A Drosophila toolkit for HA-tagged proteins unveils a block in autophagy flux in the last instar larval fat body

Protein functional analysis is fundamental in cell and developmental biology. Multiple transgenic strains with a protein of interest (POI) that harbors different tags are needed for in vivo functional analysis (Kanca et al., 2017), e.g. a transgenic line with an epitope tag-fused construct for immunostaining or immunoprecipitation (Vandemoortele et al., 2019), a GFP- or RFP-fused construct for live imaging (Dunst and Tomancak, 2019), a biotin ligase-fused construct for proximity labeling (Bosch et al., 2021), or a RFP-GFP tandem fluorescent-fused construct to monitor autophagic degradation in lysosomes (Kimura et al., 2007). Transgenesis methods introducing a tagged construct at a secondary site are well established in Drosophila (Venken and Bellen, 2014). Nevertheless, it is still labor-intensive work to generate multiple transgenic strains for each gene of interest.

Autophagy, an intracellular degradation pathway in eukaryotes, is induced by environmental and developmental stimuli (Klionsky et al., 2021). Developmentally programmed autophagy is seen in a variety of Drosophila tissues, including the salivary gland, muscle and fat body (Fujita et al., 2017; Berry and Baehrecke, 2007; Rusten et al., 2004). It is thought that developmental autophagy allows the degradation and turnover of the cytosolic materials required for tissue remodeling during metamorphosis (Fujita et al., 2017). In the fat body, autolysosomes, a hybrid organelle of autophagosome and lysosome, are enlarged and accumulate massively at the last larval and white prepupal stage (Rusten et al., 2004; Butterworth et al., 1988; Butterworth and Forrest, 1984). The steroid hormone ecdysone, the master regulator of insect development, drives developmental autophagy in the fat body through downregulation of the Akt/PI3K axis (Rusten et al., 2004). However, it is unclear whether the robust induction of autophagy by ecdysone is the most important cause of the accumulation of enlarged autolysosomes. Autophagic flux has not yet been explored; therefore, it is still possible that a delay in autolysosome turnover results in the accumulation of enlarged autolysosomes in the last instar larval and prepupal fat body.

Here, we have developed a Drosophila toolkit of HA Frankenbody, a genetically encoded single-chain variable fragment for the HA epitope tag (Zhao et al., 2019). As the FlyORFeome project generates and distributes a comprehensive in vivo genome-wide UAS-ORF-3xHA library of Drosophila (Bischof et al., 2012), the HA Frankenbody toolkit is powerful for functional analysis of a POI. This tool has allowed us to carry out live-imaging of HA-tagged constructs in Drosophila tissues. By exploiting the pH sensitivity of GFP, GFP-mCherry tandem fluorescently tagged (TF) HA Frankenbody visualizes the autophagic degradation of HA-tagged proteins. Additionally, TF-tagged HA Frankenbody reveals a block in autophagic flux in the last instar larval and prepupal fat bodies, probably due to insufficient acidification of lysosomes. Altogether, a HA Frankenbody toolkit enables various protein localization and dynamics studies, including the autophagic degradation assay, by simply crossing HA Frankenbody transgenic flies, HA Frankenflies, with HA-tagged POI flies.

We generated a series of transgenic lines harboring fluorescent protein-fused HA Frankenbody under UAS or QUAS enhancer sequences (Table 1) to express the HA Frankenbody in Drosophila tissues. Binary expression systems allow the expression of genes in a tissue of interest (Brand and Perrimon, 1993; Riabinina et al., 2015). To express both the HA-tagged protein and the HA Frankenbody construct under the control of the UAS sequences, the progenies need three transgenes: (1) a tissue-specific GAL4 driver; (2) a UAS gene of interest (GOI)-HA; and (3) a UAS-HA Frankenbody (Fig. 1A). For this purpose, the UAS-HA Frankenbody was combined with several GAL4 lines, such as Cg-GAL4, DMef2-GAL4, pnr-GAL4 or da-Gal4 (Table 1 and Fig. 1A). These are drivers for fat bodies, muscle cells, larval epidermal cells and broad tissues. We used larval epidermal cells (LECs) to assess the system because LECs are mononuclear and form a monolayer suitable for observing intracellular organelles. First, we expressed HA Frankenbody-GFP in the absence of HA-tagged protein. HA Frankenbody-GFP diffusely localized in both cytosol and nucleus, in the same way as GFP, in the absence of an HA-tagged construct (Fig. 1B; LacZ, control). Next, we co-expressed HA Frankenbody-GFP and HA-tagged organelle markers. γCOP, TOM20 and βTub97EF are markers for Golgi, mitochondria and microtubules, respectively. The localization of HA Frankenbody-GFP changed upon expression of γCOP-3xHA, TOM20-3xHA or βTub97EF-3xHA (Fig. 1B). HA Frankenbody-GFP signal colocalized entirely with the anti-HA immunostaining signal (Fig. 1B, magnified and line-plot). We also confirmed their colocalization in larval body wall muscles (LBWMs) (Fig. S1). These results show that the HA Frankenbody visualizes HA-tagged proteins in Drosophila tissues.

