613 DE genes on overnight starvation in Upd2-WT flies (Figure 4C, S4C, Data S1A). Strikingly, only 119 DE genes (were detected in Upd2-AIM (7% of the changes in WT state; Figure 4C, S4D, Data S1B). Given the transcriptional response of Upd2-AIM flies to starvation was muted, we examined using GO analysis^48,49 which signature pathways were altered in Upd2-WT versus Upd2-AIM. We observed that in control flies (Upd2-WT), pathways involved in reproductive physiology, innate immunity, and cellular metabolism (specifically fatty acid biosynthesis and pentose phosphate pathway) were altered (Figure S4E). Notably, immune and reproductive pathways were the top enriched categories in the baseline difference between Upd2-AIM and control (Figure S4B). In contrast, in Upd2-AIM flies starvation, the immunity and reproduction pathways were not the top pathways significantly enriched by starvation-induced DE genes (Figure S4F). However, the pentose phosphate pathway and glutathione metabolic process were most significantly altered between fed and starved states in Upd2-AIM (Figure S4F). In yeast, an adaptation strategy to prolonged glucose deprivation is to downregulate the pentose phosphate pathway (PPP); this allows energy to be conserved instead of generating NADPH for nucleotide precursors⁵¹. Consistent with this finding, in adult Drosophila, we observed that during starvation, the PPP genes show reduced expression (Figure S4H). Importantly, we observed that a few PPP genes (CG7140, Rpe) were already downregulated in Upd2-AIM, even in the fed state (Figure S4G). The bulk transcriptome analyses indicate that constant adipokine retention, even in a normal fed state, leads to changes in immune, reproductive, and metabolic pathways that resemble those seen in a starved state. In sum, we conclude that constitutive adipokine retention of Upd2-AIM flies likely results in a transcriptional environment that is conducive to adapting to nutrient depletion.

To explore whether adipokine retention causes fat storage alterations, we visualized lipid content in adult abdominal adipocytes of Upd2-WT and Upd2-AIM flies using a neutral lipid stain. We observed a significant increase (p=0.0006) in lipid accumulation in mutant flies relative to wild-type (Figure 4D, D’). Measurement of neutral triglyceride (TAG) content showed significantly higher TAG in Upd2-AIM flies compared to controls (Figure 4E; p<0.0001). During overnight starvation, both Upd2-WT and Upd2-AIM flies exhibited efficient TAG lipolysis (Figure 4E; p<0.0001). Thus, constitutive adipokine Upd2 retention extends resilience by storing excess energy reserves as neutral fat and efficiently breaking it down during starvation.

After overnight starvation flies display a significant increase in feeding events that we refer to as hunger-driven feeding (HDF) response⁵². Hence, we wondered whether constitutive adipokine retention in Upd2-AIM increases feeding events even at baseline, akin to a starved state. We measured feeding events using the FLIC assay⁵³ in Upd2-WT and Upd2-AIM flies at baseline and after overnight starvation. Upd2-AIM flies showed a significant increase in feeding events compared to baseline (Figure 4F; p<0.0001 for Upd2-WT, p=0.0017 for Upd2-AIM). Even at baseline, Upd2-AIM displayed increased feeding akin to the hunger-driven feeding response of starved flies (Figure 4F; p=0.0029). Overnight starvation did not significantly increase feeding events in Upd2-AIM flies compared to their already hyperphagic baseline behavior (Figure 4F). These results suggest that constitutive adipokine retention causes hyperphagia at baseline, providing advantages in nutrient-deprived states and enhancing starvation resilience in adult Drosophila.

LC3 family proteins are required for leptin secretion in human cells.

Human leptin can functionally substitute for Upd2 function in flies⁶. The evolutionary conservation between Drosophila Atg8 and mammalian LC3-family proteins and the autophagy pathway is well-established^54–57. Therefore, we explored whether our observations regarding the acute changes in nutrient-sensitive Upd2 secretion in Drosophila cells is recapitulated in cultured human cells for leptin. We transiently transfected HEK293T cells with GFP-tagged human leptin. The cells were cultured in either complete media (fed) or nutrient-deprived state (EBSS). A GFP-sandwich ELISA on the conditioned media assayed leptin secretion. Culturing in EBSS for 4 hours resulted in a 72% reduction in leptin secretion (Figure 5A), suggesting nutrient depletion negatively impacts leptin secretion.

Figure 5. LC3/GABARAP is required for Leptin secretion in human cells.

A, B) GFP sandwich ELISA to assess Leptin secretion under different nutrient conditions or drugs. C) GFP pull-down with Leptin-WT::Myc or Leptin-AIM::Myc and EGFP::LC3. D) GFP sandwich ELISA to assess Leptin secretion with various AIM mutations. E, F, F’) HEK293T cells transfected with Leptin-WT or Leptin-AIM, stained for Myc, GM130, and DAPI. Quantification of Leptin’s colocalization with GM130. G, G’) Polarization factor calculation for Leptin::Myc distribution. See also Figure S5. Data shown as mean ± SEM. Statistical significance: unpaired t-test or Mann Whitney U test.

Next, we examined whether pharmacological activation (3-MA)³⁴ or inhibition of autophagy (Torin1)³⁵ impacted leptin secretion. We observed a dose-sensitive increase in leptin secretion upon 3-MA treatment and a dose-dependent decrease in secretion upon Torin treatment (Figure 5B), implying autophagy initiation is antagonistic to leptin secretion. These observations in mammalian cells recapitulate the nutrient-sensitive secretion observed in Drosophila cells for Upd2, where adipokine secretion is negatively regulated by nutrient depletion.

We then examined whether LC3 family proteins regulate human leptin secretion, similar to Atg8’s role in Upd2 secretion. Human leptin has two putative AIM/LIR sequences near its N-terminus (Figure 5C). We generated point mutations to these motifs (AIM1: WgtL→AgtA; AIM2: WpyL→ApyA) and transiently transfected plasmids with Myc-tagged leptin-WT or Myc-tagged leptin with AIM mutations (leptin-AIM) into HEK293T cells, along with GFP-tagged LC3 (EGFP-LC3). Co-immunoprecipitation assays showed that leptin-WT complexes with LC3, but AIM mutations impair leptin’s ability to complex with LC3 (Figure 5C), indicating that leptin complexes with LC3 in an AIM-dependent manner in human cells.

We investigated whether AIM mutations in leptin impact its secretion. Using a quantitative GFP sandwich ELISA, we found that compared to baseline leptin-WT, AIM mutations caused a reduction in secretion for AIM1 (50% reduction; p=0.1367) and a significant reduction for AIM2 (72% reduction; p=0.0228). Mutating both AIM sequences resulted in a significant reduction (93%; p=0.0039) in leptin secretion (Figure 5D), thus leptin’s AIM sequences are important for its secretion.

To understand if AIM mutations reduce secretion due to changes in protein stability, localization, or levels, we examined the localization of leptin-AIM versus leptin-WT in HEK293T cells. Leptin-WT localizes asymmetrically to a perinuclear region positive for the secretory compartment marker GM130 (Figure 5E, 5G’; Figure S5). In contrast, mutations to both AIM motifs result in increased leptin accumulation (Figure 5F; p<0.0001), reduced co-localization with GM130 (Figure 5F’; p=0.0062), loss of directionality of leptin accumulation (Figure 5G, G’; p<0.0001; Figure S5), and diffuse cytosolic localization (Figure 5E). Our findings suggest LC3 is required for leptin’s secretion in mammalian cells.

Leptin is secreted via small extracellular vesicles (sEVs) in human cells.

To understand how leptin is secreted in an LC3-dependent manner, we performed comparative analyses of protein-protein interactors between WT and AIM leptin. We immunoprecipitated (IP) leptin-WT::GFP and leptin-AIM::GFP from HEK293T cells and subjected the immunocomplexes to LC-MS. We then performed unbiased proteomic analyses to identify proteins present specifically in the secretion-competent leptin-WT and absent in the leptin-AIM pulldown, which we further analyzed using the STRING database ^58,59. Based on our analysis cutoff, we found that 31 proteins bound uniquely to leptin-WT but not to leptin-AIM (Figure 6A). These proteins fell into three functional interaction clusters (Figure 6B, C) – Cluster 1 (endoplasmic reticulum (ER) chaperone complex), Cluster 2 (mRNA metabolism) and Cluster 3 (extracellular exosome). Gene ontology (GO) enrichment analysis^48,49 showed that the following GO terms displayed a 10–100-fold enrichment only with leptin-WT: ER chaperone, granule lumen, extracellular exosome, RNA binding proteins (RBPs- spliceosomes) (Figure 6C, Data S2D–F). We were intrigued to observe that 13 proteins involved with mRNA metabolism were enriched only in leptin-WT immunocomplexes (Figure 6B). Specifically, we noted that secretion-competent leptin associates with two RNA binding proteins (Figure-6B-green) — HNRNPK and SNRPA1- secreted by LDELS²¹.

