A conserved myotubularin-related phosphatase regulates autophagy by maintaining autophagic flux

Cells rely on macroautophagy (hereafter autophagy) to deliver sequestered cytoplasmic cargoes to lysosomes for degradation. Basal autophagy levels maintain cellular homeostasis, and activated autophagy mitigates several cellular stresses, including nutrient deprivation. Autophagy begins with the formation of an isolation membrane that elongates and encloses cargoes to eventually form a closed double-membrane autophagosome that fuses with lysosomes to become an autolysosome (Yu et al., 2018). Lysosomal enzymes, including cathepsins, break down autophagic cargoes into macromolecules, which are later effluxed into the cytoplasm for use in anabolic cellular processes. Pioneering yeast experiments defined the core autophagic molecular mechanisms (Tsukada and Ohsumi, 1993). However, metazoans such as Drosophila melanogaster possess an enhanced repertoire of autophagic machinery beyond those required in yeast, thus providing a multicellular in vivo experimental system to study autophagy (Mulakkal et al., 2014; Zhang and Baehrecke, 2015).

From initiation to completion, autophagy is regulated by >30 conserved core autophagy-related (Atg/ATG) proteins (Galluzzi et al., 2017). Autophagy initiation and autophagosome biogenesis require the Atg1–ULK1 serine–threonine kinase complex, the vacuolar protein sorting (Vps34) class III phosphoinositide (PI) 3-kinase (PI3K) complex, and the Atg8 conjugation system. Autophagosome–lysosome fusion utilizes dozens of proteins, including many shared with the endocytic pathway and the autophagosomal SNARE protein Syntaxin 17 (Takáts et al., 2013; Itakura et al., 2012). When lysosomal enzymes complete autophagic cargo degradation, then autolysosome reformation contributes to regeneration of the lysosomal pool, which further promotes homeostasis (Yu et al., 2010). Both excessive and insufficient autophagy can promote disease by dysregulating autophagic flux, a term that describes the quantity of autophagic degradation (Dowling et al., 2015; Levine and Kroemer, 2019; Yoshii and Mizushima, 2017), making rate control a key component to each step of autophagy.

Vesicular membrane identity and trafficking rely largely upon PI identity, concentration, and coincidence (Schink et al., 2016). PI kinases (PIKs) and phosphatases can switch the 3-, 4-, and 5-hydroxyl group phosphates of the inositol ring to generate seven distinct cellular PIs. Membrane-embedded PIs recruit effectors to control membrane dynamics via multiple mechanisms, including membrane fusion, scission, and shaping. In the endolysosomal system, PI(3)P is the major PI species on endosomes following PI conversion from PI(4,5)P₂, which is required for endocytosis (Wallroth and Haucke, 2018). Recycling endosomes rely on a PI(3)P loss (Campa et al., 2018) and PI(4)P gain (Ketel et al., 2016), while degradative PI(3)P-positive endosomes require PI(3,5)P₂ production by the PI(3)P 5-kinase PIKFYVE for endosomal sorting complexes required for transport (ESCRT)-III recruitment and delivery to multivesicular bodies (Kim et al., 2014; Hasegawa et al., 2017). Similar PI conversion events occur during uptake and degradation of either extracellular fluids or particles by macropinocytosis or phagocytosis, respectively (Zoncu et al., 2009), and phagosome maturation coincides with PI(3)P reduction (Levin et al., 2015). Likewise, autophagy progression from start to finish relies upon PI identity changes. Autophagy initiation requires recruitment of the class III PI3K complex I to preautophagosomal structures (Funderburk et al., 2010), which results in PI(3)P production and recruitment of autophagosome-forming effectors (Lystad and Simonsen, 2016). Meanwhile, autophagosome–lysosome fusion requires multiple PI conversion events. For example, fusion requires production of PI(4)P at autophagosomes by PIK4IIα (Wang et al., 2015), as well as lysosomal PI(3,5)P₂ synthesis by PIKFYVE (de Lartigue et al., 2009) and hydrolysis by INPP5E (Hasegawa et al., 2016). After fusion, a transient PIP5K-mediated PI(4,5)P₂ increase on the lysosomal surface recruits clathrin machinery (Rong et al., 2012), which is followed by inositol polyphosphate-5-phosphatase (OCRL)-mediated PI(4,5)P₂ turnover to PI(4)P for autophagic lysosome reformation (De Leo et al., 2016).

Both PI(3)P and PI(3,5)P₂ facilitate degradative membrane-driven cellular processes. Thus, as described above, enzymes that catalyze their synthesis or turnover play critical roles in trafficking membrane-bound cargoes to the lysosome. Myotubularin-related phosphatases (MTMRs) are a conserved family of 3-PI phosphatases that are associated with diseases (Raess et al., 2017) and regulate membrane trafficking during endocytosis and autophagy (Nicot and Laporte, 2008; Robinson and Dixon, 2006). Of the 16 human MTMRs, 9 contain the catalytically active CX₅R motif, and 7 are catalytically dead pseudo-phosphatases due to a cysteine residue substitution. The MTMR family of enzymes are grouped into subfamilies, and the Drosophila genome harbors a single orthologue in each subfamily. One human MTMR subfamily consists of human MTMR6, MTMR7, and MTMR8. These all partner with MTMR9, an inactive pseudophosphatase that fails to bind PIs on its own (Zou et al., 2012b) but enhances substrate binding, catalytic activity, and protein stability of its active partners in vitro and in cells (Zou et al., 2009a, 2012b).

Here, we identify CG3530/dMtmr6 as a regulator of autophagy in Drosophila. We reveal that dMtmr6, which is homologous to the human MTMR6 subfamily of myotubularin-related 3-phosphatases, functions as an essential regulator of autophagic flux in multiple cell types and is required for Drosophila development. Down-regulation of dMtmr6 leads to autophagic vesicle accumulation, promotes lysosome biogenesis, and impairs both endocytosis and phagocytosis. Importantly, we show that MTMR8 has similar function in autophagy in higher animal cells. This work establishes dMtmr6 as an essential regulator of autophagic flux.

Enzymes that regulate autophagy represent potential drug targets for human disorders (Rubinsztein et al., 2012), and cycles of phosphorylation and dephosphorylation provide regulatory cues for autophagy and other cellular processes. However, the phosphatases that regulate autophagy remain largely unexplored. We sought to identify novel phosphatases that regulate autophagy by screening the Drosophila PI phosphatases because of their links to membrane trafficking and autophagy. We used the flippase/flippase-recognition target (FLP/FRT) system (Golic and Lindquist, 1989; Theodosiou and Xu, 1998) to simultaneously express GFP and RNAi that target phosphatases in clones of cells within tissues that uniformly express the autophagy reporter mCherry-Atg8a.

Drosophila larval intestine enterocytes contain pmCherry-Atg8a puncta 2 h after puparium formation (APF) due to high levels of developmental autophagy (Denton et al., 2009). Therefore, we evaluated the effects of PI phosphatase RNAi on mCherry-Atg8a expression in 2-h-APF midguts (Table S1). We identified multiple previously described regulators of autophagy, such as PTEN (not shown), as well as CG3530/Mtmr6 (dMtmr6 hereafter), a previously undescribed regulator of autophagy. Expression of double-stranded RNA targeting dMtmr6 generates cells that contain Atg8a puncta in feeding third-instar larvae (L3) when autophagy remains inactive in neighboring WT cells (Fig. 1, A and B). At 2 h APF, dMtmr6-knockdown enterocytes contain both increased and enlarged mCherry-Atg8a puncta compared with neighboring WT cells (Fig. 1, B–D).