Next, we observed the HA Frankenbody-GFP in live tissues. HA Frankenbody-GFP and HA-tagged protein were co-expressed in LECs, LBWMs or fat bodies, and observed in live tissues by confocal microscopy (Fig. 2A-C, Fig. S2). LECs and LBWMs were observed through the cuticle in intact animals. Conversely, fat bodies were dissected, as they were too far from the cuticle for a high-magnification objective lens. Upon co-expression of 3xHA-tagged organelle markers, HA Frankenbody-GFP showed the typical localization pattern associated with each organelle (Fig. 2, Fig. S2). Collectively, HA Frankenbody allows imaging of HA-tagged constructs in live Drosophila tissues.

We have recently reported that the spinster-positive tubular autolysosomal (tAL) network functions in the remodeling of the abdominal muscles during metamorphosis (Murakawa et al., 2020). The tAL network is fragile and highly sensitive to dissection and fixation processes (Murakawa et al., 2020; Johnson et al., 2015). Hence, it is almost impossible to observe the tAL network by immunostaining. In contrast to that, HA Frankenbody-mCherry visualized the spin-3xHA-positive tAL network in 20 h after puparium formation (APF) dorsal internal oblique muscles (DIOMs) (Fig. 2D). Subsequent to dissection and fixation, the structure was lost (Fig. 2D), showing that HA Frankenbody is suitable for imaging of fragile structures that are highly sensitive to fixation.

The tandem fluorescent protein-tagged Atg8 homolog, GFP-mCherry-LC3/Atg8, allows the assessment of autophagy flux because of the pH sensitivity of GFP (Kimura et al., 2007). Analogous to this reporter, we generated a transgenic fly carrying UAS-HA Frankenbody-GFP-mCherry (Table 1) to monitor the autophagic degradation of HA-tagged protein. Both GFP and mCherry fluoresce at the neutral pH of the cytosol (Fig. 3A left). On the other hand, only mCherry fluoresces at the acidic pH in lysosomes because the pKa is around 6.0 for GFP, and it is quenched in acidic conditions (Fig. 3A right). To validate the TF system, both HA Frankenbody-TF and HA-tagged constructs were expressed in the larval fat bodies. Third instar larvae (3IL) were fed or starved for 4 h, and fat bodies were observed (Fig. 3B,C). In fed condition, expression of LacZ, GAPDH1-3xHA or TOM20-3xHA induced only a low level of mCherry-single-positive puncta (Fig. 3B,C). In contrast, expression of HA-ref(2)p, an established autophagy cargo protein (Komatsu et al., 2007; Nezis et al., 2008), induced robust formation of mCherry-single-positive puncta, suggesting HA-ref(2)p is degraded in the lysosome even under fed conditions (Fig. 3B,C). We obtained a similar result in 1-week-old adult indirect flight muscles (Fig. S3A,B). Furthermore, nutrient starvation significantly elevated the lysosomal degradation of the TF-tagged HA Frankenbody (Fig. 3B,C). We confirmed that the mCherry-single-positive puncta were colocalized well with LysoTracker dye and Cathepsin L (Fig. S3C). These results indicate that mCherry accumulated in the acidic lysosomal compartments.

To examine whether the degradation of the TF reporter depends on autophagy, we used the Atg5 null to examine the formation of mCherry-single-positive puncta in the fat body co-expressing HA-ref(2)p and HA Frankenbody-TF. Atg5 is an essential gene for autophagy (Mizushima et al., 1998; Kim et al., 2016). As shown in Fig. 3D,E, loss of Atg5 significantly blocked mCherry-single-positive puncta formation. We further confirmed that the degradation of HA Frankenbody-TF depended on autophagy even in the absence of HA-fused construct (Fig. S3D,E).

We noticed that, under fed conditions, HA-ref(2)p expression caused the bright spherical structures positive for both GFP and mCherry (Fig. 3B,D, Fig. S4A), and the number and size of the GFP-positive spherical structures significantly decreased in parallel with their degradation upon starvation (Fig. 3D,E, Fig. S4A,B). In addition, the Frankenbody-TF-positive spherical structures were negative for Lysotracker dye (Fig. S4C). These results indicate that the spherical structures are not lysosomal compartments but liquid or gel-like ref(2)p-droplets in the cytosol (Pircs et al., 2012; Kageyama et al., 2021).

Next, we examined whether HA Frankenbody-TF expression triggers degradation of HA-fused constructs. In the starved larval fat bodies, HA Frankenbody-TF expression had little noticeable effect on the protein level of GAPDH1-3xHA or TOM20-3xHA (Fig. S5A,B). In contrast, the level of HA-ref(2)p was elevated by HA Frankenbody-TF expression (Fig. S5C), suggesting that HA Frankenbody-TF stabilizes HA-ref(2)p. To test whether the stabilization depends on autophagy, we examined the loss of Atg5 on HA-ref(2)p level in the presence or absence of HA Frankenbody-TF. The elevation of HA-ref(2)p induced by HA Frankenbody-TF was not dependent on Atg5 (Fig. S5D), indicating that HA-ref(2)p is stabilized in an autophagy-independent mechanism. Furthermore, it has been shown that p62/ref(2)p is also degraded by the ubiquitin-proteasome system (Song et al., 2016). Therefore, it is plausible that HA Frankenbody-TF binds and stabilizes HA-ref(2)p through protection from the ubiquitin-proteasome system. Altogether, the above results demonstrate that the HA Frankenbody-TF visualizes the lysosomal degradation of HA-tagged protein of interest in Drosophila tissues.