Figure 6. Leptin is secreted via sEVs in human cells.

A) Venn diagram of Leptin-WT::GFP vs. Leptin-AIM::GFP interactors. B) STRING analysis of unique Leptin-WT interactors. See also Data S2. C) GO terms associated with Leptin-WT interactors. Enrichment for the GO term is displayed on the X-axis and the color of the dot represents the significance, with blue indicating an FDR=0.05 and red indicating an FDR<0.0001. D) Western blots from lysates and sEVs in HEK293T cells transfected with Leptin-WT::EGFP or Leptin-AIM::EGFP to demonstrate the presence/absence of leptin in sEVs. Markers for each fraction confirm the purity of sEVs. See also Figure S6.

To confirm the presence of leptin in sEVs, we subjected conditioned media of transiently transfected HEK293T cells to sEV purification by differential centrifugation according to standard protocols ⁶⁰. To assess the size and purity of the sEVs we isolated, we used Transmission Electron Microscopy (TEM) ^60,61. We observed spherical vesicles that correlated with the expected size range of exosome-like vesicles (50–90 nm)⁶² (Figure S6). The sEV fractions were analyzed by western blotting to assess the enrichment of known sEV markers and the absence of intracellular proteins. As expected, the sEV fractions were enriched for markers including TSG101 and Flotillin but not for the ER-resident protein Calnexin (Figure 6D), suggesting that the quality of the sEV prep was appropriate. Though Leptin-AIM::GFP was detected in the lysate, consistent with its significantly reduced secretion (Figure 5D), we were unable to detect it in the sEV fractions (Figure 6D). Altogether, our results suggest that LC3 regulates human leptin secretion, and that leptin is enriched in secreted sEV fractions of human cells in an LC3 dependent fashion.

LC3-mediated extracellular vesicle loading (LDELS) pathway is required for leptin secretion from human primary adipocytes.

To establish that LC3 regulates leptin secretion in a physiologically relevant context, we examined its role in endogenous leptin secretion from human primary subcutaneous adipocytes. We cultured and differentiated human pre-adipocytes to subcutaneous adipocytes, then performed knockdowns of LC3s (LC3A, B, C) and GABARAPs (GABARAP, GABRAPL1, GABRAPL2) using siRNAs validated in previous studies. An ELISA for endogenous leptin showed a significant reduction in leptin secretion upon knockdown of LC3 (p<0.0001), GABARAP (p<0.0001), or both (p<0.0001) (Figure 7A). Adiposity, measured via AdipoRed analysis, was unaffected by siRNA treatment (Figure S7A), suggesting that LC3 family proteins are required for leptin secretion from human primary adipocytes.

Figure 7. Leptin secretion in primary human adipocytes likely occurs via the LDELS pathway.

A-C) Endogenous Leptin ELISA in primary human adipocytes with siRNA treatments targeting LC3/GABARAP (A), autophagy proteins (B), or LDELS pathway proteins (C). See also Figure S7 and Table S1. D) Summary model of Leptin secretion involving LC3 interaction and LDELS pathway. Data shown as mean ± SEM. Statistical significance: unpaired t-test on biological replicates.

Next, we investigated whether LC3 acts alone or in concert with other autophagy (Atg) pathway proteins. We knocked down proteins involved in all stages of autophagy—initiation, phagophore elongation, and autophagosome maturation. Knockdown conditions did not significantly affect adiposity, as assessed by AdipoRed analysis (Figure S7B). ELISA results showed that proteins involved in the early initiation steps of autophagy (Ulk1, RB1CC1/FIP200) were dispensable, whereas knockdown of proteins involved in LC3 lipidation and membrane conjugation reduced leptin secretion (Figure 7B). This indicates that autophagy initiation factors do not regulate leptin secretion, but a subset of Atg-pathway genes involved in LC3 lipidation and membrane expansion are required.

LDELS²¹ is one of the possible pathways by which LC3 and Atg-factors mediate secretion^63,64. Our proteomic data revealed that leptin associates with RNA-binding proteins adopting an LDELS route (Figure 6, Data S2). We tested the impact of knocking down three key LDELS pathway proteins: Atg7 (part of the LC3-conjugation machinery), neutral sphingomyelinase 2 (nSMase2/SMPD3), and factor associated with nSMase2 activity (FAN). Knockdown of Atg7, FAN (NSMAF), and nSMase2 (SMPD3) significantly reduced endogenous leptin secretion (Figure 7C), without affecting adiposity (Figure S7C). Thus, leptin secretion utilizes an LDELS secretory route requiring Atg8 proteins that promote LC3-lipidation, summarized in Figure 7D.

In conclusion, our results drawn from a range of systems - Drosophila and human cell systems, in vivo fly adipocytes, and human primary subcutaneous adipocytes- establish that Atg8/LC3 proteins, in a nutrient-sensitive manner, regulate leptin secretion in both flies and humans.

Discussion

The precise secretory route of leptin and its coupling to the adipocyte nutrient state was unclear. Studies in 3T3-L1 and rat primary adipocytes suggest leptin is enriched in the endoplasmic reticulum and traffics to the Golgi^65–67. However, they show that leptin does not traffic via known adipocyte secretory routes such as GLUT4-positive vesicles^65–67. In rat adipocytes, leptin is present in low-density small intracellular vesicular compartments⁶⁸. However, the nature of leptin secretory vesicle and nutrient-sensitive mechanisms of leptin loading into secretory vesicles has remained poorly understood. Our study addresses this knowledge gap by identifying a nutrient-sensitive mechanism mediating leptin secretion from fly and human adipocytes. We show the requirement for LC3/GABARAP family of proteins and exosome loading proteins including SMPD3 and NSMAF in leptin secretion from human adipocytes. The enrichment of components of extracellular exosomes in proteomics analyses of leptin’s interactome (Fig. 6B, C) led us to the discovery that leptin is indeed present in exosome-like vesicles (Fig. 6D).

Atg8/LC3 enable organismal adaptation to starvation because of their conserved roles in autophagy from yeast to mammals ^69–72. Prior studies established that Atg8 is enriched on autophagosomes 4–6 hours after nutrient deprivation^73,74. We observe that Upd2 and leptin secretion significantly reduce within 4 hours of nutrient deprivation in Drosophila (Figure 1B) and human (Figure 5A) cells. We reason that when Atg8/LC3 is recruited to autophagosomes¹⁷, it limits Atg8/LC3’s availability for Upd2 and leptin secretion. In support of this, the application of the autophagy activator, Torin1, significantly reduces both Upd2 and leptin secretion (Figure 1C, 5B). Overall, it appears that autophagy activation – either physiologically or pharmacologically- is antagonistic to adipokine secretion. We show that a subset of other Atg proteins – in lipidation specifically- are required for leptin secretion from human adipocytes. In contrast, autophagy initiation factors (Ulk1 & FIP200) are dispensable to leptin secretion (Schematic Figure 7D); the autophagy initiation factors were dispensable for LDELS secretion as well^75,76. To gather further evidence that autophagy initiation is antagonistic to leptin secretion, it will be important to test whether genetic activation of autophagy by expressing constitutively active AMPK or phosphomimic Ulk1 inhibits leptin secretion⁷⁷. Future work is required to tease out the precise mechanisms by which LC3 switches between autophagy and adipokine secretion.

Our lab previously reported that Golgi reassembly stacking protein (GRASP) is required for Upd2 secretion in a nutrient-sensitive manner in adult Drosophila adipocytes and S2R+ cells³². Recently, in addition to their role in unconventional secretion⁷⁸, GRASP’s role in the ER-Golgi interface is to restrict improper activation of autophagy⁷⁹. Here we find activation of autophagy inhibits Upd2/leptin secretion. It is possible that GRASP’s role in inhibiting basal autophagy⁷⁹ contributes to Upd2 and leptin secretion, and merits further exploration.