We sought to determine if dMtmr6 regulates autophagy either specifically in larval intestine developmental autophagy or more broadly in multiple cell contexts, including stress-induced autophagy following nutrient deprivation. Therefore, we investigated if dMtmr6 functions in amino acid starvation–induced autophagy in the Drosophila larval fat body (Scott et al., 2004). We examined fat body tissue from either fed L3 that lack autophagy or animals grown on 20% sucrose under nutrient-restriction conditions for 4 h to induce autophagy. As in midguts of the intestine, dMtmr6-knockdown fat body cells contain Atg8a puncta in fed larvae, and Atg8a puncta are significantly larger in dMtmr6-knockdown cells than in adjacent WT cells from starved larvae (Fig. 1, E–H). These data indicate that dMtmr6 functions in both developmental and starvation-induced autophagy in Drosophila.

To better understand the function of dMtmr6, we deleted the dMtmr6 open reading frame using CRISPR-Cas9 (Fig. 2 A). Knockdown of dMtmr6 in all tissues using both strong and weak RNAi strains results in lethality at either the early larval stage or early pupal stage, respectively (Fig. 2 B). Importantly, homozygous loss of dMtmr6 (dMtmr6Δ/Δ) is larval lethal, as is dMtmr6Δ/Df (Fig. 2 B). However, some trans-heterozygous mutant animals (dMtmr6Δ/dMtmr6^KG01267) survive until adult stage. dMtmr6^KG01267 is a mutant allele made by P-element insertion in the 5′ UTR of the dMtmr6 B and C isoforms, but not A (Fig. 2 A), which suggests that the P-element insertion is a weak allele. To evaluate the effects of dMtmr6 loss on autophagy, we measured protein levels from whole larvae (feeding L3). Trans-heterozygous dMtmr6Δ/KG and dMtmr6Δ/Df animals have increased levels of the cargo receptor ref(2)p (p62 in mammals) and lipidated Atg8a compared with control heterozygous animals lacking one allele of either the null dMtmr6Δ or weak allele dMtmr6 ^KG01267 (Fig. 2, C–C″). We then used the FLP/FRT system to generate homozygous dMtmr6-null mutant cell clones within WT tissue. As with dMtmr6-RNAi, dMtmr6 knockout results in increased amounts of Atg8a in feeding larval midguts (Fig. 2, D and D′). However, we obtain significantly fewer dMtmr6Δ-null mutant cell clones than control cell clones in both the midgut and fat body (Fig. 2 E), suggesting that homozygous loss of dMtmr6Δ compromises cell viability. These data suggest that dMtmr6 regulates autophagy and is critical for Drosophila development and that at least in some contexts, dMtmr6 is required for cell viability.

Our data indicate that dMtmr6 influences autophagic flux, but it is unclear at what stage flux is altered because increased Atg8a puncta could result from elevated levels of autophagic membrane biogenesis, impaired degradation, or both. We analyzed Drosophila fat body tissue stained with antibody to detect the autophagy cargo adapter ref(2)p to determine if lysosomal degradation of autophagic cargoes is impaired when we down-regulate dMtmr6. Under normal conditions, ref(2)p is internalized within the autophagosome and degraded by the autolysosome. In dMtmr6-knockdown fat body cells, ref(2)p levels increase compared with neighboring WT cells from fed and starved animals (Fig. 3, A–C). In addition, dMtmr6-knockdown cells contain larger ref(2)p spots than WT cells from starved animals (Fig. 3 D). These findings indicate that dMtmr6 influences the rate of autophagy at the stage of initiation, degradation, or both.

dMtmr6 is an MTMR protein that is most similar to the human MTMR6 subfamily, which includes MTMR6 (65% similar, 47% identical), MTMR7 (64% similar, 47% identical), and MTMR8 (72% similar, 55% identical; Fig. 3 E and,Fig. S1). The predicted catalytic domain of dMtmr6 and each of the human MTMR6-subfamily members is conserved (Fig. 3 E). However, a key difference between fly and human is that dMtmr6 encodes an N-terminal FYVE domain, while members of the human MTMR6-subfamily encode a coiled-coil domain (Fig. 3 F). Previous studies suggest a negative role for human MTMR8 in autophagy and for MTMR6 in apoptosis (Zou et al., 2009a, 2012b). However, these studies focus on biochemical aspects of MTMR8, MTMR6, and their inactive binding partner, MTMR9, and provide very little information about how MTMR8 could function in autophagy. We transfected both COS-7 and HeLa cells with siRNA to down-regulate either MTMR6 or MTMR8 and confirmed a role for MTMR8 in autophagy in primate cells. As shown previously (Zou et al., 2009a), MTMR6 knockdown promotes HeLa cell death, but, unexpectedly, MTMR6 knockdown in COS-7 cells did not influence viability (data not shown). Like dMtmr6 knockdown, down-regulation of MTMR8 produces increased numbers of large LC3B puncta in COS-7 cells (Fig. 3, G–J). Recruitment of ubiquitinated cargoes to autophagic membranes requires the autophagy adapter molecule p62 (Drosophila ref(2)p). Importantly, reduced MTMR8 function resulted in increased numbers of large p62 puncta in COS-7 cells (Fig. 3, G, H, K, and L) and HeLa cells (not shown). We also observed increased lipidated-LC3B (LC3B-II) and p62 protein levels in MTMR8 siRNA–transfected COS-7 cells in both serum-fed and serum-deprived conditions (Fig. 3, M–M″), which suggests that MTMR8-depleted cells can respond to nutrient deprivation. Furthermore, the inhibition of lysosomal function by addition of Bafilomycin A1 caused a further increase in LC3B-II and p62 levels (Fig. 3, M–M″), which supports our hypothesis that MTMR8 inactivation impairs autophagic flux. Similar to what we observed in Drosophila, we hypothesize that MTMR8 may also regulate the rate of either autophagosome biogenesis or lysosomal degradation of cargoes in higher organisms.

Genes whose molecular products facilitate autophagic flux are expressed through transcription factor EB (TFEB; Drosophila Mitf) mediated transcription. When inactive TFEB remains in the cytoplasm, but in response to starvation or during lysosome dysfunction, TFEB translocates to the nucleus to promote expression of both lysosome and autophagy genes (Napolitano and Ballabio, 2016). We hypothesized that if MTMR8 inactivation impairs lysosome function, then we should see evidence of TFEB activation in MTMR8-depleted cells. Indeed, we detected more TFEB in the nucleus of serum-starved than serum-fed control HeLa cells, while MTMR8-depleted cells displayed nuclear TFEB in both serum-starved and serum-fed conditions (Fig. S2, A–E). If MTMR8 and dMtmr6 function as orthologues, then dMtmr6 knockdown should produce activation of the Drosophila TFEB orthologue, Mitf. As expected, known gene targets of Mitf increased in Drosophila fat body dMtmr6-knockdown compared with control (Lucif-RNAi; Fig. S2 F). Despite nutrient availability, TFEB activation occurs in cells and tissues depleted of MTMR8 or dMtmr6. These results suggest that inactivation of either MTMR8 or dMtmr6 may elicit lysosome dysfunction.

We further explored the possibility that lysosome dysfunction exists in the absence of either dMtmr6 or MTMR8. We reasoned that fusion between autophagosomes and lysosomes could be impaired, which would indicate an autolysosome formation defect and could explain increased ref(2)p/p62 and lipidated Atg8a/LC3B levels. To determine if autolysosomes form, we used the tandem GFP-mCherry-Atg8a fluorescent reporter, which is a transgene expressed under control of the Atg8a promoter (Lee et al., 2016). Dual-punctate fluorescence of both GFP and mCherry reflects autophagosomes, but because GFP is pH sensitive, mCherry signal alone reflects autolysosomes. In Drosophila fat body from fed animals, Atg8a reporter expression is diffuse throughout the cytoplasm in control tissue, and very few puncta form autophagosomes or autolysosomes (Fig. 4 A). By contrast, dMtmr6-depleted tissue contains numerous autophagosomes and autolysosomes (Fig. 4, B and C). These data indicate that the number of autolysosomes increase when dMtmr6 function is reduced.