By using time course microscopy to observe fat bodies expressing HA Frankenbody-TF from early 3IL to white prepupa, we found that both GFP and mCherry fluoresce in the enlarged structures in the white prepupal fat bodies (Fig. 4A-C, white dotted circle in B). For quantification, the bright spherical structures positive for both GFP and mCherry were excluded (Fig. 4B, cyan arrowhead) because they were LysoTracker-negative liquid or gel-like ref(2)p-droplets (Fig. S4, Fig. 3D,E). Parallel to the increase of GFP and mCherry double-positive objects, the number of mCherry-single-positive objects decreased (Fig. 4B,C, white arrowhead). These results suggest that the GFP and mCherry double-positive structures are lysosome-related structures with relatively higher pH. To test this possibility, white prepupal fat bodies expressing both HA Frankenbody-TF and HA-ref(2)p were stained with LysoTracker Deep Red: a dye for acidic organelles. As shown in Fig. 4D, the enlarged GFP and mCherry double-positive structures were also positive for LysoTracker dye, indicating that they are acidic lysosomal compartments. Although the LysoTracker probe accumulates and fluoresces in the acidic compartments, its intensity is largely independent of pH (Guha et al., 2014). In contrast, the LysoSensor dye exhibits a change in fluorescent intensity with luminal pH. Therefore, we compared the intensity of LysoSensor Green in the compartments in starved 3IL or white prepupal fat bodies. The LysoSensor showed a significantly lower intensity in the white prepupal fat body than the starved 3IL fat body (Fig. 4E,F), indicating that lysosomal acidification is regulated in the prepupal fat bodies. We further confirmed that lysosomal protease activity, as monitored by Magic Red, also decreased more in the white prepupal fat body than it did in the starved 3IL fat body (Fig. 4G,H). From these results, we conclude that lysosomal function is developmentally downregulated in the white prepupal fat body.

It is thought that robustly induced autophagy is the dominant cause of the enlargement of lysosomal compartments in the last instar larval fat body (Rusten et al., 2004). To test this model, we examined whether there was a correlation between autophagy induction and the size of the lysosome. The number of GFP-Atg8 puncta (autophagosomes) and the size of LysoTracker Red-positive compartments (lysosomes) were quantified in starved 3IL or white prepupal fat bodies (Fig. 5A-D) (Kabeya et al., 2000). The ratio of the lysosome area per number of autophagosomes was significantly higher at the white prepupal stage than at the starved 3IL stage (Fig. 5D), suggesting that not only autophagy induction but also other mechanisms contribute to the enlargement of lysosomes. Next, we tested the loss of Atg5 on the size of Lamp1-3xmCherry-positive lysosomes, to examine further the contribution of autophagy to the enlargement of lysosomes (Hegedűs et al., 2016). Strikingly, the enlargement was observed even in Atg5-null conditions (Fig. 5E-F). The size of the lysosomes was affected only modestly by the loss of Atg5 in the white prepupal fat body. Next, we performed a flux assay using chloroquine, a well-known lysosomotropic agent that increases lysosomal pH and induces vacuolation (Zhou et al., 2020). Chloroquine feeding significantly increased the size of the Lamp-positive lysosomes in white prepupal LECs (Fig. S6A-C). In sharp contrast, chloroquine feeding did not induce the enlargement of lysosomes nor the decrease in lysosomal protease activity in the fat body at the same stage (Fig. S6A,B,D-F), suggesting lysosomal function is downregulated in the white prepupal fat body.

Which trafficking pathway does play a role in the enlargement of lysosomes? Besides autophagy, endocytic pathways also deliver cargoes to the lysosomes: therefore, we tested temporal blockade of endocytosis on the enlargement of lysosomes. For this purpose, Rab5^S43N, a dominant-negative mutant of Rab5, was temporally expressed in the late larval fat body by Lsp2-GAL4 (Cherbas et al., 2003). Consistent with Fig. 5E, loss of Atg5 or FIP200/Atg17 hardly affected the size of LysoTracker Red-positive lysosomes (Fig. 5G,H,I). In sharp contrast, Rab5^S43N overexpression significantly reduced the size of lysosomes (Fig. 5G,J). These results show that endocytosis, but not autophagy, contributes dominantly to the enlargement of lysosomes in the prepupal fat body.

We explored the biological significance of the downregulation of lysosomes in the fat body. For this purpose, we tested fat body-targeted Mitf overexpression by Lsp2-GAL4 on the enlargement of lysosomes and the development of Drosophila. Mitf is the single MiTF/TFE family gene in Drosophila and is a master regulator of the autophagy-lysosome pathway (Zhang et al., 2015; Bouché et al., 2016). Thus, Mitf overexpression upregulates lysosomal functions through transcription activation of a series of lysosome-related genes, including v-ATPase subunits required for the acidification of lysosomes (Bouché et al., 2016). We confirmed the transcriptional activation of lysosome-related genes by Mitf overexpression in the white prepupal fat body (Fig. 6F, bottom row). Mitf overexpression markedly suppressed the increase in lysosome pH and the enlargement of lysosomes in the white prepupal fat body (Fig. 6A-D), probably through the forced activation of lysosomes. It is important to note that Lsp2-GAL4 drives expression only after late 3IL. At 13 days after egg laying, almost all progenies became adults in control (Fig. 6E). By contrast, fat body-targeted Mitf overexpression induced severe lethality at the pupal stage (Fig. 6E). The Mitf overexpression-induced lethality was independent of Atg9, an essential gene for autophagy (Wen et al., 2017) (Fig. S7), indicating that autophagy is not involved in the Mitf-induced lethality. These data suggest that the downregulation of lysosomes in the last instar larval and white prepupal fat body plays a role in the metamorphosis of Drosophila.