From flies to humans^1,6,7,45, it has been shown that excessive adipokine production is detrimental to organismal adaptation to starvation. Consistent with this, we uncover that Upd2 retention observed in the Upd2-AIM allele increases starvation resilience (Figure 4D, 4E). We propose many mechanisms are in play to increase the starvation resilience of Upd2-AIM flies. First, the Upd2-AIM flies have higher baseline fat stores. Prior studies have shown that excessive fat stores extend the lifespan of starvation. For instance, when the triacylglycerol lipase brummer (bmm) is mutated in flies, it results in excessive fat storage⁸⁰ and extends starvation survival. Secondly, a potential reason for increased TAG accumulation in the basal state could be that Upd2-AIM flies display hyperphagic behaviors, suggesting that adipokine retention increases hunger (Figure 4F). Future work is needed to resolve how increased upd2 retention results in hyperphagic behaviors in flies. Third, we found that even in the fed state, the bulk transcriptome of the Upd2-AIM fed flies was distinct compared to starved flies (Figure S4A). It is unclear what underlies the starkly different transcriptional profiles (Figure S4A) we observe in Upd2-AIM flies. An intriguing possibility is that intercellular retention of Upd2/Leptin could act in an ‘autocrine’ fashion to activate the STAT transcription pathway. When Atg8 is knocked down in Upd2-AIM flies, Upd2 protein is not only increased in the cytosol but also in the nucleus and nuclear lamina (Figure 3C, D). Hence, Upd2 may fine-tune gene expression by binding to other DNA and chromatin binding factors in the nucleus. In multiple independent proteomic surveys of Upd2 immunocomplexes, we identified ten bona fide DNA binding and chromatin binding factors that complexed with Upd2 (Kelly and Rajan, unpublished results); whether these interactors are functional will be tested in future work.

There are significant physiologic differences between upd2Δ and Upd2-AIM alleles. Previously, we reported that upd2Δ mutants have reduced insulin signaling, reduced lipogenesis, and reduced TAG stores⁶. Despite reduced TAG stores, upd2Δ alleles exhibit starvation resilience⁶. Whereas in the Upd2-AIM allele, where the Upd2 protein is still present but retained within adipocytes, there is a rewiring of physiology to better adapt to starvation. In sum, we propose that there are likely to be many potential mechanisms by which starvation resilience could occur during reduced adipokine signaling. Whether the increased starvation resilience of adipokine retention comes at the cost of other physiological trade-offs in immunity, reproduction, sleep, and cognition will be explored in future work.

In conclusion, Atg8/LC3 regulates nutrient surplus signals Upd2/leptin and is pivotal to the starvation response. Hence, we propose that Atg8/LC3 acts as a ‘bidirectional’ switch protein. By coupling scarcity and surplus mechanisms, Atg8/LC3 proteins enable organisms to manage nutrient flux effectively. In the future, decoding other bidirectional molecular switches and their mechanisms of action will enable us to develop new strategies to treat disorders that arise from nutrient imbalance states, such as obesity and anorexia. Our study illustrates the power of using the Drosophila model as a discovery tool to uncover fundamental, evolutionarily conserved principles in organismal cellular metabolism.

STAR Methods:

RESOURCE AVAILABILITY

Lead Contact

Requests for further information, reagents, and resources should be directed to and fulfilled by Lead Contact, Akhila Rajan (akhila@fredhutch.org).

Materials Availability

Materials and Drosophila strains generated in this study are available from the lead contact and corresponding author, Akhila Rajan (akhila@fredhutch.org).

Data and Code Availability

• All data is available in the main text or the supplementary materials.

• All data reported in this paper will be shared by the lead contact, Akhila Rajan (akhila@fredhutch.org), upon request.

• This paper does not report any original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Experimental Animals: Drosophila melanogaster

Only males were used in experiments at 5–10 days post-eclosion. Flies were cultured in a humidified incubator at 25°C with a 12h light-12h dark cycle and were fed a standard lab diet containing per liter 15 g yeast, 8.6 g soy flour, 63 g corn flour, 5g agar, 5g malt, 74 mL corn syrup. For acute starvation, adult male flies, seven days old, were subjected to 4-hour starvation on 0% sucrose in agar at 29°C. For survival curves, flies were subjected to 1% sucrose in an agar diet. For RNAi experiments, flies were raised at 18°C until seven days post-eclosion, after which they were shifted to 29°C for seven days.

The following fly strains used in this study were from our previous work ^6,32 and obtained from other previously published work and the Bloomington Drosophila stock center (BDSC): upd2Δ3–62 (upd2Δ¹⁰), LppGAL4 ⁸¹, UAS-Dicer; r4-mCherry-Atg8a, Act5C>CD2>GAL4, UAS-GFPnls/TM6B (kind gift from Helene Knævelsrud, University of Oslo) [Arsham and Neufeld, 2009⁸²]. The control UAS strain used for over-expression experiments is UAS-Luciferase ⁶. RNAi lines from the TRiP facility at Harvard Medical School (http://www.flyrnai.org/TRiP-HOME.html), including Luciferase-RNAi (JF01801), Atg8-RNAi #1 (JF02895, BDSC 28989), Atg8-RNAi #2 (HMS01328, BDSC 34340).

The following UAS lines were generated for this study: UAS-Upd2::GFP, UAS-Upd2-AIM::GFP, UAS-V_HH-Myc::Atg8-WT, UAS- V_HH-Myc. All transgenes of the same gene were inserted at the same att site to control expression levels—either attp40 or attP2. For nanobody based interaction experiments in Figure 3A, flies of genotype LppGAL4; UAS- V_HH-Myc::Atg8-WT (experiment) and LppGAL4; UAS- V_HH-Myc (Control) were crossed to UAS-Upd2-WT::GFP or UAS-Upd2-AIM::GFP, the crosses were set at 25°C and the adult flies were dissected and stained at Day 7. For the clonal analysis experiments in Figure 3D and S3, virgin females of Upd2-WT::HA; +; UAS-Luciferase-RNAi and were crossed to males of yw hsflp; UAS-Dicer; r4-mCherry-Atg8a, Act5C>CD2>GAL4, UAS-GFPnls/TM6B ⁸²). Crosses were set at 25C for 24–48 hours, parents were transferred to fresh vials, and vials with embryos were incubated at 29°C till eclosion. F1 flies were aged to 5–7 days at 29°C. Leaky expression at 29°C of the heat-shock inducible FLP leads to stochastic “flip out” activation of GAL4 expression ⁸³ in a minority of fat body cells, activating UAS-driven expression of GFP and the inserted RNAi, allowing assessment of these cells in otherwise normal animals. Fat bodies were fixed and imaged as noted. UAS RNAi experiments and temporal over-expression in Figure 3 flies were generated with the following genotype: Upd2-crGFP; Ppl-Gal4, tubGAL80ts was crossed to UAS- Atg8-RNAi or UAS-Luciferase-RNAi for 2E; after eclosion at 18°C, adult male flies were maintained at 29°C for 7–10 days to inactivate the GAL80 and allow for Gal4 expression. Then, 10–12 adult male flies were dissected and stained per genotype.

The following genomic tag-knockin lines into the endogenous Upd2 locus were generated for this study: Upd2-WT::GFP, Upd2-WT::HA, and Upd2-AIM::HA. GFP or HA-tagged Upd2 CRISPR lines were developed in collaboration with WellGenetics Inc. using modified methods of Kondo and Ueda ⁸⁴. To generate the HA-tagged Upd2 genomic knock-in used in Figure 3B, the following strategy was used. In brief, the gRNA sequence (see Key Resource Table) was cloned into U6 promoter plasmids for HA lines. Cassette 3xHA RFP containing 3xHA and 3xP3 RFP and two homology arms were cloned into pUC57 Kan as donor templates for repair. upd2/CG5988 targeting gRNAs and hs-Cas9 were supplied in DNA plasmids, together with donor plasmid for microinjection into embryos of control strain w1118. F1 flies carrying selection markers of 3xP3 RFP were further validated by genomic PCR and sequencing. CRISPR generates a break in upd2/CG5988 and is replaced by cassette 3xHA RFP. 3XP3 RFP, which facilitates genetic screening, was flipped out by Cre recombinase. 46 bp) remains after excision between the Stop codon and 3’UTR. To generate the Upd2-AIM point mutations, as seen in Figure 3C, CRISPR/Cas9 mediated genome editing by homology-dependent repair (HDR) using 1 guide RNA and a dsDNA plasmid donor was used. We introduced point mutations in the Upd2-WT::HA flies. To generate the AIM mutations, to replace W310 to A, TGG to GCG, and L313 to A, CTG to GCT of Upd2-WT::HA, using a targeting design without deletion. The gRNA sequence (see Key Resource Table) was cloned into the U6 promoter plasmid. Genomic PCR and sequencing methods to verify CRISPR alleles of upd2/CG5988 by testing if the knock-in cassette PBacDsRed is correctly used as a template for homology-directed repair and has been incorporated into guide RNA cutting site. For the GFP knock-in line of Upd2-WT::GFP used in Figure 3E, gRNA sequence (see Key Resource Table) was used.