Reduced function of both dMtmr6 in Drosophila and MTMR8 in COS-7 cells results in large Atg8a/LC3B and ref(2)p/p62 structures. To further determine the nature of the large Atg8a-positive puncta generated by dMtmr6 depletion, we performed transmission electron microscopy (TEM) analyses of Drosophila midguts 2 h APF. In control animals undergoing developmental autophagy in midgut enterocytes, the cytoplasm contains autophagosomes and autolysosomes (Fig. 4 D). By contrast, dMtmr6-knockdown enterocytes possess large electron-poor vesicles throughout the cytoplasm (Fig. 4 E). We classified these single-membrane structures as autolysosomes (Fig. 4, F and G), but it is possible that these structures may be amphisomes that form by fusion of autophagosomes with late endosomes. Importantly, we observe similar large autolysosome structures in COS-7 cells transfected with MTMR8 siRNA, but not in control cells (Fig. 4, H–K).

To confirm that the large structures we observe by TEM are autolysosomes and that they occur in another mammalian cell line, we costained HeLa cells with LC3B and LAMP1 antibodies to detect coincidence of autophagic and lysosomal markers, respectively. In serum-fed HeLa cells transfected with control siRNA, we observe very few LC3B-positive spots, and LAMP1 puncta appear throughout the cytoplasm (Fig. 4 L). By contrast, MTMR8 siRNA–transfected HeLa cells contain large puncta that feature both LC3B and LAMP1 that colocalize and cluster near the perinuclear region (Fig. 4, M–R). These data indicate that depletion of both dMtmr6 in fly tissues and MTMR8 in primate cells promotes the formation of large autolysosomes.

Autophagosomes deliver their cargoes to lysosomes for degradation, and impaired lysosome function can lead to autophagosome accumulation and prevent autolysosome resolution. Therefore, we evaluated lysosomes in larval fat body using LysoTracker, an acidophilic dye that labels acidic organelles, to determine if dMtmr6 influences lysosome homeostasis. In fat body from fed larvae, dMtmr6-knockdown cells contain more acidic organelles than neighboring WT control cells (Fig. 5, A and B). Although dMtmr6-knockdown cells possess increased numbers of acidic vesicles, this may not reflect an increase in functional lysosome numbers. Therefore, we evaluated cathepsin protease activity in dMtmr6-knockdown cells to directly evaluate lysosomal enzyme function. We tested cathepsin B protease activity using Magic red substrate, which fluoresces upon cleavage by cathepsin B. Compared with neighboring WT cells in fat body from fed larvae, Magic red fluorescence levels increase in dMtmr6-knockdown cells (Fig. 5, C and D). We also measured cathepsin L protein levels from fed or starved larval fat body tissues. During starvation-induced autophagy, lysosome biogenesis increases and pre- and procathepsin L levels rise in Luciferase-RNAi and Atg1-RNAi control tissues (Fig. 5 E). However, in dMtmr6-depleted tissues from fed animals, pre- and procathepsin L protein levels are greater than in controls (Fig. 5 E), indicating that cathepsin L biogenesis increases as a result of dMtmr6 knockdown. Together these data suggest that, although lysosomes display functional qualities in dMtmr6, lysosome homeostasis is altered.

Because lysosomal homeostasis is altered in dMtmr6 knockdown cells, we reasoned that aspects of the endolysosomal pathway may be altered. To test whether dMtmr6 functions in endocytosis, we monitored fluid-phase endocytosis in larval fat body via uptake of Texas red (TR)-avidin. Control cells contained TR-avidin–positive puncta throughout the cytosol, whereas dMtmr6-knockdown cells (GFP positive) possessed little to no endocytic tracer (Fig. 6, A and B), suggesting that dMtmr6 depletion impairs early endocytic uptake. However, we observed no change in localization of the endosomal marker Rab7 (Fig. S2, G–I). Thus, dMtmr6 knockdown alters endocytosis in a manner that does not appear to impact Rab7 localization.

Intracellular localization of FYVE domain–containing proteins represents PI(3)P-enriched sites (Gillooly et al., 2000). However, Drosophila PI(3)P reporters remain limited to endosomal compartments, and we lack tools that are specific to autophagic PI(3)P. Nonetheless, we investigated how dMtmr6 depletion effects the widely used GFP-2xFYVE endosomal PI(3)P reporter, which utilizes two copies of the FYVE domain from the early endosomal protein Hrs fused to EGFP (Wucherpfennig et al., 2003). When we misexpressed GFP-2xFYVE and depleted dMtmr6 in fat body of fed larvae, we detected increased GFP-2xFYVE levels in cellular regions that are both perinuclear and away from the nucleus (aponuclear) when compared with Lucif-RNAi controls (Fig. S3, A–F). Interestingly, fat body from fed dMtmr6-knockdown animals contained perinuclear GFP-2xFYVE distribution similar to fat body from starved control animals, even though aponuclear levels remained high in dMtmr6-RNAi conditions (Fig. S3, B–G). Starvation further increased perinuclear and decreased aponuclear GFP-2xFYVE localization in dMtmr6-depleted cells (Fig. S3, C–H). Despite alterations in GFP-2xFYVE localization, we saw reduced colocalization between GFP-2xFYVE and Atg8a in dMtmr6-knockdown cells compared with controls in both fed and starved conditions (Fig. S3, I–M). Together, these data indicate that dMtmr6 depletion impacts localization of PI(3)P-containing vesicles, but these vesicles remain distinct from large Atg8a-bearing structures.

Phagocytosis requires both Vps34-mediated PI(3)P generation and lysosomal degradation of extracellular cargoes (Jeschke and Haas, 2016). Given the role of dMtmr6 in autophagy and endocytosis, we hypothesized that dMtmr6 might also influence phagocytosis. To investigate this, we tested the ability of Drosophila dMtmr6-knockdown embryonic macrophages to engulf apoptotic corpses, which are generated throughout the embryo as part of normal development. Corpse phagocytosis primes macrophages for subsequent immune responses such as recruitment to wounds and bacterial uptake (Weavers et al., 2016). Importantly, dMtmr6 knockdown in embryonic macrophages compromises phagocytosis, as indicated by the decreased vacuole number compared with control macrophages (Fig. 6, C and D). These data suggest a role for dMtmr6 in phagocytosis.

We next investigated the relationship between dMtmr6 and upstream regulators of autophagy. Cell growth and the factors that regulate this process, such as the ribosomal protein kinase p70 S6 and the serine/threonine kinase Akt (protein kinase B), often have a reciprocal relationship with autophagy. Since our data indicate that cells with decreased dMtmr6 function are prone to autophagy, we tested if growth signaling is altered in these cells. Indeed, fat body from feeding larvae possess reduced levels of both phosphorylated p70 S6 Kinase and phosphorylated AKT in dMtmr6-depleted tissue compared with controls (Fig. 7 A). These data suggest that fat body exists in an autophagy-prone state in the absence of dMtmr6.

We queried the relationship between dMtmr6 and the Atg1 and Vps34 kinases that are conserved regulators of autophagy (Galluzzi et al., 2017). As expected, Atg1 knockdown alone resulted in ref(2)p accumulation (Fig. 7 B) and inhibited Atg8a puncta formation (Fig. 7, F and D). Surprisingly, fat body with double knockdown of Atg1 and dMtmr6 possessed Atg8a lipidation, ref(2)p accumulation (Fig. 7 B), and increased numbers of Atg8a-positive structures (Fig. 7, C and D), but these Atg8a puncta were significantly smaller (Fig. 7, C–E). These data suggest that dMtmr6 functions in at least a partially Atg1-independent manner.