Last, we performed a comparative RNA-seq to explore transcriptional regulation of lysosome-related genes in the fat body. As shown in Fig. 6F, the mRNA level of several lysosome-related genes, including v-ATPase subunits, spinster and lysosomal α-N-acetylglucosaminidase (CG13397), significantly dropped in the white prepupal fat body compared with 3IL fat body. As it is known that hypofunction of them leads to lysosomal dysfunction and swelling (Dermaut et al., 2005; Rigon et al., 2021), their transcriptional downregulation might be a key to understanding the developmental regulation of lysosomes in the fat body.

A toolkit of HA Frankenbody allows a versatile localization and dynamic analysis of HA-tagged POIs in Drosophila. A combination of HA Frankenflies and the Zurich FlyORFeome library enables large-scale localization screening in vivo. The FlyORFeome project aims to develop a comprehensive in vivo genome-wide UAS-ORF library. Notably, the UAS-transgenes were fused with a 3xHA tag at the C terminus (Bischof et al., 2012). Figs 1 and 2 demonstrate that our strategy is powerful enough for in vivo subcellular localization analysis. Cytosolic HA Frankenbody does not recognize the HA tag exposed in the luminal or extracellular space. However, we think this feature is advantageous for analyzing the membrane topology of a POI. If an inserted HA tag is recognized by HA Frankenbody, one can tell that the HA-fused segment is exposed to the cytosol.

Immunostaining is a standard protocol for the localization analysis of epitope-tag fused constructs. Nevertheless, chemical fixation and permeabilization by detergents often cause artifacts such as changes in the morphology of organelles, loss of epitopes or mislocalization of proteins (Richter et al., 2018). As such, we analyzed the tAL network, which is highly sensitive to a normal immunostaining process (Murakawa et al., 2020). A combination of spin-3xHA and HA Frankenbody-mCherry succeeded in imaging this tubular network in live muscle cells (Fig. 2D). Hence, the HA Frankenbody system is especially suitable for fragile structures labeled with HA-tagged POI where fluorescent protein-tagging is not a viable option.

TF-tagged HA Frankenbody enables monitoring of autophagic degradation of HA-fused POI in lysosomes (Fig. 3). HA-ref(2)p, a fly homolog of p62/SQSTM1 that is an established autophagy cargo, was more efficiently degraded compared with other constructs in both muscles and fat bodies (Fig. 3B,C and Fig. S3A,B). We also confirmed that the degradation of HA-ref(2)p depends on autophagy (Fig. 3D,E). Using a combination of TF-tagged HA Frankenbody and HA-fused FlyORF library allows the assessment of autophagic degradation of POIs without establishing a new tandem-fluorescent protein-fused transgenic line. Autophagy specifically sequesters target organelles, including ER, mitochondria, peroxisomes and lysosomes, for degradation (Morishita and Mizushima, 2019); however, the physiological roles of the organelle-phagy are not fully understood, especially in multicellular organisms. Thus, it would be interesting to examine the degradation of markers for each organelle in various tissues, developmental stages and nutrient conditions by the TF-tagged HA Frankenbody strategy. in addition, TF-tagged HA Frankenbody is a powerful tool for exploring cargo specificity in autophagy. Most selective autophagy cargoes have a short linear sequence motif, called LIR, LRS or AIM, that physically interacts with Atg8 family proteins (Kalvari et al., 2014). It is now feasible to test all predicted AIM-containing proteins by combining TF-tagged HA Frankenbody and the FlyORF library. A systematic approach would provide insights into cargo recognition by Atg8 on the autophagic membrane.

We observed accumulated and enlarged autolysosomes in the last instar larval and white prepupal fat body, consistent with previous reports (Rusten et al., 2004; Butterworth et al., 1988; Butterworth and Forrest, 1984). However, we propose here a new mechanistic model explaining this phenomenon. Our data support a model where decreased autophagic flux is the dominant cause of autolysosome accumulation. First, lysosomal acidification was downregulated in prepupal fat bodies (Fig. 4E,F). Second, blockade of lysosomal activity by chloroquine feeding did not significantly affect the size and activity of lysosomes at this stage (Fig. S6). Third, autophagy was dispensable for the enlargement of lysosomes (Fig. 5E-G), indicating that lysosomal activity is developmentally downregulated in an autophagy-independent manner.