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Coating Antibody for GFP sandwich ELISA	Allele Biotechnology	CAT#ACT-CM-GFPTRAP
Detection Antibody for GFP sandwich ELISA	Rockland	CAT#600-401-215
Anti-Rabbit secondary antibody	Abcam	CAT#ab136636
Rabbit-anti-RFP	Rockland	CAT#600-401-379
Chicken-anti-GFP	Abcam	CAT#ab13970
Mouse-anti-Lamin	DSHB	ADL67.10
Rat-anti-HA	Chromotek	#7C9 RRID: AB_2631399
Rabbit-anti-GM130	Abcam	CAT#ab52649
Mouse-anti-Myc	Abcam	CAT#ab32
GFP antibody	Proteintech	CAT#50430-2-AP
Calnexin antibody	Cell Signaling Technology	CAT#2679
TSG101 antibody	GeneTex	CAT#GTX70255
Flotillin-2 antibody	Cell Signaling Technology	CAT#3436
Bacterial and virus strains
JM109 competent E. coli	Promega	CAT#L2005
Chemicals, peptides, and recombinant proteins
Recombinant GFP protein	Vector Labs	CAT#MB-0752
Torin 1	Abcam	CAT#ab218606
3-MA	Sigma Aldrich	CAT#M9281
Critical commercial assays
Human Leptin Quantikine ELISA kit	Bio-techne	CAT#DLP00
Dharmacon Accell siRNA delivery media	Horizon Discovery	CAT#B-005000-500
DharmaFECT kB transfection reagent	Horizon Discovery	CAT#T-2006-01
Total Exosome Isolation reagent	Invitrogen	CAT#4478359
Effectene Kit	QIAGEN	CAT#301427
Direct-zol RNA miniprep kit	Zymo Research	CAT#R2071
iScript cDNA Synthesis	Bio-Rad	CAT#1708891
Deposited data
Analyzed Data	This paper	N/A
Experimental models: Cell lines
D. melanogaster S2R+	DGRC	CAT#150; RRID: CVCL_Z831
Human HEK293t	ATCC	CRL-3216
Human subcutaneous preadipocytes	Lonza	CAT# PT-5020
Experimental models: Organisms/strains
D. melanogaster upd2Δ3–62	Hombria et al.10	N/A
D. melanogaster Lpp-GAL4	Brankatschk, M. and Eaton, S. 79	N/A
D. melanogaster yw hsflp; UAS-Dicer;r4-mCherry-Atg8a, Act5C>CD2>GAL4, UAS-GFPnls/TM6B	Arsham and Neufeld. 80	Gift from Helene Knævelsrud
D. melanogaster UAS-Luciferase	Rajan, A. and Perrimon, N. 6	RRID:BDSC_35788
D. melanogaster Atg8-RNAi #1	DRSC/TRiP	RRID: BDSC_28989
D. melanogaster Atg8-RNAi #2	DRSC/TRiP	RRID: BDSC_34340
D. melanogaster UAS-Upd2-AIM::GFP	This paper	N/A
D. melanogaster UAS-VHH-Myc	This paper	N/A
D. melanogaster UAS-VHH-Myc::Atg8-WT	This paper	N/A
D. melanogaster UAS-VHH-Myc	This paper	N/A
D. melanogaster UAS-Upd2-WT::GFP	This paper	N/A
D. melanogaster UAS-Upd2-AIM::GFP	This paper	N/A
D. melanogaster UAS-Luciferase-RNAi	DRSC/TRiP	RRID: BDSC_31603
D. melanogaster Upd2-WT::GFP CRISPR knockin	This paper	N/A
D. melanogaster Upd2-WT::HA CRISPR knockin	This paper	N/A
D. melanogaster Upd2-AIM::HA CRISPR knockin	This paper	N/A
Oligonucleotides
gRNA for HA-tagged Upd2: TCCAATGAGTCTTGAGCCCT[CGG]/GCCGAGGGCTCAAGACTCAT[TGG]	This paper in collaboration with WellGenetics	N/A
gRNA for HA-tagged Upd2-AIM lines: TATTGCGTCGCGGTCGCCGC[AGG]	This paper in collaboration with WellGenetics	N/A
gRNA for GFP knock-in line of Upd2WT::GFP: TCCAATGAGGTCTTGAGCCCT[CGG]	This paper in collaboration with WellGenetics	N/A
dsRNA LacZ forward: TTTCCAGTTCCGTTTATCCG	This paper	N/A
dsRNA LacZ reverse: CAAAAATCCATTTCGCTGGT	This paper	N/A
dsRNA Atg8a #1: forward:GTACAGGGAGCCCATGGTA	This paper	N/A
dsRNA Atg8a #1 reverse: CGAGAAGGCTCCCAAGG	This paper	N/A
dsRNA Atg8a #2 forward: GCCCTGCGTATCAGATCAAT	This paper	N/A
dsRNA Atg8a #2 reverse: CGATGAGAATGTTTACGGCA	This paper	N/A
siRNAs for Leptin ELISAs, see Table S1.	Horizon Discovery	See Table S1.
Recombinant DNA
pENTR-D/TOPO	Thermo Fisher	CAT#K240020
Atg8 cDNA	DGRC-Gold Collection	CAT#LD05816
pENTR-Upd2	Rajan et al. 32	N/A
pCMV-leptin-DDK-Myc	Origene	CAT#RC209259
PDONR221	Thermo Fisher	CAT#12536017
pGEX::dt	Gift from Susan Parkhurst	N/A
pAc-Upd2::GFP	Hombria et al.10	N/A
pACRenilla::Luciferase	Hombria et al. 10	N/A
Upd2::GFP	Hombria et al. 10	N/A
Leptin::GFP	This paper	N/A
Myc::Leptin	This paper	N/A
Myc:Leptin-AIM	This paper	N/A
hLep-WT::Myc	This paper	N/A
hLep-AIM::Myc	This paper	N/A
EGFP::LC3	Addgene	#21073
Software and algorithms
Graphpad Prism 7	GraphPad Software Inc.	https://www.graphpad.com/features
R Version 4.4.0	R Core Team	https://www.r-project.org/
Image J	Schneider et al. 83	https://imagej.net/ij/
MATLAB R2020b	The MathWorks Inc.	https://www.mathworks.com/products/matlab.html

Cell lines

Drosophila S2R+ cells, HEK293T cells, and Human subcutaneous preadipocytes (Lonza Catalog #: PT-5020) were used for all cell culture-related experiments. Drosophila S2R+ cells were maintained in Schneider’s medium (GIBCO), 10% heat-inactivated FBS (SIGMA), and 5% Pen-Strep (GIBCO) at 25˚C. For starvation, cells were incubated in Schneider’s Insect Medium without Amino Acids (United States Biological, Cat# S0100–01.10) for the indicated amount of time. HEK293T cells were cultured in DMEM with 10% FBS and 1% penicillin-streptavidin supplement. For starvation experiments, HEK293T cells were cultured in EBSS (Earl’s Balanced Salt Solution) for the indicated amount of time. Human Subcutaneous Preadipocytes were utilized for human Adipocyte experiments. Preadipocyte cells were maintained in preadipocyte growth media-2 (Lonza Cat #PT-8002).

METHOD DETAILS:

Cloning and Transgenic Flies

All cloning was done using the Gateway^® Technology. For Atg8, entry cDNA was PCR amplified from Atg8 cDNA (clone LD05816 DGRC-Gold collection) and cloned into pENTR-D/TOPO using BP reaction (Gateway^® BP Clonase II enzyme mix, Cat#11789–020, Invitrogen). For Drosophila Upd2 variants, pENTR-Upd2 ³², made for previous work in our lab, was used. For site-directed mutagenesis of putative AIM sites, pENTR-Upd2 was mutagenized to convert tryptophan and leucine encoding codons to alanine. For human leptin, an ORF clone (Origene; pCMV-leptin-DDK-Myc; CAT#: RC209259) was used for mutagenizing the AIM sequences. Mutagenesis was done using the Q5^® Site-Directed Mutagenesis Kit from NEB (Cat # E0554S). The sequence of oligonucleotides used for this mutagenesis reaction is available on request. The entry vectors were then moved using LR clonase reaction (Gateway^® LR Clonase^® II Enzyme mix, Cat#11791–020, Invitrogen) into destination vectors compatible with fly transformation or cell culture with the appropriate C-terminal tags for Upd2 and N-terminal V_HH tags for Atg8. Transgenic flies were generated by using the microinjection service provided by Bestgene Inc. The Myc-tagged human leptin clones were subcloned to include a GFP tag for ELISA.

