Drosophila WASH is required for integrin-mediated cell adhesion, cell motility and lysosomal neutralization

The Wiskott-Aldrich syndrome protein (WASP) family of nucleation promoting factors (NPF) controls actin dynamics by activating the ubiquitous Arp2/3 complex, a nucleation machinery that initiates actin filament branches on the sides of pre-existing filaments. One of three highly conserved WASP-family members is the Wiskott-Aldrich syndrome protein and SCAR homolog (WASH; CG13176 in flies) protein that is found across multiple eukaryotic kingdoms (Veltman and Insall, 2010). As all class-I NPFs, WASH possesses a VCA domain at its C-terminal end that activates the Arp2/3 complex in vitro (Linardopoulou et al., 2007). It differs considerably from other members of this class at its N-terminal part as it has a WASH-homology domain (WAHD), comprising a WASH homology domain 1 (WAHD1) and a tubulin-binding region (TBR) (Gomez and Billadeau, 2009). In cells, WASH exists in a stable pentameric complex known as the WASH regulatory complex (SHRC) that contains four additional subunits: strumpellin (encoded by KIAA0196), FAM21, SWIP (encoded by WSB1) and CCDC53. This complex appears to be structurally very similar to the WAVE regulatory complex (WRC), which controls the activity and localization of WAVE (Jia et al., 2010). Similar to the WRC, knockdown of individual WASH complex subunits results in the degradation of other WASH complex components (Derivery et al., 2009; Gomez and Billadeau, 2009; Jia et al., 2010). However, the SHRC seems to be constitutively active in contrast to the WRC, thus possibly being regulated by, so far, unknown mechanisms (Derivery and Gautreau, 2010; Gomez and Billadeau, 2009).

First insights into the cellular function of WASH-mediated actin polymerization emerged from the subcellular localization of WASH and from RNA interference (RNAi) studies in mammalian cell culture. These studies have demonstrated that WASH is an endosomal protein that regulates several sorting and maturation steps of endocytic vesicles. Upon WASH depletion, endosomes change shape from normally spherical to an enlarged and tubular shape (Derivery et al., 2009; Duleh and Welch, 2010; Gomez and Billadeau, 2009), indicating that WASH plays a role in regulating endosome tubulation and scission. WASH colocalizes with Rab4, Rab7 and Rab11, suggesting an important role in both early and late endosomal recycling pathways (Derivery et al., 2009). In line with this notion, wash-deficient cells show trafficking defects in different endocytic routes, including endosome-to-Golgi transport through the retromer complex (Gomez and Billadeau, 2009; Harbour et al., 2012; Duleh and Welch, 2012; Piotrowski et al., 2013; Zech et al., 2011) and endosome-to-plasma-membrane recycling of various cargos, such as integrins and the epidermal growth factor (EGF) and transferrin receptors (Derivery et al., 2009; Zech et al., 2011). Thus, WASH activity is thought to facilitate and maintain the segregation of endosomal sorting domains.

First evidence for the in vivo role of WASH in lysosomal maturation and autophagy comes from loss-of-function studies in Dictyostelium (Carnell et al., 2011). Normally, endosomes and, in particular, lysosomes get acidified by the activity of the vacuolar (V)-ATPase, a large multimeric ATP-dependent H⁺ pump (Maxson and Grinstein, 2014). An acidic environment provides not only an optimal pH for a wide range of acidic hydrolases but also ensures that internalized receptors release their ligands so that they can recycle back to the plasma membrane (Luzio et al., 2007; Pryor and Luzio, 2009). Dissociated ligands are then delivered to late endosomes, whereas the receptors are sorted into recycling vesicles destined for the plasma membrane. The successive acidification of endosomes further leads to highly acidic lysosomes where proteins are degraded.