During autophagy, Vps34 kinase activity contributes to PI(3)P production on autophagic membranes, and Vps34 mutant cells fail to form GFP-Atg8a puncta (Juhász et al., 2008). Because dMtmr6 is a putative PI(3)P phosphatase, we reasoned that dMtmr6 could influence the pool of PI(3)P that is supplied by Vps34. We used the FLP/FRT system to generate Vps34-null mutant cell clones in fat body expressing dMtmr6-RNAi (Lsp2-GAL4 > UAS-dMtmr6-RNAi) and evaluated Atg8a puncta formation in starved animals. Vps34 null mutant cells that express dMtmr6-RNAi possess fewer Atg8a puncta than cells expressing dMtmr6-RNAi alone (Fig. S4, A and B). In addition, we observed similar results when we tested if the loss of Atg9 could suppress Atg8a puncta formation in dMtmr6-knockdown tissue (Fig. S4, C and D). Taken together, these data suggest that dMtmr6 negatively regulates Atg8a membrane association and that dMtmr6 function is at least partially independent of either Vps34, Atg9, or Atg1 function.

Despite documentation that MTMR8 hydrolyzes PI(3)P and PI(3,5)P₂ in vitro (Zou et al., 2012a, 2012b), we lack evidence that the MTMR8 catalytic domain mediates autophagy. In fact, MTMR8 could influence autophagy by various membrane interactions, including PI(3)P or PI(3,5)P₂ hydrolysis (CX₅R motif) and membrane localization (PH-GRAM domain), which are required for other MTMR family members (Amoasii et al., 2012). To test these possibilities, we modified MTMR8 in HeLa cells using CRISPR-Cas9 to generate a cysteine to serine mutation (MTMR8-C338S), which is the amino acid change seen in the catalytically dead MTMR family members, and a PH-GRAM domain truncation (MTMR8ΔPH; Fig. 8 A; Lorenzo et al., 2006). Similar to the LAMP1 and LC3B puncta alterations in MTMR8-knockdown cells (Fig. 4, M–S), LAMP1 and LC3B vesicles were larger in serum-fed MTMR8-C338S cells than in MTMR8-WT cells (Fig. 8, B–E). Furthermore, fewer LAMP1 puncta, more LC3B puncta, and increased LAMP1/LC3B colocalization exist in MTMR8-C338S cells compared with MTMR8-WT control cells (Fig. 8, F–H). Although not significant, MTMR8-C338S influences autophagic flux in serum-fed conditions compared with control cells (Fig. S5). Furthermore, LysoTracker staining appears brighter in MTMR8-C338S cells compared with MTMR8-WT cells, but puncta number and size remains unchanged (Fig. S5, B–E). These data suggest that MTMR8 catalytic activity influences autophagic flux.

We next investigated the requirement of the MTMR8PH-GRAM domain for autophagy. In striking comparison to both MTMR8-WT and MTMR8C338S cells, MTMR8ΔPH cells contain enlarged LAMP1 and LC3 puncta, reduced LAMP1 and increased LC3B puncta numbers, and significant colocalization between these two autophagic markers (Fig. 8, B–I). Moreover, MTMR8ΔPH cells have more LysoTracker puncta than both MTMR8-WT and MTMR8-C338S cells, and LysoTracker staining significantly overlaps with LAMP1 (Fig. S5, D–G). Thus, MTMR8 catalytic function influences autophagy, but MTMR8 membrane localization via the PH-GRAM domain appears to play a more critical role in autolysosome homeostasis.

Previous data indicate that MTMR8 favors PI(3)P, but can hydrolyze both PI(3)P and PI(3,5)P2 (Zou et al., 2012b). To investigate a role for MTMR8 in PI(3)P-regulation at autophagic membranes, we evaluated localization of WIPI2, a protein that binds early autophagic structures at PI(3)P-enriched sites (Dooley et al., 2014). We found a significant increase in WIPI2 puncta numbers in both MTMR8C338S and MTMR8ΔPH cells compared with MTMR8-WT control HeLa cells (Fig. 8, J–O), which suggests that MTMR8 regulates PI(3)P levels at autophagic membranes. Furthermore, MTMR8C338S cells contain significantly more WIPI2 puncta than MTMR8ΔPH cells (Fig. 8, K–M and O), indicating that the catalytic cysteine residue is important for decreasing PI(3)P levels, while the PH domain is likely important for MTMR8 recruitment to PI(3)P-bearing membranes. Moreover, WIPI2 and LC3B colocalize more in MTMR8C338S cells and less in MTMR8ΔPH cells (Fig. 8, K, N, and O). We expected that impaired PI(3)P hydrolysis by MTMR8C338S could lead to increased WIPI2 recruitment, more LC3B conjugation, and increased WIPI2/LC3B colocalization. However, it is surprising that WIPI2/LC3B colocalization decreases in MTMR8ΔPH cells compared with MTMR8-WT control cells when WIPI2 puncta numbers increase, suggesting a possible role for the MTMR8 PH domain in WIPI2/LC3B localization.

Previous work indicates that MTMR8 localizes to the perinuclear region (Lorenzo et al., 2006). We surmised that MTMR8 might localize to a perinuclear lysosomal pool to facilitate autolysosome resolution. We initially used CRISPR-Cas9 to tag the MTMR8 C-terminus with 3xFLAG, but we were unable to detect FLAG by either immunofluorescence or immunoblotting. Therefore, we overexpressed MTMR8. Unlike Lorenzo et al., we saw peripheral localization, and MTMR8 failed to overlap with perinuclear-localized LC3B puncta (Fig. S5 H). Consistent with previous work (Zou et al., 2012b) and our evidence that MTMR8 influences autophagic flux (Fig. 3), p62 protein levels decreased and lipidated LC3B levels increased in cells overexpressing MTMR8 (Fig. S5 I). While MTMR8 mutant and overexpression differentially influence p62 levels, LC3B-II levels increase under both conditions. These data suggest that MTMR8 overexpression activates autophagy and promotes degradation, while MTMR8 inactivation impairs degradation, which activates autophagy through what is likely a TFEB-mediated feedback loop. Taken together, these data suggest that MTMR8 influences autophagic flux by promoting lysosome-mediated degradation.

We identified Drosophila CG3530/Mtmr6 as a positive regulator of autolysosome homeostasis and show that MTMR8 has a similar function in both human and monkey cells. dMtmr6 maintains autophagic flux in multiple fly tissues under different conditions, including stress- and developmentally induced autophagy. We demonstrated that dMtmr6 and MTMR8 play a critical role in regulating the amount and size of cellular Atg8a/LC3B-positive vesicles. Furthermore, these structures correlated with lysosome markers and demarcated the existence of autolysosomes. Depletion of dMtmr6 in fly tissues or MTMR8 in mammalian cells influences endolysosomal homeostasis and promotes autolysosome accumulation.

MTMRs dictate intracellular membrane identity and modulate membrane trafficking events, including autophagy, through PI(3)P and PI(3,5)P₂ degradation. However, differential expression of MTMRs implies that they may have cell- and tissue-specific functions (Raess et al., 2017). According to bioinformatics studies from public data at the Genotype-Tissue Expression database, the roles and requirements for the different MTMRs are thought to differ according to tissue and cell type. For example, MTMR8 is expressed at higher levels in reproductive tissues than in brain tissues, which instead, express MTMR7. Meanwhile, MTMR6 is most highly expressed in the adrenal gland and testis. We tested HEK293T cells (kidney) in our analyses (not shown) and found that MTMR8 knockdown had a less significant effect on autophagic structures than in HeLa cells (cervical). These data are consistent with the data of Raess et al. (2017), which show lower MTMR8 expression in kidney and higher MTMR8 expression in cervical tissues. Like MTMR8, publicly available RNA sequencing of Drosophila tissues indicates that dMtmr6 is highly expressed in reproductive tissues, and both MTMR8 and dMtmr6 are moderately expressed in tissues of the digestive tract (Thurmond et al., 2019). Unlike MTMR8, however, dMtmr6 is highly expressed in head of adult flies, hinting at a role for dMtmr6 in the brain and potentially the central nervous system. Interestingly, dMtmr6 muscle-specific knockdown impairs Drosophila flight (Schnorrer et al., 2010).