For generating the V_HH tagged constructs, the sequence for the V_HH tag and appropriate linkers, utilized in a prior study by Strutt and colleagues ⁴¹, was appended to either Myc (control: V_HH-Myc ) or Atg8-WT (V_HH-Myc::Atg8-WT). Double-stranded sequences for these constructs (available on request) were synthesized using Codex DNA Inc. Synthetic DNA platform and cloned into PDONR221 using BP clonase (Gateway^™ BP Clonase^™ II Enzyme mix; Catalog number: 11789020, Invitrogen). Then, to generate expression clones, the PDONR vectors were cloned into Expression competent S2 cell culture vectors from the Murphy Collection available from DGRC using LR clonase or into Gateway Vectors for fly transgenesis.

GST pull-down assays

Drosophila Atg8 was synthesized in vitro using the TNT quick-coupled transcription-translation kit (Cat #1170, Promega) in the presence of ³⁵S-labeled methionine. The in vitro translated lysate was diluted in Hepes-LS buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100, and 1 mM DTT) with Complete EDTA-Free protease inhibitor (Cat #11697498001, Roche) and stored at −80°C.

GST-tagged Upd2-WT and Upd2-AIM were cloned into a pGEX vector backbone and transformed into JM109 competent E.coli (Cat #L2005, Promega). Protein expression was induced using 0.3 mM IPTG (Cat # I2481C, GoldBio). After 3 hours, pelleted cells were lysed using sonication. Lysates were incubated with glutathione-sepharose beads and nutated for 3 hours at 4°C. Beads were washed and stored at 4°C.

For the GST pull-down assay, IVT-Atg8 was incubated in GST-fused glutathione sepharose for 1 h at 4°C to remove nonspecific binding proteins. The precleared lysate was individually incubated with GST, Upd2-WT, or Upd2-AIM in Hepes-LS for 1 h at 4°C. The beads were then washed three times in Hepes-LS buffer, and the bound fractions were analyzed by SDS-PAGE followed by autoradiography (Magie and Parkhurst, 2005; Magie et al., 2002; Rosales-Nieves et al., 2006).

Survival Assays

Survival curves were done using flies harvested in a 24-hour time frame and aged for 7 days. Ten males per vial were flipped onto 1% sucrose agar starvation food. The number of dead flies was recorded each day until every fly had died. Flies were kept in a 25°C incubator with a 12-hour light-dark cycle for the entirety of the experiment.

Drosophila lipid droplets

Adult fat bodies were dissected in ice-cold PBS and then fixed for 20 minutes in 4% formaldehyde in PBS at room temperature. Following fixation, tissues were washed five times in 0.5% BSA and 0.5% Triton X-100 in PBS (PAT). Tissues were blocked in PAT + 5% NDS for 1 hr at room temperature, then incubated overnight with LipidTOX (1:500, Thermo Fisher Cat#H34477). Following incubation, samples were washed five times in PAT, then mounted on slides in Slow fade gold antifade. Images were acquired with the Leica Stellaris 5 confocal system.

Triglyceride Measurements

TAG assays were carried out as previously described in (Rajan et al., 2017). In brief: Flies were homogenized in PBST (PBS + 0.1% Triton X-100) using 1mm zirconium beads (Cat#ZROB10, Next Advance) in a Bullet Blender^® Tissue homogenizer (Model BBX24, Next Advance). Samples were heated to 70°C for 10 minutes, then centrifuged at 14,000 rpm (in a refrigerated tabletop centrifuge). 10.0 μl of the supernatant was applied to determine the level of TAG in the sample, using the following reagents obtained from Sigma: Free glycerol (cat # F6428–40ML), Triglyceride reagent (cat# T2449–10ML), and Glycerol standard (cat# G7793–5ML).

Feeding behavior

Age-matched flies were given a normal diet or HSD for 7 days after an initial 7 days of development on a normal diet. Then, 16 hr prior to feeding behavior assessment, half of the flies from each treatment were moved to starvation media 0%. During the 3-hour assessment window of feeding behavior, individual flies were placed in a single well of FLIC and supplied with a 5% sucrose liquid diet for all FLIC experiments. Detailed methods for how FLIC operates can be found in Ro et al., 2014. Fly feeding was measured for the first 3 hr in the FLIC, and all FLICs were performed at 10 am local time. For each FLIC, half of the wells (n = 6/FLIC) contained the fed group, and the other half contained the starved group of flies for direct comparison. A total of 12–30 flies were measured for analysis. Any signal above 40 (a.u.) was considered a feeding event. Analysis of feeding events was performed using R.

Tissue culture:

Drosophila S2R+ cells were passaged to 60–80% confluency. Cells were cultured in 96 well plates for transfections related to ELISA experiments and coimmunoprecipitations in 6-well plates. For ELISAs, cells were transfected with 20ng/well of pAc-upd2::GFP 3 or indicated Upd2 variants, 10ng/well pACRenilla::Luciferase, and 150ng of dsRNA/well for knockdown experiments.

For imaging experiments, cells were seeded on poly-D-lysine coated 8-well chambered culture slides for fixed imaging (MatTek CCS-8). The T7 flasks, which were 100% confluent, were used for seeing cells at 1:10 dilution at 400μl per well of the 8-well chamber dish. Transfections were done with plasmids indicated. For 8-chamber slides, 100ng/well of each plasmid was transfected. Transfections were done using the Effectene kit (Cat# 301427, QIAGEN) per the manufacturer’s instructions.

Drosophila S2R+ dsRNA production and cell treatments

Amplicons for dsRNAs were designed using the SnapDragon dsRNA design tool (https://www.flyrnai.org/snapdragon) and in vitro transcribed (IVT) using MEGAscript^® T7 Transcription Kit (Cat# AMB1334–5, ThermoFisher). IVT reactions were carried out as per the protocol provided by the DRSC (available at: http://www.flyrnai.org/DRSC-PRS.html). The sequence of amplicons used in this study can be found in the table below. LacZ dsRNA was used as a control. All dsRNA knockdown experiments were carried out using two independent dsRNAs per gene. We found that this produced a knockdown efficiency of >85% (based on qPCR analysis) in S2R+ cells. S2R+ transfected with dsRNAs were incubated for four days to allow for gene knockdown. On the 4th day, the media was changed, and the ELISA assay was carried out on the 5th day. Note that the data is represented as percent change in Upd2/Leptin secretion normalized to transfection efficiency with 0% change indicating baseline secretion level. See the ELISA assay procedure below.

Treatment of cells with drugs

For drug treatment experiments, the media was replaced with media containing the drug on day 3 after transfection with Upd2::GFP or Leptin::GFP. Four hours later, the conditioned media was used for ELISA or imaging. Drugs used in this study include Torin1 (Cat# ab218606, Abcam) and 3-MA (Cat#M9281, Sigma Aldrich). Stock solutions of the drugs were made in DMSO as per the manufacturer’s instructions and used at a working concentration indicated in Figure 1C and 5B legends. DMSO-treated replicates were used as controls. Note that for ELISAs, the data is represented as percent change in Upd2 or Leptin secretion normalized to transfection efficiency, with 0% change indicating baseline secretion level.

ELISA assays

GFP sandwich ELISA assay was used for detecting Upd2::GFP. On day 1, 96 well medium bind polystyrene plates (Cat#CLS3368–100EA, Sigma) were incubated overnight at 4°C with coating antibody (Cat# ACT-CM-GFPTRAP, Allele Biotechnology) diluted in 0.01M pH8.0 bicarbonate buffer at a concentration of 1μg/ml. On day 2, plates were washed briefly with PBS, blocked for 30 minutes with 1% BSA block in PBS, coated with conditioned media, and incubated overnight at 4°C. Recombinant GFP protein (Cat# MB-0752, Vector labs), diluted in S2R+ cell growth media (64ng/μl to 4 ng/μl), was used in every ELISA plate as a positive control to ensure linearity of GFP readings. On day 3, the plates were washed with PBS+0.05% Tween-20 (PBS-T) blocked with 1% BSA in PBS for 30 minutes at RT. GFP detection antibody (Cat# 600–401-215, Rockland) was added to the diluted 1:1000 in 0.1% BSA in PBS-T. Plates were washed with PBS-T and incubated with secondary HRP conjugated anti-Rabbit secondary antibodies (Cat# ab136636, Abcam) diluted at 1:5000 in 1% BSA block. Plates were washed in PBS-T with a final wash in PBS. For detection, each well was incubated with 100μl 1-step Ultra-TMB ELISA substrate (Cat# 34028, Pierce), which was previously equilibrated to RT, for approximately 5–15 minutes until detectable blue colorimetric reaction occurred. The reaction was stopped with 2N sulphuric acid, and absorbance was measured at 450nm. The TMB readings were normalized to transfection efficiency as measured from Renilla Luciferase assays (see below). Renilla Luciferase assay: On day 2 of the ELISA assay, after the conditioned medium was transferred for use in ELISA assays, cells were re-suspended in 50μl of PBS and incubated with 50μl/well Renilla-Glo^® Luciferase reagent (Cat# E2710, Promega) for 10 minutes and read using a multiwell luminometer.