Before exocytosis of indigestible material can occur, lysosomes have to be neutralized by retrieving the V-ATPase from the lysosomal membrane. This task is performed by WASH, as it interacts with the V-ATPase and generates a newly polymerized network that, in turn, pushes the V-ATPase out of the membrane (Carnell et al., 2011). Lysosomal neutralization requires a functional WASH-VCA domain, suggesting that Arp2/3-mediated actin polymerization is essential for the V-ATPase retrieval (Carnell et al., 2011). Further studies in Dictyostelium and in mammalian cells suggest a conserved role of WASH in regulating the lysosomal network required for efficient phagocytic and autophagic digestion (Kolonko et al., 2014; King et al., 2013; Gomez et al., 2012). Dictyostelium cells lacking WASH are unable to digest their cytoplasm in order to survive phases of starvation (King et al., 2013). The phenotypic analysis of wash knockout mice confirms an important in vivo function of WASH in regulating autophagy (Xia et al., 2014). WASH deficiency results in embryonic lethality and mutant mice show extensive autophagy (Xia et al., 2014). Therefore, a model has been proposed in which WASH inhibits autophagy by suppressing ubiquitination of Beclin1/Atg 6, a core component of the conserved Vps34 complex, which promotes autophagosome formation when active (Xia et al., 2014). Remarkably, different from its activity in V-ATPase retrieval and retromer-mediated endosomal sorting, the function of WASH in autophagy does not require its Arp2/3-activating activity. A truncated WASH protein lacking the VCA domain still rescues the accelerated autophagy induction in wash−/− mouse embryonic fibroblasts similar to full-length WASH (Xia et al., 2014). Surprisingly, rather than finding accelerated autophagy, Zavodszky and colleagues have recently reported impaired autophagy in HeLa cells that had been depleted for WASH (Zavodszky et al., 2014a,b). Thus, the exact role of WASH as an NPF in autophagy remains unclear so far.

In Drosophila, a role of WASH (known as Washout in flies) in lysosomal function and autophagy has not been reported yet. The initial analysis of a deletion mutant suggests an essential function of WASH in Drosophila oogenesis and larval development (Linardopoulou et al., 2007; Liu et al., 2009). However, whether the observed lethality and the oogenesis defects are due to a loss of WASH function has not been addressed in these studies. The role of WASH in developmental dispersal of immune cells during Drosophila embryogenesis is even unclear. RNAi-mediated knockdown of WASH partially affects the stereotypic migration of embryonic macrophages (Verboon et al., 2015), but the loss-of-function mutant macrophages are indistinguishable from wild type (Evans et al., 2013). By contrast, a more recent study demonstrates a conserved function in retromer-dependent endocytic recycling of the luminal protein Serpentine in Drosophila trachea development (Dong et al., 2013). wash mutant embryos show defects in tracheal tube length, phenocopying rab9 and vps35 mutants (Dong et al., 2013).

Here, we analyzed the function of WASH in Drosophila. Different from previous studies, we show that WASH is dispensable for Drosophila development. Viable mutant females show neither reduced fertility nor defects in the morphology of developing egg chambers. Instead, we found a requirement of WASH in integrin-dependent cell spreading of macrophages, and we provide further evidence for a conserved role in generating F-actin patches, in driving late endosomal recycling and in lysosome neutralization that is required for efficient phagocytic and autophagic clearance.

The Drosophila melanogaster genome contains a single wash gene located on the second chromosome (2R) at the cytological location 48E6 (Fig. 1A). It consists of a single exon encoding a 75-kDa protein. In previous studies, a wash deletion mutant (washΔ185) was generated through imprecise excision of the P-element EY15549 inserted between codon 11 and 12 of the wash open reading frame (ORF) (Linardopoulou et al., 2007). We confirmed the 1029-bp deletion in washΔ185 mutant flies (Fig. 1B; Linardopoulou et al., 2007). This deletion removes more than half of the coding region, resulting in a premature stop codon. Flies homozygous for this deletion chromosome die at the late third-instar larvae stage, indicating an essential function of WASH in Drosophila development (Linardopoulou et al., 2007; Liu et al., 2009). However, whether the lethality is due to a loss of WASH function was not addressed in those studies (Linardopoulou et al., 2007; Liu et al., 2009). The existence of viable flies bearing the 10-kb P-element insertion EY15549 in the wash ORF already suggests that the lethality of homozygous washΔ185 mutant animals might be caused by second-hit mutations. This discrepancy prompted us to test trans-heterozygous flies with the sequence-mapped deficiency Df(2R)BSC699, which removes the complete wash gene locus (Fig. 1A,B).

Trans-heterozygous washΔ185/Df(2R)BSC699 flies, but also the homozygous viable P-element insertion strain, completely lacked WASH expression (Fig. 1C). Remarkably, they were viable with no apparent developmental defects, suggesting that lethality was caused by a second-site mutation. Consistently, we could establish a viable wash mutant fly stock (termed washΔ185* from now onwards) by separating the lethal mutation from the washΔ185 chromosome (see Materials and Methods). washΔ185* mutant females produced normal eggs and were completely fertile (Fig. 1E–I′). PCR and western blot analyses confirmed the complete loss of WASH expression in these flies (Fig. 1D). WASH was strongly expressed in wild-type ovaries, and this expression was completely lost in washΔ185* mutant animals (Fig. 1D). The phenotypic analysis of homozygous washΔ185* mutant developing eggs did not reveal any morphological defects (Fig. 1F,H). Phalloidin staining of developing mutant egg chambers further confirmed a wild-type morphology and organization of actin-rich structures, including ring canals and actin baskets (Fig. 1G,I). wash mutant egg chambers also exhibited no significant defects in border cell migration, a well-established model for invasive collective cell migration (Fig. 1J). Thus, we conclude that WASH function is not required for Drosophila oogenesis.