Tissue and cell specificity at least partially account for the diverse regulatory functions of the MTMR6 subfamily members and their binding partner, MTMR9. Human MTMR6 facilitates PI(3)P turnover during micropinocytosis (Maekawa et al., 2014), negatively regulates the Ca²⁺-activated K⁺ channel KCa3.1 in CD4 T cells (Srivastava et al., 2005, 2009), and apoptosis in HeLa cells (Zou et al., 2009a, 2012a). Similar to dMtmr6, which regulates AKT signaling in Drosophila fat body (Fig. 7 A), both MTMR7 and MTMR9 regulate CD4 T cell AKT signaling and differentiation (Guo et al., 2013), and MTMR7 regulates AKT signaling in C2C12 myoblasts. MTMR7 protein levels inversely correlate with tumor grade in colorectal cancer (CRC), and in CRC cell lines, MTMR7 inhibits proliferation and insulin-mediated AKT-ERK1/2 signaling (Weidner et al., 2016). In zebrafish embryos, MTMR8 regulates actin filament modeling (Mei et al., 2009) and both muscle (Mei et al., 2009) and vascular development (Mei et al., 2010). It is important to note that the mouse genome lacks MTMR8 and contains only MTMR6 and MTMR7. In human cells, the MTMR8–MTMR9 complex reduces PI(3)P levels and inhibits autophagy (Zou et al., 2012b). MTMR9 mutations are associated with obesity (Yanagiya et al., 2007), metabolic syndrome (Hotta et al., 2011), and multiple prediabetes-associated symptoms (Tang et al., 2014). While a specific link to MTMR8 in disease remains to be determined, alterations in the MTMR8 locus have been implicated in glioblastoma multiforme (Waugh, 2016) and intellectual disability (Holman et al., 2013). Furthermore, a rare type of X-linked spastic paraplegia (SPG16; OMIM accession no. 300266) is mapped to an uncharacterized locus at Xq11.2, which is where MTMR8 resides. Future studies could benefit from mining genome- and exome-sequencing databases for the influence of this MTMR family on disease.

An additional difference exists between MTMR8 and dMtmr6. MTMR8 encodes one human isoform while dMtmr6 encodes three (Fig. 2 A). We have not experimentally tested the significance of dMtmr6 splice isoforms. However, publicly available data of fly gene expression during development indicate that there is ubiquitous expression of dMtmr6 B and C isoforms throughout development (Thurmond et al., 2019). The dMtmr6 A isoform is expressed at low levels during early embryonic development, ceases during late embryonic development, and increases progressively throughout the L3 stage. In adult flies, isoform A expression is lower in females than in males. Potentially relevant to our findings, wandering third-instar fat body expresses both B and C, but not A. However, upon pupa formation, when numerous tissue types undergo developmental autophagy, all three isoforms are expressed. Thus, dMtmr6 isoform A could play a specific role in developmental autophagy, while isoforms B and C function more generally. Our data indicate that dMtmr6 is required for survival (Fig. 2 B), yet some trans-heterozygous animals that are dMtmr6^KG01267/Df(2R) survive to adults. The dMtmr6^KG01267 allele retains isoform A, which indicates that, in the absence of isoforms B and C, isoform A could function in a compensatory way to support survival. Future experiments should explore the possibility of isoform-specific dMtmr6 functions in development and cellular trafficking.

Despite evidence of autophagy initiation, dMtmr6 and MTMR8 depletion elicits ref(2)p/p62 accumulation, indicative of impaired autophagic cargo degradation (Fig. 2 C; and Fig. 3, A–D and G–M). However, autophagosome–lysosome fusion remains intact based on GFP signal reduction in dMtmr6 fat body expressing the tandem-fluorescent Atg8a reporter (Fig. 4, A–C). This reporter recapitulated the aberrant large autophagic puncta we saw with mCherry-Atg8a alone in dMtmr6-depleted fat body from fed animals and supplied evidence that upon fusion, lysosomal pH remains low enough to quench the GFP signal from autophagic structures. Furthermore, dMtmr6-depleted cells possess elevated levels of LysoTracker and cathepsin B protease activity compared with adjacent WT cells (Fig. 5, A–D). These data imply that in dMtmr6-deficient cells, lysosomes harbor some functional characteristics that are required for degradation. Similarly, MTMR8-deficient and mutant cells possess numerous and large vesicles that are positive for LC3B, LAMP1, and LysoTracker, indicative of acidified autolysosomes (Fig. 4, L–R; Fig. 8, B–I; and Fig. S4, B–G). Thus, in dMtmr6- and MTMR8-depleted cells, autophagosomes form and fuse with lysosomes, which suggests that they promote postfusion endolysosomal homeostasis. Despite aberrant localization, the endosomal marker GFP-FYVE failed to colocalize with large Atg8a-positve vesicles in dMtmr6-knockdown tissue (Fig. S3). This suggests that although autophagic vesicles lack endosomal markers in dMtmr6-depleted cells, inactivation of dMtmr6 may interfere with endocytic trafficking via a direct or indirect role at endocytic vesicles or degradative lysosomes, respectively. Taken together, our data indicate that dMtmr6/MTMR8 may regulate autophagy at the autolysosome. Future studies should investigate if dMtmr6/MTMR8 mediates autolysosome reformation.

We also evaluated Drosophila cathepsin L protein levels (Fig. 5 E). Lysosomal cathepsins are synthesized as inactive preproenzymes that undergo subsequent maturation steps (Stoka et al., 2016). Following preprocathepsin synthesis in the ER, the N-terminal signal peptide is cleaved and glycosylated to generate procathepsin. Glycosylation ensures that cathepsins remain inactive until low pH triggers propeptide unfolding and either successive autocleavage or cleavage by other active cathepsins. We showed that cathepsin L undergoes maturation in dMtmr6-depleted Drosophila fat body from fed and starved animals, but the balance between preprocathepsin, procathepsin, and mature cathepsin L differs compared with WT fat body under the same conditions (Fig. 5 E). Indeed, high levels of prepro- and procathepsin L exist in dMtmr6-knockdown tissue. For this, one possible explanation is that dMtmr6 deficiency influences cathepsin L processing. However, we can detect mature cathepsin L, and an alternative possibility is that dMtmr6 depletion prompts lysosome biogenesis. This idea is further supported by our evidence of augmented gene expression of known targets of Mitf (TFEB) in dMtmr6-depleted tissues (Fig. S2 F), denoting activation of autophagy and lysosome biogenesis (Sardiello et al., 2009). We note that dMtmr6-knockdown elevates ref(2)p transcript levels, which could also compound ref(2)p protein levels. It appears that MTMR8-depleted cells undergo a similar response because TFEB localized to the nucleus of HeLa cells under both serum-fed and serum-starved conditions.