Human HEK293T cells were cultured in DMEM with 10%FBS and 1% penicillin-streptavidin supplement. At the start of the experiment, cells were trypsinized and seeded onto a 24-well plate or 8-well MatTek tissue culture slides for imaging (MatTek CCS-8) and grown to confluence.

Human Subcutaneous Preadipocytes (Lonza Catalog #: PT-5020) were utilized for human Adipocyte experiments. Preadipocyte cells were maintained in preadipocyte growth media-2 (Lonza Cat #PT-8002). For experiments, 15,000 preadipocyte cells per well were seeded into a 24-well plate and grown to 90% confluency. Once confluent, cells were then switched to differentiation media (Lonza Cat #PT-8002) for seven days to allow for differentiation into mature primary adipocytes and the development of lipid droplets. All cells were housed in a 37°C Incubator at 5% CO2 and 90% humidity.

Leptin ELISA

Cell culture media of mature adipocytes was replaced 5 days prior to ELISA. For siRNA knockdown experiments, media was replaced with 500uL of Accell siRNA delivery media (Horizon Discovery, Dharmacon Accell siRNA delivery media Cat B-005000–500) with 0.5uM of each siRNA following manufacturer’s instructions. See Table S1 for siRNAs used in this study. The supernatant was then collected and used to measure leptin concentration after 5 days of treatment with siRNA. ELISA’s for leptin were performed using the bio-techne Human Leptin Quantikine ELISA kit (DLP00). On the same day as the supernatant was collected, an adipored assay was performed with the remaining cells (Lonza Adipored Assay Reagent #PT-7009) for each sample used in the ELISA to confirm similar levels of adipocyte development and intracellular triglyceride was present amongst samples being compared. Adipocyte differentiation was quantified via intracellular triglyceride levels determined through the use of Lonza AdipoRed Assay Reagent. In brief, after the supernatant for the Leptin ELISA assay was collected, the remaining cells were then washed in PBS, followed by adding 1mL PBS mixed with 30μL AdipoRed. Cells were incubated for 15 minutes, followed by immediate fluorescent measurement using the SpectraMax i3 plate reader with an excitation of 485nm and an emission of 572nm. For each ELISA assay, samples had two technical replicates at three different dilutions to confirm accurate concentration based on a standard curve of known Leptin concentrations.

Cell immunofluorescent staining:

Drosophila S2R+ cells:

For immunohistochemistry, cells were fixed for 20 minutes in 4% formaldehyde, washed in PBS for 5 quick changes, permeabilized in PBS+ 0.1% Triton-X-100 for 3X washes 5 minutes, subsequently washed with PBS+0.1% Triton-X-100 (Cell wash). Blocked for 30 minutes at room temperature (RT) with gentle agitation in Cell wash+ 5% Normal donkey serum (Block). Primary antibodies – Rabbit-anti-RFP (1:500, Rockland, #600–401-379); Chicken-anti-GFP (1:2000, Abcam, #ab13970); mouse-anti-Lamin (1:100; ADL67.10 DSHB) – were diluted in block incubated overnight at 4oC. Washed 3X-5X for 15 mins each the following day in the fat wash at (RT) incubated with appropriate secondary antibody at a concentration of 1:500 (from Jackson Immunoresearch) in the block for 2–4 hours at RT. Washed 3X-5X for 5–15 minutes in cell wash, mounted with SlowFade Gold antifade reagent with DAPI (Cat# S36938, Invitrogen). Images were captured with Zeiss LSM 800 or Zeiss Elyra 7 with Lattice SIM confocal systems and analyzed with Zeiss ZenLite or ImageJ⁸⁵.

For live imaging, cells were incubated with ProLong Live antifade reagent (Cat# P36974, ThermoFisher) and NucBlue (Cat# R37605, Thermofisher) as per manufacturer’s instructions prior to imaging in appropriate media condition. Images were captured in time-lapse mode on Zeiss LSM800 in the sequential scan setting.

Drosophila adult adipose tissue:

For fat body preparation, incisions using dissection scissors (Cat# 500086, World Precision Instruments Inc) were made to release the ventral abdomen from the rest of the fly body. Flies used for dissection were adult males, 7–10 days old. Dissections were done in Ringer’s medium (1.8 mM CaCl2, 2 mM Kcl, 128 mM NaCl, 4 mM MgCl2, 35 mM sucrose, 5 mM HEPES). The tissue was fixed in 4% formaldehyde for 20 minutes and rinsed with PBS. For immunohistochemistry, the fixed fat tissue was permeabilized in PBS+ 1.0% Triton-X-100 for 3X washes for 5 minutes and subsequently washed with PBS+0.3% Triton-X-100 (Fat wash). Blocked for 30 minutes at room temperature (RT) with gentle agitation in Fat wash+ 5% Normal donkey serum (Block). Primary antibodies, diluted in block, included Rabbit-anti-RFP (1:500, Rockland, #600–401-379); Chicken-anti-GFP (1:2000, Abcam, #ab13970); rat-anti-HA (1:500; Chromotek, #7C9); mouse-anti-Lamin (1:100; ADL67.10 DSHB) and incubated overnight at 4°C. Washed 3X-5X for 15 mins each the following day in the fat wash at (RT) incubated with appropriate secondary antibody (from Jackson Immunoresearch) at a concentration of 1:500 in the block for 2–4 hours at RT. Washed 3X-5X for 5–15 minutes in the fat wash, mounted in SlowFade Gold antifade reagent with DAPI (Cat# S36938, Invitrogen). Images were captured with the Zeiss LSM 800 confocal system and analyzed with Zeiss ZenLite or ImageJ.

Human HEK293T cells:

Hek293T cells were trypsinized and seeded onto 8-well MatTek tissue culture slides for imaging (MatTek CCS-8) with rat tail collagen I (ThermoFisher Cat# A1048301). Rat collagen was used to allow the cells to adhere and spread on the glass slides. 1:80 dilution collagen: PBS. 200uL per MatTek well and incubated at 37C for at least two hours. MatTek was then either placed in the fridge for future use or liquid was removed, and MatTek was used immediately, after a quick wash, and seeded with HEK293T cells. Upon reaching confluency, cells were transfected with either a Myc-tagged wildtype leptin or a Myc-tagged leptin with the AIM regions point mutated. Transfection was performed using the Effectene transfection reagent (Qiagen). Cells were incubated in the transfection media for 24 hours and then replaced with standard-supplemented DMEM media for another 24 hours before the tissue was fixed. The tissue was fixed in 4% formaldehyde for 30 minutes and rinsed with PBS. HEK293T cells were then washed with PBS+0.3% Triton-X-100 (Cell wash) three times for 20 minutes each wash. Cells were blocked for 1 hour in Cell wash + 5% NDS (Block). Primary antibodies, diluted in block, included Rabbit-anti-GM130 (1:200, #ab52649); mouse-anti-myc (1:500, #ab32) and incubated overnight at 4°C. Cells were then washed five times for 10 minutes each the following day in Cell wash at room temperature. Cells were again subject to block for 30 minutes. The appropriate secondary antibodies (Jackson Immunoresearch) and DAPI were diluted in a block at a concentration of 1:500 for 2 hours. Afterward, cells were washed five times for 5–15 minutes in the Cell wash, mounted with Prolong Diamond Antifade (Thermo Fischer #P36961). The HEK293T cell imaging was performed with Leica Stellaris 5 with deconvolution Lightning software.