Previous studies have highlighted an important role of WASH in efficient phagocytic and autophagic digestion in Dictyostelium (King et al., 2013). In Drosophila, different types of circulating immune cells can be found. The most abundant cell types are plasmatocytes, the professional phagocytes in flies. They fulfill similar roles as mammalian monocytes and macrophages, including phagocytosis of invading pathogens and clearance of apoptotic bodies controlling development and inflammation (Stramer et al., 2005; Wood and Jacinto, 2007). Western blot analysis confirmed specific expression of WASH in macrophage-like plasmatocytes (Fig. 2A; from here on referred to as macrophages). In these cells, WASH was most prominent at large F-actin-coated subcellular structures (Fig. 2B,B′), concentrated in the perinuclear region (Fig. 2B′). The punctate signals were absent in wash mutant macrophages, indicating antibody specificity (Fig. 2C).

Interestingly, we found that wash mutant macrophages spread properly on Concanavalin A (ConA)-coated surfaces (Fig. 2D–F), but displayed prominent defects in cell flattening and spreading on uncoated surfaces (Fig. 2C). Detailed quantification analysis confirmed a significantly reduced spreading area of both larval and pupal washΔ185* macrophages on uncoated glass cover slips (Fig. 2F). Re-expression of a WASH–EGFP transgene in washΔ185* mutant macrophages significantly restored cell spreading defects, indicating that the defects resulted from a loss of WASH function (Fig. 2G). Moreover, RNAi-mediated suppression of WASH and the known WASH complex components Strumpellin, SWIP, Fam21 and CCDC53 resulted in comparable spreading defects on uncoated surfaces (Fig. S1).

ConA-induced spreading is controlled by remodeling of the actin cytoskeleton upon binding of lectins to the polysaccharide side chains of plasma membrane proteins and lipids (Rogers et al., 2003); however, cell spreading on uncoated glass surfaces depends on integrin-mediated cell adhesion (Jani and Schöck, 2007). Because wash mutant cells spread properly on lectin-coated surfaces, we suggest that WASH plays a role in integrin-mediated cell spreading rather than in regulating lamellipodial cell protrusions. In Drosophila, the main β-integrin is βPS-integrin, also known as Myospheroid (Mys), which is required for cell spreading and cell migration of embryonic and pupal macrophages (Comber et al., 2013; Moreira et al., 2013). Structured illumination microscopy (SIM) revealed an overall punctate expression pattern of endogenous βPS-integrin in macrophages. The strongest signal was observed in the perinuclear region compared to a rather moderate concentration along the leading edge of the cells (Fig. 2H). wash mutant cells show an increased accumulation in the perinuclear region (Fig. 2I). Quantitative measurement of fluorescence intensities further confirmed an increased signal of βPS-integrin around the nuclei at the expense of peripheral staining in wash mutant cells (Fig. 2J).

To further stimulate integrin-mediated cell adhesion, macrophages were plated on surfaces coated with the ECM protein vitronectin. Cells spread to similar extents on vitronectin; however, they exhibited numerous βPS-integrin-positive punctae that were often enriched along the leading edge and formed prominent foci characteristic for focal adhesion sites in mammalian cells (Fig. 2K; Movie 1). Under these conditions βPS-integrin clusters colocalize with the reporter fusion protein EGFP–FAT, containing the focal targeting sequence (FAT) domain of Focal Adhesion Kinase (FAK; Fig. 2L). Stress-fiber-like actin filament bundles were seen emerging from these bright foci, as is typical for focal adhesion sites (Fig. 2L). Such dynamic focal adhesion sites were also formed in vivo in migrating pupal macrophages expressing EGFP–FAT imaged from prepupae (Movie 1). Similar to uncoated surfaces, washΔ185* mutant cells show cell spreading defects on vitronectin (Fig. 2M). Moreover, loss of WASH function also resulted in a reduced length of vitronectin-induced cell adhesion foci along the cell periphery marked by EGFP–FAT (Fig. 2N,O; quantification in Fig. 2P; Movie 2). Thus, our data support a conserved role of Drosophila WASH in integrin-mediated cell adhesion.