Phosphorylated p70 S6K and Akt levels are reduced in Drosophila dMtmr6-knockdown fat body from fed animals compared with control animals (Fig. 7 A), which likely explains the influence of bMtmr6/MTMR8 on Mitf/TFEB. Indeed, both phospho-p70 S6K and phospho-Akt are indicators of mTORC1 activity and phosphorylation is indicative of ample nutrients and growth factors (Liu and Sabatini, 2020). When active, mTORC1 localizes to lysosomes and inhibits TFEB nuclear localization (Settembre et al., 2012), and Akt represses TFEB nuclear localization independent of mTORC1 (Palmieri et al., 2017). Conversely, starvation and lysosome stress prompts mTORC1 dissociation from lysosomes, TFEB nuclear translocation, and autophagy induction (Settembre et al., 2012). During prolonged starvation, cells require mTORC1 reactivation to replenish the lysosome pool (Yu et al., 2010). Furthermore, active mTORC1 is required for lysosomal tubulations, which is a critical step in autolysosome reformation (Yu et al., 2010). Once reactivated, mTORC1 functions upstream of PIK3C3 produce a small pool of autolysosomal PI(3)P, and loss of PIK3C3 activity, despite active mTORC1, prevents tubule scission. Our genetic epistasis analyses indicated that dMtmr6 functions at least in a partially class III PI3K Vps34–independent manner (Fig. S3 A and B). As mentioned above, our data suggest that dMtmr6 functions at the autolysosome to facilitate degradation. In Drosophila blood cells, the class II PI3K, Pi3K68D coregulates endolysosomal size and cortical dynamics with mtm, an MTMR family member, via a pool of PI(3)P (Velichkova et al., 2010). However, endolysosomal size in blood cells is also Atg1 and Vps34-dependent, unlike dMtmr6. The results of our phosphatase screen indicate that mtm does not play a role in fat body or developmental midgut autophagy (Table S1). Moreover, we do not think that dMtmr6 plays a role in cortical remodeling of blood cells. We were able to observe actin filaments (as seen by LifeAct localized accumulation) at extending protrusions in both control and dMtmr6-depleted blood cells (Fig. 6 C), which strongly suggests that dMtmr6 does not play a direct role in actin filament turnover.

We showed that dMtmr6-depleted macrophages contained fewer phagocytic vacuoles than control blood cells (Fig. 6 C). Phagocytosis requires Vps34 to produce PI(3)P and, like autophagy, terminates as lysosomal hydrolases degrade cargoes (Jeschke and Haas, 2016). During Drosophila embryonic development, apoptotic corpses are removed by macrophages, a process that primes these blood cells for ensuing immune responses later in life (Weavers et al., 2016). Our findings imply that dMtmr6 supports macrophage corpse uptake. Similarly, we saw impaired uptake of TR-avidin by fluid-phase endocytosis in dMtmr6-depleted fat body cells (Fig. 6 A). In the nematode Caenorhabditis elegans, MTM-1, MTM-6, and MTM-9 facilitate fluid-phase endocytosis in coelomocytes (Dang et al., 2004; Xue et al., 2003), and MTM-1 negatively regulates apoptotic corpse engulfment (Neukomm et al., 2011; Zou et al., 2009b). Thus, MTMRs have an evolutionarily conserved role in these diverse but related trafficking processes.

Here, we demonstrate that dMtmr6 and MTMR8 positively regulate late-stage autophagy and play a critical role in regulating autophagic flux. Our analyses revealed critical requirements for dMtmr6 and MTMR8 in preventing expansion of the Atg8a/LC3B compartment, leading to autolysosome accumulation, and impairing autophagic flux in multiple cell and tissue types. Defects in autophagic flux underlie lysosomal storage disorders and neurodegenerative diseases, including Parkinson’s and Huntington’s. Thus, it remains important to identify novel drug targets to modulate autophagy levels. Future studies of MTMR6 family members will reveal if they are useful targets to mitigate human diseases.

Drosophila were raised on standard cornmeal-molasses-agar medium. We performed fly crosses and experiments at 25°C unless noted otherwise. We used UAS-Luciferase as the WT control for RNAi experiments. We used CRISPR-Cas9 to delete the dMtmr6 open reading frame (Well Genetics). To generate RNAi-expressing cell clones, we crossed yw hsFlp; pmCherry-Atg8a; Act>CD2> Gal4, UAS-GFP (nls) virgin females to UAS-RNAi males. To generate tissue-specific knockdown, we crossed Np1-Gal4 virgins for intestinal knockdown or CG-Gal4 virgins for fat body knockdown to UAS-RNAi males. For dMtmr6 mutant cell clones we crossed yw hsFlp; FRT42D, ubi-GFP virgin females to FRT42D, dMtmr6Δ males. For tandem-fluorescent Atg8a analyses, we used the EGFP-mCherry-Atg8a that is expressed ubiquitously under the control of the Atg8a promoter (Lee et al., 2016). For Vps34 and Atg9 mutant clones, we crossed yw hsFlp; FRT42D, ubi-RFP; Lsp2-Gal4 to either yw hsFlp; FRT42D, Vps34^Δm22/cyo; dMtmr6-RNAi^BL38340 or yw hsFLP; FRT42D, Atg9^d51/cyo; dMtmr6-RNAi^BL38340. The following fly stocks used were from the Bloomington Drosophila Stock Center (BDSC): UAS-CG3530^HMS01807, UAS-Luciferase³⁵⁷⁸⁸, UAS-GFP-myc-2xFYVE⁴²⁷¹², and CG-Gal4⁷⁰¹¹. The sequences for BDSC knockdown strains are available via the Drosophila RNAi Screening Center and Transgenic RNAi Project (DRSC/TRiP) at https://fgr.hms.harvard.edu. We used RNAi lines from the Vienna Drosophila RNAi Center (VDRC; UAS-Atg1^GD16133, UAS-CG3530^GD26216). The sequences for VDRC knockdown strains are available at http://stockcenter.vdrc.at/control/main. We thank the following researchers for mutant flies: Gabor Juhász, Eotvos Lorand University, Budapest, Hungary for Vps34^Δm22 (Juhász et al., 2008) and Guang-Chao Chen for Atg9^d51 (Wen et al., 2017).

We induced RNAi clones in midgut or fat body cells by crossing virgin females of y w hsFlp; pmCherry-Atg8a; Act >CD2 >GAL4 (> is FRT site), UAS-nlsGFP/TM6B (reference) or y w hsFlp; +; Act >CD2 >GAL4, UAS-nlsGFP/TM6B to indicated transgenic RNAi lines. Following 1-d egg lays, we heat-shocked embryos at 37°C for 10 min. White prepupae were placed on wet filter paper for 2 h before dissection.

We transferred feeding L3 to 20% sucrose (starved) for 4 h or kept larvae in food (fed).

We dissected midguts or fat body in PBS (Gibco), fixed tissues in 4% formaldehyde in PBS 0.1% Triton X-100 (PBS-TX) overnight at 4°C, blocked with 1% BSA and 0.5% cold fish skin gelatin, and incubated overnight at 4°C with primary antibodies in PBS-TX. For immunolabeling, we used rabbit anti-GABARAP (1:100; Cell Signaling Technologies; 13733), chicken anti-GFP (1:500; Abcam; ab13970), and rabbit anti-ref(2)p (1:100; Abcam; ab178440). We used Hoechst dye to stain DNA and the following secondary antibodies (1:250): anti-rabbit Alexa Fluor 546 (Thermo Fisher Scientific; A-11035), anti-mouse 546 (Invitrogen; A-11003), or anti-rat Alexa Fluor 488 or 546. We mounted samples in Vectashield (Vector Labs). We imaged samples with a Zeiss LSM 700 confocal microscope equipped with a Plan-Apochromat 63×/1.40 Oil differential interference contrast (DIC) M27 objective and EC Plan-Neofluar 40×/1.30 Oil DIC M27 objective at room temperature (22°C) using Zeiss Zen Software. For LysoTracker red staining, we dissected tissues in PBS, transferred samples to LysoTracker red diluted in PBS (1:1,000) on ice for 5 min, mounted samples in 80% glycerol (in PBS), and imaged immediately at room temperature (22°C) with a Zeiss AxioImager Z1 microscope equipped with a Plan-Apochromat 20×/0.8 M27 objective and an Axiocam MR3 camera using Axiovision Imaging software. For Magic red staining, we dissected tissues in PBS, followed by a 30-min incubation in Magic red (diluted 1:10 in PBS; ImmunoChemistry; Magic red Cathepsin B Assay, 937) and a quick wash in PBS, and then imaged immediately. For mCherry-Atg8a and GFP-2xFYVE imaging, we fixed samples briefly with 4% formaldehyde in PBS-TX, mounted in Vectashield and imaged samples with a Zeiss AxioImager Z1 microscope (described above). Image processing was preformed using ImageJ (National Institutes of Health) by cropping, converting to RGB, generating maximum intensity projections, and adding scale bars. We calculated P values using a paired t test for adjacent cells, an unpaired t test for two separate genotypes under one condition, and an ordinary two-way ANOVA for comparison of multiple genotypes or conditions. Data are plotted as either min to max or the mean ± SD.