The next question to be addressed is the mechanism of the downregulation of lysosomal activity in the last instar larval and prepupal fat body. As lysosomal acidification was altered compared with the early 3IL stage (Fig. 4E,F), the expression level of vha genes, encoding v-ATPase subunits required for lysosomal acidification, or assembly of V0 and V1 V-ATPase subcomplex, could be attenuated (Collins and Forgac, 2020). The regulation of lysosomal pH is complex, and it could be affected by anions, cations, membrane voltage, lysosomal lipids, etc. (Mindell, 2012; Sillence, 2013). Therefore, it is possible that the expression of other lysosome-related genes, such as spinster, is developmentally regulated in the fat body (Dermaut et al., 2005). To test the possibilities, we performed a comparative RNA-seq of 3IL and white prepupal fat bodies. We predict that Mitf is dispensable for the downregulation of lysosomes in the fat body, because most of the genes regulated by Mitf were not transcriptionally affected in the white prepupal fat body (Fig. 6F). The mRNA levels of several vha genes, spinster and lysosomal α-N-acetylglucosaminidase (CG13397) were significantly decreased in white prepupal fat bodies (Fig. 6F). CG13397 is a fly homolog of human α-N-acetylglucosaminidase that is a causative gene of Mucopolysaccharidosis type IIIB – a lysosomal storage disorder (Rigon et al., 2021). Further studies are needed to address whether the transcriptional regulation contributes to the enlargement of lysosomes in the last instar and white prepupal fat body.

Rab5-mediated endocytosis was indispensable for the swelling of lysosomes (Fig. 5G,J). Therefore, it is also possible that increased endocytic flux to lysosomes is a driver of the enlargement to some extent. It should be examined whether the endocytic activity is upregulated more in the white prepupal fat body than in the 3IL fat body. However, this is challenging because lysosomal activity is significantly decreased in the white prepupal fat body. Alternatively, an elevated basic metabolite could cause the downregulation of lysosomes. Metabolome analysis of early 3IL and white prepupal fat bodies would provide valuable clues to the mechanism behind the lysosomal swelling.

What is the biological significance of the block in autophagic flux in fat bodies? Forced activation of lysosomes by Mitf overexpression suppressed the accumulation of enlarged lysosomes in the fat body (Fig. 6C,D). Moreover, Mitf overexpression in the last instar larval fat body induced severe lethality at the pupal stage (Fig. 6E), although its precise mechanism has not yet been revealed. These results suggest that the downregulation of lysosomes in the fat body has a beneficial role in Drosophila development. Therefore, we hypothesize that the last instar larvae store autolysosomes in fat bodies as a nutrient source for the next harsh starvation period: the pupal stage. The accumulated autolysosomes in the larval fat body might be degraded and recycled in the pupal stage. Supporting this idea, time-course electron microscopy of the fat body showed that the number of autophagic vacuoles peaked at 1 day APF and then decreased during the remainder of metamorphosis (Butterworth et al., 1988). Alternatively, degradation of endocytosed hemolymph proteins or plasma membrane proteins in the fat body may have some kind of deleterious effect, and the ecdysone hormone blocks their degradation for proper development. The exciting questions to be explored are: (1) what are the dynamics of the enlarged autolysosomes in the pupal fat body?; (2) what is the effect of forced blockade of autolysosome turnover throughout the pupal period?; and (3) what is the mechanism of how fat body-targeted Mitf overexpression induces pupal lethality?

Besides the utilities described in this paper, HA Frankenbody could be applied to proximity labeling of HA-tagged POI by HA Frankenbody-APEX2 or BioID (Bosch et al., 2021) and visualization of HA-POI in the extracellular space in vivo by a secreted form of HA Frankenbody, etc. Our study provides a new platform for the functional analysis of HA-tagged proteins in vivo. This strategy will be widely used, contributes to the fly community, and is also being applied to in vivo studies in other model organisms, such as worms, amphibians, zebrafish, plants and mice.

The following antibodies were used: mouse anti-HA (1:1000; clone TANA2; M180-3, MBL), rabbit polyclonal anti-GFP (1:1000; 598, MBL), mouse anti-insect Cathepsin L (1:200; clone 193702; MAB22591, R&D Systems), anti-mouse IgG Alexa Fluor 594 conjugate (1:400; A11005, Thermo Fisher Scientific), anti-mouse IgG Alexa Fluor 405 conjugate (1:400; A48255, Thermo Fisher Scientific), HRP-conjugated AffiniPure goat anti-mouse IgG (1:10,000; 115-035-166, Jackson ImmunoResearch), HRP-conjugated AffiniPure goat anti-rabbit IgG (1:10,000; 111-035-144, Jackson ImmunoResearch), LysoTracker Red DND-99 (Thermo Fisher Scientific), LysoTracker Deep Red (Thermo Fisher Scientific), Lysosensor Green DND-189 (Thermo Fisher Scientific), Magic Red Cathepsin B assay kit (ImmunoChemistry Technologies) and Chloroquine diphosphate (TCI).