Immunoprecipitation, Mass Spectrometry, and Proteomics:

For Immunoprecipitation (IP) of human leptin from HEK293T cells, cells were transfected with GFP-tagged Leptin-WT and Leptin-Myc. Protein for each condition was prepared by lysing 1 well of a 6-well dish 4 days after transfection. 2mg/ml was used per IP experiments performed with camelid antibodies GFP-Trap Magnetic Agarose (Cat# gtma-20, Chromotek) as per the manufacturer’s protocol. Immunoprecipitates were electrophoresed approximately 1 to 2 centimeters into an SDS-PAGE gel. Coomassie-stained gel bands were proteolytically digested with trypsin as described (in Cheung et al.2017; ⁸⁶). De-salted peptides were analyzed by LC-MS/MS using an Orbitrap Eclipse with FAIMS (field asymmetric ion mobility spectrometry) mass spectrometer.

RNAseq:

RNA prep:

Total RNA was prepared from 30 flies per genotype in triplicates using the Direct-zol RNA miniprep kit (Zymo Research, cat#R2071). cDNA prepared with iScript cDNA Synthesis (Bio-Rad, cat#1708891), and 1mg RNA applied per reaction.

RNA quality control:

Total RNA integrity was checked using an Agilent 4200 TapeStation (Agilent Technologies, Inc., Santa Clara, CA) and quantified using a Trinean DropSense96 spectrophotometer (Caliper Life Sciences, Hopkinton, MA).

Co-Immunoprecipitation in HEK293T cells

HEK293T cells were seeded at a dilution of 1:5 in 2 medium bind polystyrene 6-well plates and incubated overnight at 37°C. The next day, one of the 6-well plates was transfected with hLEP-WT::Myc and EGFP::LC3 (Addgene #21073), while the other plate was transfected with hLEP-AIM::Myc and EGFP::LC3, using Dharmacon DharmaFECT transfection reagent. 24–48 hours post-transfection, cells were harvested using cell scrapers and washed with ice-cold 1xPBS to remove residual media and floating cells. Cells were lysed in 200 μL of ice-cold lysis buffer (10mM Tris/Cl pH 7.5, 150mM NaCl, 0.5mM EDTA, 0.5%Nonidet P40 Substitute, 0.09% sodium azide) (provided in the ytmak-20 ChromoTek Myc-Trap Magnetic Agarose Kit) supplemented with 100X HALT protease inhibitor (Cat #78430, Thermo Scientific). The lysates were transferred into pre-cooled microcentrifuge tubes and placed on ice at 4°C for 30 min. The cell suspension was extensively pipetted every 10 minutes. Lysates were then centrifuged at 17,000 × g for 10 minutes at 4°C to pellet cell debris, and the supernatants were collected and transferred to pre-cooled tubes. ~200 μL of cell lysates were then incubated with 50 μL of ChromoTek Myc-Trap Magnetic Agarose beads overnight at 4°C. Bound proteins were eluted from the beads by adding 25 μL of 4x Laemmli Sample Buffer and TCEP Bond Breaker (Cat #77720, Thermo Scientific). The elution was achieved by heating the samples at 95°C for 5 minutes. Immunoprecipitated proteins were resolved by SDS-PAGE and transferred onto a nitrocellulose membrane for immunoblotting. Membranes were probed with appropriate primary antibodies against Myc and GFP, followed by HRP-conjugated secondary antibodies. Protein bands were visualized using chemiluminescence detection systems.

HEK293T Western Blots

HEK293T cells were seeded at a 1:5 dilution onto three 6-well medium bind polystyrene plates and incubated overnight at 37°C in DMEM complete media to allow for attachment and growth. On the following day (day 2), one plate was transfected with leptin-WT::Myc, while the second plate was transfected with leptin-AIM::Myc, using Dharmacon DharmaFECT kB transfection reagent (Cat #T-2006–01, Horizon Discovery). A third plate was left untransfected to serve as a negative control. After 24 hours, 500 μL of 0.25% Trypsin (GIBCO) was added to each well and incubated at 37°C for 1 minute. The trypsin reaction was stopped by adding 1000 μL of DMEM complete media to each well. Cells were then collected from each plate into labeled 15 mL conical tubes and centrifuged at 500 × g for 5 minutes. The supernatant was aspirated and discarded. For cell lysis, 150 μL of a mixture containing 200 μL of diH2O and 2 μL of HALT Protease Inhibitor Cocktail, along with 50 μL of 5x RIPA Buffer, was added to each tube containing the cell pellet. The cells were transferred to 1.5 mL Lobind Eppendorf Tubes, labeled accordingly, and incubated on a rocking platform for 30–60 minutes at 4°C. Subsequently, the tubes were centrifuged at 10,000 × g at 4°C for 10 minutes, and the supernatant was carefully collected into separate 1.5 mL Eppendorf Tubes. Protein quantification was performed on all three lysate samples using the Pierce^™ BCA Protein Assay Kit. Samples were then prepared for western blotting by adjusting the concentration to 2 μg/μL using 4x Laemmli Sample Buffer and TCEP Bond Breaker, followed by heating at 95°C for 5–10 minutes.

Equal amounts of protein (10 μg per lane) were loaded onto SDS-PAGE gels. Precision Plus Protein^™ Dual Color Standards (Cat #1610374, Bio-Rad) were included as molecular weight markers. Following electrophoresis, proteins were transferred from the gel to a nitrocellulose membrane using a wet transfer system. The membrane was blocked in Starting Block^™ Blocking Buffer (Cat #37538, Thermo Scientific) in TBS-T (Tris-buffered saline with 0.1% Tween-20) for 1 hour at room temperature with gentle agitation to prevent non-specific binding. Primary antibodies against the target protein and loading control were diluted in blocking buffer and TBS-T; GFP (Cat #50430–2-AP, Proteintech) 1:5000, Calnexin (Cat #2679, Cell Signaling Technology) 1:1000, TSG101 (Cat #GTX70255, GeneTex) 1:500, Flotillin-2 (Cat #3436, Cell Signaling Technology) 1:1000, and incubated with the membrane overnight at 4°C. Membranes were washed and incubated with HRP (horseradish peroxidase)-conjugated secondary antibodies diluted in blocking buffer and TBS-T for 1 hour at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection system (Cat #AB65623, Abcam). Images were captured using a chemiluminescence imaging system.

Small Extracellular Vesicle (sEV) Collection and Preparation

Post transfection of HEK293T cells with hLEP-WT::Myc or hLEP-AIM::Myc, conditioned media from three 6-well plates was centrifuged at 2000 × g at 4°C for 30 minutes to remove cells and debris. The resulting supernatant, containing cell-free culture media, was carefully transferred to new 15 mL conical tubes without disturbing the pellet. To precipitate exosomes, 0.5 volumes of Invitrogen^™ Total Exosome Isolation reagent (Cat# 4478359, Invitrogen) was added to each tube. The mixture was homogenized by vortexing and then incubated at 4°C for 3 days to allow exosome precipitation. Following the incubation period, the culture media/reagent mixtures were transferred to Polypropylene Centrifuge Tubes (Beckman Coulter) and subjected to ultracentrifugation at 10,000 × g for 1 hour at 4°C. The resulting pellet containing exosomes was located at the bottom of the tube, while the supernatant was aspirated and discarded. The exosome pellet was then resuspended in 100 μL of ice-cold 1x PBS. Protein quantification was immediately performed using the Pierce^™ BCA Protein Assay Kit. For downstream analysis, samples were prepared for western blotting by mixing with 4x Laemmli Sample Buffer and TCEP Bond Breaker and subsequently heated to 95°C for 5–10 minutes to denature proteins and break disulfide bonds. For TEM, the resulting exosome pellet post ultracentrifugation was resuspended and fixed in 100 μL 2.5%glutaraldehyde/2%PAF/0.1M cacodylate.

Negative Staining for TEM

200 mesh Formvar/Carbon coated glow discharged grids were floated over 50ul sample suspensions for 30 min. Grids were briefly rinsed with ½ strength Karnovsky’s EM fixative, 0.1M Cacodylate buffer, and distilled water. Grids were negatively stained with 1% uranyl acetate, wicked with damp filter paper, and allowed to air dry overnight. Samples were examined on a ThermoFisher Talos L120c transmission microscope (Thermo Fisher Scientific, Waltham, MA) at 120kV. Digital images were acquired with a Ceta 16M CMOS 4kx4k digital camera system.

QUANTIFICATION AND STATISTICAL ANALYSIS:

Survival Analysis:

Survival analysis was performed using the Survival Curve module of GraphPad Prism. A Mantel-cox test was the test used to determine statistical significance. 100–150 flies were used per genotype per curve. Data consolidated from 3 independent experiments is shown.