To visualize endocytosis, we dissected fat body from feeding L3, incubated it ex vivo with TR-avidin (Invitrogen; A820) diluted in Schneider’s media to a concentration of 80 µg/ml for 20 min, and then chased with 0.5% BSA in cold PBS for 10 min before overnight fixation with 4% formaldehyde. We washed the tissue three times (10 min per wash) with 0.1% Tween-20 in PBS and mounted in Vectashield (Vector Labs). We imaged samples with a Zeiss LSM 700 confocal microscope with 40×/1.30 Oil DIC M27 objective at room temperature (22°C) using Zeiss Zen Software and processed images using ImageJ .

We cultured COS7 (ATCC; CRL-1651) and HeLa cells (ATCC; CCL-2) in DMEM supplemented with 10% FBS and penicillin/streptomycin at 37°C and 5% CO₂. We applied experimental conditions and collected cells at a specific time point for either immunostaining or Western blotting. For immunostaining, we fixed cells in 4% PFA for 20 min, permeabilized in 10 µg/ml digitonin (Sigma; D141) for 15 min at room temperature, and blocked with 5% goat serum for 60 min at room temperature. We incubated cells overnight at 4°C in the following primary antibodies (diluted in 5% goat serum): rabbit anti-LC3B (1:1,000; Abcam; ab48394), mouse monoclonal anti-LAMP1 (1:1,000; BD Biosciences; 555798), mouse anti-Lamp2 (0.5 µg/ml; DSHB; H4B4), rabbit anti-p62 (1:2,000; MBL; PM045), rabbit anti-TFEB (1:1,000; Cell Signaling Technologies; 4240), and mouse anti-WIPI2 (1:500; Abcam; ab105459). After three washes with PBS, we incubated cells for 1 h at room temperature with Hoechst to stain nuclei and the following fluorescently labeled secondary antibodies (1:250): anti-rabbit Alexa Fluor 546 (Thermo Fisher Scientific; A-11035), anti-mouse 546 (Invitrogen; A-11003), or anti-rat Alexa Fluor 488 or 546. We mounted specimens with Vectashield, examined with a Zeiss LSM 700 confocal microscope with 63×/1.40 Oil DIC M27 objective at room temperature (22°C) using Zeiss Zen Software, and processed images using ImageJ by cropping, converting to RGB, generating maximum intensity projections, and adding scale bars. For serum starvation, we incubated cells in EBSS (Gibco). For Bafilomycin treatment, we incubated cells for 12 h in Bafilomycin A1 (20 µM; Sigma-Aldrich; B1793).

For knockdown experiments, we transfected cells with either control or siRNA oligos using Invitrogen Lipofectamine RNAiMAX (Thermo Fisher Scientific; 13778150) and cultured for 72 h before analyses. We purchased double-stranded siRNAs from GenePharma, and sequences are listed in Table S2. For MTMR8-overexpression experiments, we transfected cells with Lipofectamine 2000 Transfection Reagent (Thermo Fisher Scientific; 11668019), either as vehicle only or to deliver Myc-DDK-MTMR8 (OriGene; RC201912), and cultured for 24 h before analyses.

To generate MTMR8-C338S and MTMR8ΔPH HeLa cells, we delivered a CRISPR-Cas9 ribonucleoprotein complex containing a tracrRNA–crRNA complex, which is a two-part gRNA that consists of a CRISPR RNA (crRNA) that recognizes the DNA target sequence and a transactivating crRNA (tracrRNA) that functionally base-pairs with the crRNA to form a functional gRNA, along with Cas9 enzyme, and single-stranded donor oligonucleotides (ssODNs), using electroporation with the Neon transfection system (Thermo Fisher Scientific; MPK5000; Iyer et al., 2019). We acquired tracrRNA, crRNA, Cas9, and ssODNs from Integrated DNA Technologies and identified positive single-cell clones by PCR amplification and Sanger sequencing. The gRNA and ssODN sequences are as follows: MTMR8-C338S gRNA, 5′-GGT​CCC​ATC​CAT​CAG​AAC​AA-3′; MTMR8-C338S ssODN, 5′-TCT​TGT​CTT​CAC​AGG​CAG​TGA​AGG​TAG​AAA​AGG​CCA​GTG​TCT​TAG​TAC​ATA​GTT​CTG​ATG​GAT​GGG​ACC​GCA​CAG​CAC​AAG​TCT​GCT​CAG​TGG​CTA​GCA​T-3′; MTMR8ΔPH gRNA–5′, 5′-GGT​AGA​AAA​CGT​GAA​ATT​GG-3′; MTMR8ΔPH gRNA–3′, 5′-GCA​GTG​AAA​TAT​AAA​CCT​CA-3′; MTMR8ΔPH ssODN, 5′-GGT​AAC​TTA​TTC​AGT​ATT​TTT​TCC​TCC​AAT​AAC​AGG​TAG​AAA​ACG​TGA​AAA​GGT​TTA​TAT​TTC​ACT​GCT​CAA​GCT​TTC​TCA​GCC​AGG​TAG​TTA​TGA​TAG​T-3′.

We washed mammalian cells three times with PBS, added lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100) supplemented with protease inhibitor cocktail (Thermo Fisher Scientific; 78425), and incubated for 30 min on ice. We centrifuged homogenates at 13,000 rpm for 10 min at 4°C. We subjected supernatants to SDS-PAGE electrophoresis, transferred proteins to 0.45-mm Immobilon-P polyvinylidene difluoride membranes (Millipore; IPVH00010), and detected signals using the following primary antibodies: mouse monoclonal anti-LC3 (clone 4E12, 1:1,000, MBL, M152-3), rabbit polyclonal anti-p62 (1:1,000; MBL; PM045), mouse monoclonal anti-actin (1:2,000; Proteintech; clone 7D2C10, 60008–1-Ig).

For fly tissues, we dissected fat body from larvae in PBS. We homogenized fat body in Laemmli buffer (0.1% glycerol, 2% SDS, 0.125 M Tris, pH 6.8, 0.05% β-mercaptoethanol, and 0.05% bromophenol blue) and boiled for 10 min at 100°C. Equal amounts of proteins were separated on 12% SDS polyacrylamide gels, and we transferred proteins to 0.45 mm Immobilon-P polyvinylidene difluoride membranes (Millipore; IPVH00010) according to standard procedures. We used the following primary antibodies: rabbit anti-GABARAP (1:1,000; Cell Signaling Technology; 13733), rabbit anti-Ref(2)p (1:1,000; Abcam; ab178440), mouse anti-β-tubulin (1:50; Developmental Studies Hybridoma Bank; E7), or mouse anti-β-actin (1:2,000; Proteintech; 60008–1-Ig). We used the following secondary antibodies: HRP goat anti-mouse IgG (H+L; 1:5,000; Thermo Fisher Scientific; 31430) and HRP goat anti-rabbit IgG (H+L; 1:5,000; Invitrogen, 62–6120). We performed three independent biological replicates.