Flies were reared at 25°C unless otherwise stated. For tissue-targeted gene expression, the following drivers were used: DMef2-GAL4 for muscle, Cg-GAL4 for the fat body, pnr-GAL4 for larval epidermal cells and da-GAL4 for broad tissues. UAS-LacZ was used as a control. Genotypes are described in Table S1. All genetic combinations were generated by standard crosses. Genotypes of flies used in this study include the following: (1) Atg5^5cc5/FM7 actin-GFP (from J. H. Lee, University of Michigan, Ann Arbor, MI, USA; Atg5 null), (2) endo_promoter-Lamp:3xmCherry1-9M (from G. Juhasz, Hungarian Academy of Sciences, Budapest, Hungary), (3) pnr-GAL4 (from D. Umetsu, Tohoku University, Sendai, Japan), (4) w; UAS-HA:Ref(2)P_7M/TM3 Sb (from S. Loh, University of Cambridge, UK), (5) y w; UAS-GFP:Atg8a (from T. Neufeld, University of Minnesota, Minneapolis, USA), (6) w; UAS-Mitf (96E) (from F. Pignoni, Upstate Medical University, Syracuse, NY, USA), (7) w¹¹¹⁸; P{w^+mC=Cg-GAL4.A}2 (Bloomington Drosophila Stock Center, BL7011; Cg-GAL4), (8) y¹ w*; P{w^+mC=GAL4-Mef2.R}3 (BL27390; DMef2-GAL4), (9) w¹¹¹⁸; P{w^+mC=UAS-lacZ.B}Bg4-1-2 (BL1776; UAS-LacZ), (10) y w; P{w[+mC]=Lsp2-GAL4.H}3 (BL6357; Lsp2-GAL4), (11) UAS-Smd2:3xHA (FlyORF, Zurich ORFeome project, F003874), (12) UAS-γCOP:3xHA (F000931), (13) UAS-βTub97EF:3xHA (F001206), (14) UAS-Tom20:3xHA (F003545), (15) UAS-Gapdh1:3xHA (F002305), (16) UAS-CkIIβ:3xHA (F004937), (17) UAS-Pex11:3xHA (F004080), (18) UAS-Atg18:3xHA (F002805), (19) UAS-Atg5:3xHA (F003001), (20) w; Atg9^d51/CyO (from GC. Chen; Atg9 null), (21) w; Atg9^Gal4KO/CyO (from G. C. Chen, Academia Sinica, Taipei, Taiwan; Atg9 null) (22) w; FIP200^3F5/TM6B (from J. H. Lee; FIP200 null) (Kim et al., 2013), (23) w; FIP200^4G7/TM6B (from J. H. Lee; FIP200 null) (Kim et al., 2013), (24) w; UAS-mCherry (from A. Kiger, University of California, San Diego, USA) and (25) w; P{w[+mC]=UAS-Rab5.S43N}2 (BL42703; UAS-Rab5^S43N). New genotypes generated during this study include the following: (26) w¹¹¹⁸; UAS-Spinster:3xHA^attP40, (27) w¹¹¹⁸; UAS-Frankenbody^HA:GFP^attP40, (28) w¹¹¹⁸; UAS-Frankenbody^HA:mCherry ^attP40, (29) w¹¹¹⁸; UAS-Frankenbody^HA:GFP:mCherry ^attP40, (30) w¹¹¹⁸; QUAS- Frankenbody^HA:GFP^attP2, (31) w¹¹¹⁸; QUAS-Frankenbody^HA:mCherry ^attP2, (32) w¹¹¹⁸; QUAS-Frankenbody^HA:GFP:mCherry ^attP2, (33) w¹¹¹⁸; UASt-HA-Frankenbody:GFP/CyO; pnr-GAL4/TM6c Sb Tb, (34) w¹¹¹⁸; UASt-HA-Frankenbody:GFP/CyO; DMef2-GAL4/TM6c Sb Tb, (35) w¹¹¹⁸; Cg-GAL4, UASt-HA-Frankenbody:GFP/CyO, (36) w¹¹¹⁸; UASt-HA-Frankenbody:mCherry/CyO; DMef2-GAL4/TM6c Sb Tb and (37) w¹¹¹⁸; Cg-GAL4, UASt-HA-Frankenbody:GFP:mCherry/CyO.

Laid eggs were reared at 21°C on standard cornmeal/yeast/agar media for 21 h. Then they were reared at 25°C until they grew to 3IL or to white prepupa. For nutrient starvation, 2IL larvae were transferred to fresh food and kept at 25°C for 24 h. The 3IL were reared in standard media (fed) or in 20% sucrose/PBS solution for 4 h (starved). For chloroquine feeding, laid eggs were grown in standard media with 3 mg/ml of chloroquine diphosphate until the indicated stage.