Triglyceride Analysis:

Three adult males were employed per biological replicate. Note: For adult TAG assays, the most consistent results, with the lowest standard deviations, were obtained with 10-day-old males. TAG readings from whole fly lysate (n=4 replicates of 3 flies each) were normalized to the number of flies per experiment. The normalized ratio from the control served as a baseline, and data are represented as a fold change of experimental genotypes with respect to the control.

Quantification for ELISA assays:

For ELISA assays, the ELISA signal readings are normalized to transfection efficiency; the data is represented as percent fold change from the control used as the baseline. Specifically, the ratio of TMB readings to Renilla Luciferase readings is calculated. This ratio from the control is used as a baseline, and the data is represented as the percent fold change of experimental conditions with respect to the control. A 2-tailed t-test quantified statistical significance on 6 biological replicates per condition. Error bars represent %SD (Standard Deviation). For TAG analysis, statistical significance was quantified by a 2-tailed t-test on 3–6 biological replicates per condition. Error bars indicate %SD (Standard Deviation). For qPCR analysis, normalized gene expression levels were analyzed using one-way ANOVA. Error bars represent the standard error of the mean (SEM).

Image acquisition, Morphometrics, and image analysis:

All cell biological experiments on adult Drosophila adipose tissue, HEK293Tand S2R+ cells, were performed 3 independent times at least. For fat imaging, 6–8 fat explants per experimental condition were used. For cells, at least 10 Z-stacks, with 5–10 cells per field of view, were captured per experimental condition. The confocal settings for laser power and gain were maintained consistently per experiment across different genotypes or dietary conditions. To prevent bias, the person performing the starved versus fed imaging was blinded to the sample’s status. Excel or GraphPad Prism 7 software was used for data quantification and generation of graphs. Image analysis pipelines were built in MATLAB R2020b, and the associated scripts are available on request to the lead author. The following four types of analysis were performed in 3D to quantify:

• The nucleus-to-whole cell intensity ratio of Upd2 in S2R+ cells: Since the S2R+ cells were transiently transfected, we used a ratiometric quantification (Nuclear vs whole cell) to assess Upd2 intensity. To generate the 3D nuclear masks, Lamin or DAPI stacks were first maximally projected along the z-axis. After global thresholding, basic morphological operations, and watershed transform, the locations of the nuclear centroids in the x-y plane were used to scan the z-stacks and reconstruct the nuclear volume. The whole-cell volume was approximated from the nuclear volume by dilation.

• Drosophila adipocytes Upd2 levels: After global thresholding, basic morphological operations, and watershed transform, the locations of the nuclear centroids in the x-y plane were used to scan the z-stacks and reconstruct each nuclear volume. The whole-cell volume was approximated from the nuclear volume by dilation. Before measuring the whole cell mean intensity of Upd2, intensities were corrected for signal attenuation along the depth of the tissue. In brief, the mean lamin pixel intensity was recorded at each z plane (when present), and the intensity decay was fitted with a single exponential after intensity normalization. The Upd2 intensity was then adjusted at each z-plane before calculating its mean intensity.

• Drosophila adipocytes clonal analysis: The clonal volume was defined after thresholding and applying morphological filters to the GFP signal. Upd2 signal was adjusted with respect to tissue depth, and the mean Upd2 pixel intensity was then measured inside and outside of the clone volume.

• Leptin analysis in HEK293T cells for mean intensity, polarization, and co-localization: Nuclei were segmented in 3D from the DAPI stack channel using our own trained Stardist3D model, a neural network-based approach to detect star convex polyhedral-shaped objects, such as nuclei (https://doi.org/10.48550/arXiv.1806.03535). To quantify the polarization of the leptin signal around the nucleus, its distribution in the equatorial region of the cell was considered. The equatorial region of each cell was defined by selecting the z-plane where the nucleus surface is the largest. The cell area was extrapolated from the nucleus area by tessellation, and the resulting cell was divided into 9 sectors with equal central angles. The integral intensity of the leptin signal was then computed for each sector. To compare the leptin distribution between cells, a circular shift was applied, such that the sector with the greatest leptin intensity faces the 0° location. To compare distributions between conditions, a polarization factor was computed after aggregating all aligned sectors and taking the ratio between the resulting highest leptin sector intensity and the total leptin intensity. Using the cell and nuclear mask, the whole cell intensity, as well as the nuclear to cell intensity ratio were also extracted. Once these cellular compartments had been defined, they were used as masks to extract intensity values of the signal of interest and to compute either the mean pixel intensity in the nucleus, the ratio of the integral intensities between the nucleus and the whole cell, or between the combined vesicles and the whole cell. Finally, colocalization between leptin and the secretory marker GM130 was quantified using the Manders overlap coefficient of leptin with respect to GM130.

Statistics:

Two-sided Wilcoxon rank-sum tests were used to assess the statistical significance of pairwise comparisons between experimental conditions. We determined that our data points were normally distributed based on two measures: i) A GraphPad outlier test did not identify any outliers in our data, and ii) the majority of our data points for a particular condition were relatively similar to one other, with only a small standard error of mean or standard deviation.

Immunoprecipitation, Mass Spectrometry, and Proteomic analysis:

Collected data were analyzed with Proteome Discoverer v2.5 to generate label-free quantification (LFQ) values, and identified peptides were filtered to a 1% false discovery rate. LFQ data were normalized to the median value across all samples. Missing data were imputed with half of the global minimum value. P-values for pairwise comparisons were calculated by t-test. All plots were constructed by ggplot2 in R. Proteins that were predicted to interact with Leptin-WT exclusively as per the IP-MS experiment and were input into the STRING database ⁵⁸. K-means clustering of the genes was conducted to categorize the genes into 3 clusters based on both known and predicted protein-protein interactions on the basis of numerous sources, including experimental repositories, computational prediction methods, and public text collections.

RNA-seq Expression Analysis:

RNA-seq libraries were prepared from total RNA using the TruSeq Stranded mRNA kit (Illumina, Inc., San Diego, CA, USA). Library size distribution was validated using an Agilent 4200 TapeStation (Agilent Technologies, Santa Clara, CA, USA). Additional library QC, blending of pooled indexed libraries, and cluster optimization was performed using Life Technologies’ Invitrogen Qubit^® 2.0 Fluorometer (Life Technologies-Invitrogen, Carlsbad, CA, USA). RNA-seq libraries were pooled (35-plex) and clustered onto an SP flow cell. Sequencing was performed using an Illumina NovaSeq 6000 employing a paired-end, 50 base read length (PE50) sequencing strategy.

RNA-seq analysis:

STAR v2.7.1⁸⁷ with 2-pass mapping was used to align 50bp paired-end reads to Drosophila melanogaster genome build BDGP6.32 and Ensembl gene annotation 104 (http://uswest.ensembl.org/Drosophila_melanogaster/Info/Index). FastQC 0.11.9 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and RSeQC 4.0.0 ⁸⁸ were used for QC, including insert fragment size, read quality, read duplication rates, gene body coverage, and read distribution in different genomic regions. FeatureCounts ⁸⁹ in Subread 1.6.5 was used to quantify gene-level expression by strand-specific paired-end read counting. Bioconductor package edgeR 3.26.8 (https://academic.oup.com/bioinformatics/article/26/1/139/182458) was used to detect differential gene expression between conditions. Genes with low expression were excluded by requiring at least one count per million in at least N samples (N is equal to the number of samples in the smallest group). The filtered expression matrix was normalized by the TMM method ⁹⁰ and subject to significance testing using a quasi-likelihood pipeline implemented in edgeR. Genes were deemed differentially expressed if absolute fold changes were above 2 and Benjamini-Hochberg adjusted p-values were less than 0.01.

Supplementary Material

1

Data S1. RNAseq data. Related to Figure 4.

Two-way comparisons between the 4 experimental conditions (Upd2-WT Fed, Upd2-WT Stv, Upd2-AIM Fed, Upd2-AIM Stv) and 6 different pair-wise comparisons have been tabulated (A-F) with fold changes and FDR.

2

Data S2. IP LC-MS data. Related to Figure 6.

Comparative proteomics data from immunoprecipitated (IP) leptin-WT::GFP and leptin-AIM::GFP from HEK293T cells. Sheets A, B) proteomics Raw data; C) Differential protein expression; Sheet C) Gene Ontology and D-F) STRING analyses of proteins that pull-down with Leptin-WT.

Highlights.

• Atg8/LC3, in fed cells, is required for Upd2/Leptin secretion from adipocytes.

• Upd2/Leptin interact with Atg8/LC3 via Atg8-interaction motif (AIM/LIR).

• LC3-mediated extracellular vesicle loading is crucial for leptin secretion.

• Atg8-mediated adipokine secretion promotes organismal starvation resilience.