Embryos of the appropriate developmental stage were collected from overnight apple juice plates, dechorionated in bleach for 90 s and mounted on double-sided sticky tape on glass slides in 10S Voltalef oil (VWR). Srp-Gal4, UAS-LifeAct-GFP/(Cyo) embryos were used as a control. We analyzed individual hemocytes, scoring the number of vacuoles at arbitrary time points over a 20-min time lapse. Imaging was performed on a Zeiss LSM880 with Airyscan laser scanning confocal microscope. Z-stacks and time-lapse images were processed and analyzed in ImageJ, Adobe Photoshop, and Adobe Illustrator software. Graphical representations were generated in GraphPad Prism 8.

For fly tissues, we dissected intestines in PBS (Gibco) 2 h APF, fixed in a solution of 2.5% glutaraldehyde and 2% PFA in 0.1M sodium cacodylate buffer, pH 7.4 (Electron Microscopy Sciences) for 1 h at room temperature followed by overnight fixation at 4°C in fresh fix. Following fixation, we washed the guts in 0.1M sodium cacodylate buffer, pH 7.4, post-fixed in 1% osmium tetroxide in distilled water for 1 h at room temperature and washed in distilled water. Preparations were stained en bloc in 1% aqueous uranyl acetate for 1 h at 4°C in the dark, washed in distilled water, dehydrated through a graded ethanol series, treated with propylene oxide and infiltrated in SPI-pon/Araldite for embedding. We cut ultrathin sections on a Leica UC7 microtome. Sections were stained with uranyl acetate and lead citrate and examined on a Phillips CM10 TEM. Images were taken down the length of the gut to ensure an unbiased approach. We reviewed all images and chose representative images for analyses. We measured area of organelles and cytoplasm in ImageJ, and graphical representations were generated in GraphPad Prism 8.

For COS-7 cells, we prefixed cells in 50% media (50% fix, 2.5% glutaraldehyde, and 2% PFA in 0.1 M sodium cacodylate buffer, pH 7.4; Electron Microscopy Sciences) for 5 min followed by fixation in full fix for 1 h at room temperature. Cells were then washed with 0.1 M cacodylate buffer, pH 7.4, post-fixed in 1% osmium tetroxide in distilled water for 1 h at room temperature, and washed in distilled water. Preparations were stained en bloc in 1% aqueous uranyl acetate overnight at 4°C in the dark and then washed in distilled water. The cells were then scraped and pelleted. Cell pellets were embedded in agarose, dehydrated through a graded ethanol series, treated with propylene oxide, and infiltrated in SPI-pon/Araldite for embedding. We cut ultrathin sections on a Leica UC7 microtome. Sections were stained with uranyl acetate and lead citrate and examined using a Phillips CM10 TEM.

We collected RNA from fat body dissected from L3 (n = 10) using Trizol Reagent (Invitrogen) and treated with DNase. We generated cDNA from 1 µg of RNA using Superscript III Reverse transcription and oligo (dT; Invitrogen; 18080051) according to manufacturer’s instruction. cDNA of indicated genes was amplified by PCR using Power SYBR Green Master Mix (Thermo Fisher Scientific; 4368577) according to manufacturer’s instructions and the following specific primers: actin5c forward primer, 5′-GGA​TGG​TCT​TGA​TTC​TGC​TGG-3′, reverse primer, 5′-AGG​TGG​TTC​CGC​TCT​TTT​C-3′, PCR product size, 146 bp; dMtmr6 forward primer, 5′-GAC​AGG​ATC​TCC​GCT​ACT​CAT-3′, reverse primer, 5′-GCA​GCG​AAG​TGT​AGA​CAT​CGT-3′, PCR product size, 98 bp; Atg1 forward primer, 5′-GTC​GGG​GAA​TAT​GAA​TAC​AGC​TC-3′, reverse primer, 5′-GCA​TGT​GTT​TCT​TGC​GAT​GAC-3′, PCR product size, 91 bp; Atg18 forward primer, 5′-GTG​TTC​GTC​AAC​TTC​AAC​CAG​A-3′, reverse primer, 5′-TGT​CCA​GGG​TCG​AGT​CCA​C-3′, PCR product size, 100 bp; Cp1 forward primer, 5′-TCA​ACT​ACA​CTC​TGC​ACA​AGC-3′, reverse primer, 5′-GCC​AGT​CCA​CAG​ATT​TGG​G-3′, PCR product size, 105 bp; Mitf forward primer, 5′-AGT​ATC​GGA​GTA​GAT​GTG​CCA​C-3′, reverse primer, 5′-CGC​TGA​GAT​ATT​GCC​TCA​CTT​G-3′, PCR product size, 115 bp; Ref(2)p forward primer, 5′-AAT​CGA​GCT​GTA​TCT​TTT​CCA​GG-3′, reverse primer, 5′-AAC​GTG​CAT​ATT​GCT​CTC​GCA-3′, PCR product size, 148 bp. We resolved the PCR products on a 1.5% agarose gel.

All experiments were performed independently at least three times, except for TEM. For mCherry-Atg8a, ref(2)p, LysoTracker red, TR-avidin, Rab7, Magic red, or GABARAP (Atg8a) antibody/stain–labeled puncta quantification, we chose at least five random images and counted the number of puncta in GFP-positive cells versus neighboring non-GFP cells. For LC3B, p62, LAMP1, or LysoTracker antibody/stain–labeled puncta, we chose at least 10 random images per genotype. We examined an average of 10–20 cells for each group. For colocalization analyses, we used the EzColocalization plugin for ImageJ to determine the Pearson correlation coefficient per image or per region of interest (Stauffer et al., 2018). Statistical significance was analyzed by paired or unpaired two-tailed t test for single comparisons and one-way ANOVA analysis for multiple comparisons. For all analyses, data distribution was assumed to be normal but was not formally tested. A P value <0.05 was considered statistically significant. Statistics were performed using GraphPad Prism 8 software.

Fig. S1 shows the full amino acid sequence alignment of CG3530, MTMR6, MTMR7, and MTMR8. Fig. S2 shows effects of MTMR8 and dMtmr6 knockdown on TFEB/Mitf signaling by immunofluorescence and gene expression changes, respectively, as well as a representative image of Rab7 localization in dMtmr6-knockdown tissue. Fig. S3 compares the effects of control versus dMtmr6-knockdown on GFP-FYVE and evaluates colocalization between GFP-FYVE and Atg8a expression in fed and starved fat body. Fig. S4 shows representative images from Atg8a-stained dMtmr6-knockdown fat body that contains Vps34- or Atg9-null clone cells stained and quantification of Atg8a puncta number. Fig. S5 shows the effects of MTMR8 mutations and MTMR8 overexpression on autophagy protein levels and lysosome reporter localization. Table S1 shows the predicted Drosophila PI phosphatases, human orthologues, reagents tested, and results of the RNAi screen for regulators of midgut autophagy. Table S2 shows the siRNA sequences that were used for cell culture experiments.

We thank the Bloomington Stock Center, the VDRC, and Well Genetics for flies; the Wolfe laboratory for use of their Neon transfection system; and the Baehrecke laboratory and Yan Zhao for constructive comments.

This work was supported by National Institutes of Health grants R35GM131689 (to E.H. Baehrecke) and 5T32CA130807-10, and Wellcome Trust fellowship 107940/Z/15/Z (to W. Wood).

The authors declare no competing financial interests.

Author contributions: E.A. Allen and E.H. Baehrecke conceptualized the experiments. E.A. Allen performed and analyzed most of the experiments. C. Amato performed and analyzed macrophage phagocytosis in embryos, T.M. Fortier performed TEM, and P. Velentzas performed Vps34 and Atg9 epistasis analyses in fat body. E.A. Allen and E.H. Baehrecke wrote the original manuscript with contributions from the other authors, including text associated with the data in Fig. 6, C and D, provided by C. Amato and W. Wood.