Standard molecular biology techniques were used to construct plasmid vectors. The DNA polymerase KOD One (TOYOBO, Tokyo, Japan) was used for PCR amplification of all the objective sequences. For the expression vector of spinster-3xHA, the spinster coding sequence was amplified with the following primer set, 5′-CACCGAATTCACCATGTCGCTGAAACACCAGAAGCAATC-3′ and 5′-CTCGAGGGCAATCTGACCGCGGCTGATC-3′, from the cDNA. Amplified sequence was inserted into pENTR-D-TOPO vector with TOPO cloning (Thermo Fisher Scientific). The inserted coding sequence was transferred to the pTWH vector with the Gateway protocol using LR Clonase II Enzyme mix (Thermo Fisher Scientific). Construction of the expression vectors of C-terminal fluorescent protein-tagged HA Frankenbody under UAS control (pUASt-attB-Frankenbody^HA-GFP, pUASt-attB- Frankenbody^HA -mCherry, and pUASt-attB- Frankenbody^HA -GFP-mCherry) was performed as follows: as a template, pCMV-15F11-HA-mEGFP was used (Zhao et al., 2019) and for pUASt-attB- Frankenbody^HA -GFP, Frankenbody^HA -GFP fragments were amplified with the following primer sets: 5′-ATAGGGAATTGGGAACACCATGGCCGAGGTGAAGCTGG-3′ and 5′-ACCCTCGAGCCGCGGCTACTTGTACAGCTCGTCCATGCC-3′. For pUASt-attB- Frankenbody^HA -mCherry, the fragments of Frankenbody^HA and mCherry were amplified with the following primer sets: 5′-ATAGGGAATTGGGAACACCATGGCCGAGGTGAAGCTGG-3′ and 5′-ACCGCTTCCACCTCCACCGCTTCCG-3′ for Frankenbody^HA; and 5′-GGAGGTGGAAGCGGTATGGTGAGCAAGGGCGAGGAGGA-3′ and 5′-TGGACGAGCTGTACAAGTGACCGCGGCTCGAGGGT-3′ for mCherry. For pUASt-attB- Frankenbody^HA -GFP-mCherry, the fragments of Frankenbody^HA -GFP and mCherry were amplified with the following primer sets: 5′-ATAGGGAATTGGGAACACCATGGCCGAGGTGAAGCTGG-3′ and 5′-GCATGGACGAGCTGTACAAGGGTGGACCG-3′ for Frankenbody^HA -GFP; and 5′-ACAAGGGTGGACCGATGGTGAGCAAGGGCGAGGA-3′ and 5′-ATGGACGAGCTGTACAAGTGACCGCGGCTCGAGGGT-3′ for mCherry. Each set of fragments was assembled with pUASt-attB vector digested by EcoRI and NotI using Gibson assembly. The construction of the expression vectors of the C-terminal fluorescent-tagged Frankenbody^HA under QUAS control (pQUASt-attB- Frankenbody^HA -GFP, pQUASt-attB- Frankenbody^HA -mCherry, pQUASt-attB- Frankenbody^HA -GFP-mCherry) was performed as follows: the fragments of Frankenbody^HA -GFP, Frankenbody^HA -mCherry, and Frankenbody^HA -GFP-mCherry were amplified with the following primer sets: 5′-GTTAGATCTCACCATGGCCGAGGTG-3′ and 5′-CTTCTCGAGCTACTTGTACAGCTCGTCCATGCCGAG-3′ for Frankenbody^HA -GFP; 5′-GTTAGATCTCACCATGGCCGAGGTG-3′ and 5′-CTTCTCGAGTCACTTGTACAGCTCGTCCATGCCGCC-3′ for Frankenbody^HA -mCherry; and 5′-GTTAGATCTCACCATGGCCGAGGTG-3′ and 5′-CTTCTCGAGTCACTTGTACAGCTCGTCCATGCCGCC-3′ for Frankenbody^HA -GFP-mCherry. The amplified fragments were purified and digested using BglII and XhoI. The digested DNA fragments were ligated to pQUASt-attB, also digested by BglII and XhoI using a Ligation-Convenience kit (NIPPON GENE). All the resultant vectors were validated by sequencing, and were then injected into embryos for P-element or phiC31 insertion (Wellgenetics).

Muscle preparations in larval fillets and pupal abdomens were performed as previously described (Ribeiro et al., 2011). 3IL or pupae were pinned on a sylgard-covered petri dish in dissection buffer [5 mM HEPES, 128 mM NaCl, 2 mM KCl, 4 mM MgCl₂ and 36 mM sucrose (pH 7.2)]. The animals were pinned flat and fixed at room temperature for 20 min (4% PFA, 50 mM EGTA in PBS). The samples were the unpinned and blocked at room temperature for 30 min (0.3% bovine serum albumin, 2% goat serum, 0.6% Triton X100 in PBS), incubated with primary antibody overnight at 4°C, washed (0.1% Triton in PBS), then incubated with Alexa Fluor568-conjugated secondary antibody (Thermo Fisher Scientific) for 2 h at room temp. The immunostained samples were washed and mounted in FluorSave reagent (Merck Millipore). LEC preparations in 3IL larvae abdomens were performed as previously described (Tenenbaum and Gavis, 2016). After dissection and fixation, the larval body wall muscles were removed from the fillet using fine forceps. Samples were then unpinned and blocked at room temperature for 30 min (0.1% Triton X100, 0.3% BSA in PBS). Subsequent steps were performed as described above for muscles.

Fat body preparations in 3IL larvae were performed as follows. 3IL larvae were dissected with forceps in the dissection buffer. The fat bodies were fixed at room temperature for 20 min (4% PFA, 50 mM EGTA in PBS), blocked at room temperature for 30 min (1% Tween20, 2% BSA in PBS). Then the samples were incubated with primary antibody overnight at 4˚C and incubated with Alexa Fluor405-conjugated secondary antibody (Thermo Fisher Scientific) for 2 h at room temp.

Staged larvae were dissected in PBS, and a lobe of the fat body was collected. The fat bodies were stained with LysoTracker Red (1:10,000 in PBS for 2 min), Lysotracker Deep Red (1:1000 in PBS for 2 min) or LysoSensor Green (1:2000 in PBS for 2 min) at room temperature.

Staged larvae were dissected in PBS with forceps. The collected fat body lobes were stained with Magic Red (1/250 in PBS) for 2 min at room temperature.

For imaging of live LECs, white prepupae were mounted between slide-glass and cover-glass with a spacer. For imaging of live LBWMs, third instar larvae were immobilized by making a tiny hole on the head with fine forceps in 4% PFA, and mounted between slide-glass and cover-glass with a spacer. Live fat bodies were mounted between slide-glass and cover-glass with PBS. LECs, LBWMs and fat bodies were observed on a confocal microscope FV3000 (Olympus) with a 60× silicone/1.30 NA UPlanSApo (Olympus). The image acquisition software used was Fluoview (Olympus). The exported images were adjusted and analyzed using ImageJ.

Fat bodies were collected from staged five larvae or pupae. The isolated fat bodies were lysed in the TRIzol (Thermo Fisher Scientific) and frozen at −80°C. The lysates were quickly thawed and homogenized through a 27-gauge needle for 10 strokes. The total RNA was collected following the manufacturer's instructions. The samples were then further purified using ReliaPrep tissue RNA Miniprep System (Promega). cDNA libraries were constructed from the total RNAs using TruSeq stranded mRNA library platform (Illumina). The libraries were sequenced with paired-end sequencing by NovaSeq6000 (Illumina). The acquired FASTQ files were quantified using salmon (Patro et al., 2017) and statistical analysis was carried out of the fold change of the gene expression levels (early 3IL versus WP, or control at WP versus Mitf OE at WP) using R's Bioconductor plug-ins (DESeq2 with independent hypothesis weighting) (Love et al., 2014; Ignatiadis et al., 2016). The threshold of the adjusted P-value was set as 0.1.

Fat bodies or whole larvae without gut were collected from three animals. The samples were lysed directly in 1.5× sample loading buffer [100 mM Tris-HCl (pH 6.8), 3% SDS, 15% glycerol and 0.03% Bromophenol Blue]. Equal amounts of proteins per sample were subjected to SDS-PAGE. The separated proteins were transferred to Immobilon-P^SQ PVDF transfer membrane (Merck Millipore) with Trans-Blot Turbo Transfer System (Bio-Rad). The membranes were blocked in blocking buffer [10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.1% Tween 20 and 1% BSA] for 30 min, incubated with the primary antibodies in Can Get Signal Solution 1 (TOYOBO) overnight at 4˚C and then incubated with secondary antibodies in blocking buffer for 2 h at room temp. Immunoreactive bands were then detected using ECL plus (GE Healthcare Life Sciences) and FUSION (Vilber Lourmat). The images were processed using ImageJ.

We analyzed up to 10 images acquired from five lobes of the fat body from five animals for each experiment. For all analysis, images were cropped to 81.59×81.59 µm and were analyzed using ImageJ macro language. Below is the analysis process in brief. In Fig. 3C,E, Fig. S3B,E, the intensity of mCherry-only positive dots was quantified. The fluorescence intensity of the GFP channel in each pixel was subtracted from the corresponding pixel of mCherry intensity. The map was binarized and the mean intensity and area of each object was analyzed. In Fig. S4B, the area of bright spherical structures positive for both GFP and mCherry was quantified. Both GFP and mCherry channel images were binarized. They were then multiplied together to extract a double-positive area. The resultant map was used to analyze the area of each object. In Fig. 4C, the number of mCherry-positive and GFP-negative acidic compartments was counted using the same procedure as Fig. 3C,E and Fig. S3E. Both GFP- and mCherry-positive large objects were manually counted because it was challenging to segment them automatically. In Figs 4F,H, 6B and Fig. S6F, the mean intensity of Lysosensor Green of lamp-3xmCh-positive objects was quantified. First, the mCherry channel was binarized and filtered to exclude noise by size and circularity. Then, the mean intensity of Lysosensor Green was calculated for each binarized mCherry-positive object. In Fig. 5B, the number of GFP-Atg8-positive puncta was counted. The GFP channel image was binarized and filtered to remove noise. The number of small puncta was then counted. To measure the size of lysosomes in the 3IL fat body, the images were binarized and filtered to remove noise. The area of each object was then quantified. In the white prepupal fat body, lysosome size was analyzed manually using Fiji, because of the same issues with segmentation described above.

Each experiment was performed at least three times and at least five different cohorts of unique flies were analyzed. All replicate experiments were performed in parallel with controls. Data are mean±s.d. When more than two genotypes or treatments were used in an experiment, the statistical analysis was performed using Tukey's test, Sidak's test or unpaired t-test on Prism8 and Prism9 software. A two-tailed unpaired Student's t-test was used to compare two means. P<0.05 was regarded as statistically significant (*P<0.05, **P<0.001, ***P<0.0001 and ****P<0.00001).

We are grateful to F. Pignoni (Upstate Medical University), S. Loh (University of Cambridge), G. Juhasz (Hungarian Academy of Sciences), G. Davis (UCSF), J. H. Lee (University of Michigan), D. Umetsu (Tohoku University), T. Neufeld (University of Minnesota), G. C. Chen (Academia Sinica), A. Kiger (UCSD), Bloomington Drosophila Stock Center, VDRC and FlyORF for reagents. We are grateful to J. Mathieu (College de France) for the helpful discussion. We thank the Biomaterials Analysis Division of the Tokyo Institute of Technology for the DNA sequencing. We are grateful to M. Landekic (McGill University) for English editing.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200243.