To further analyze a possible role of WASH in regulating integrin receptor trafficking, we examined the subcellular localization of WASH in Drosophila macrophages in more detail. WASH did not colocalize with EGFP–FAT at focal adhesion sites, suggesting that WASH itself is not a part of them but might rather regulate integrin receptor trafficking (Fig. 3A). WASH was found in close contact to βPS-integrin and surrounded βPS-integrin punctae in the perinuclear region (Fig. 3B). Live spinning disc microscopy analysis of WASH–EGFP transgene expression confirmed a dynamic localization of WASH at vesicular structures of different sizes (Fig. 3C; Movie 3). Co-expression of WASH–EGFP with Lifeact–RFP further illustrated a striking overlap of dynamic WASH and F-actin patches on these vesicles (Fig. 3D; Movie 4). Such actin patches could also be observed in vivo in migrating pupal macrophages that expressed Lifeact–EGFP (Movie 5). The formation of actin patches highly depends on WASH function. A wild-type macrophage showed at least two distinct actin patches on large circular vesicles, whereas loss or RNAi-mediated suppression of WASH completely abolished actin patch formation (Fig. 3E–G; quantification in Fig. 3H; Movie 6).

Further colocalization analysis with different YFP- or GFP-tagged Rab transgenes as markers for different stages of endosomal maturation revealed that actin-coated WASH vesicles mainly overlapped with small Rab4- and larger Rab7-marked endosomes, a compartment that also harbors the retromer complex (Fig. 3I–U; quantification of colocalization in Fig. 3M; Seaman and Freeman, 2014; Seaman et al., 2009). This suggests either that WASH regulates Rab4-mediated fast integrin receptor recycling from small early endosomes to the plasma membrane or that it controls recycling from a Rab7 late endosomal compartment, as previously proposed for α5β1-integrin recycling in mammalian cells (Zech et al., 2011). Consistently, we frequently found βPS-integrin punctae within Rab4- and Rab7-positive vesicles (Fig. 4A,B). Interestingly, we observed an increased accumulation of βPS-integrin in Rab7- but not in Rab4-marked endosomes in wash mutant macrophages (arrowheads in Fig. 4C,E). Quantification of βPS-integrin fluorescence intensity confirmed a moderate but significant increase of βPS-integrin in the lumen of late endosomes marked by endogenous Rab7 in wash mutant macrophages (Fig. 4F). Thus, we propose that WASH might act predominantly at that stage of late endosomal recycling of βPS-integrin in Drosophila, as previously suggested for α5β1-integrin recycling in mammalian cells (Zech et al., 2011).

We next analyzed a possible role of WASH in integrin-mediated cell migration in vivo. Previous data have revealed an important function of βPS-integrin-mediated cell adhesion in two-dimensional random cell migration in white prepupae, in vivo (Moreira et al., 2013). Sub-epidermal macrophages acquire a spread morphology with broad lamellipodial protrusions and collectively initiate random single cell migration (Fig. 5A, quantification in Fig. 5D–F; Movie 7; Moreira et al., 2013; Brinkmann et al., 2016). Prepupal wash mutant cells also showed a reduced cell spread area in vivo (Fig. 5D). Macrophage-specific RNAi-mediated suppression of βPS-integrin function affected cell spreading to a similar extent and nearly completely abolished cell migration (Fig. 5B, quantification in Fig. 5D–F; Movie 7). Loss of WASH function also resulted in significant albeit weaker cell migration defects with reduced migratory velocity and distance (Fig. 5C, quantification in Fig. 5E,F; Movie 7).

Additionally, we analyzed directed cell migration of macrophages upon laser-induced wounding in later pupal stages. At this stage, macrophages migrate in three dimensions rather than in two dimensions (Sander et al., 2013; Brinkmann et al., 2016). Wild-type macrophages immediately responded to a laser-ablated single cell and migrated towards the wound (ablated cell is marked by a yellow circle in Fig. 5G; Movie 8). βPS-integrin-depleted cells were able to respond and migrate towards wounds, however with strongly reduced directionality and velocity (Fig. 5H; Movie 8). By contrast, loss of WASH function had no effect on the wound response and directed cell migration during the initial phase (5–10 min after wounding; Fig. 5I; Movie 8). However, we found a reduced number of macrophages that reached the wound after 30 min by measuring the histogram-based macrophage migration score (HMMS) that we have previously established (Lammel et al., 2014; quantification in Fig. 5J). These data indicate that WASH function is important for macrophage cell motility in vivo.

Previous studies in Dictyostelium have revealed an important role of WASH in lysosome function. Given the localization of Drosophila WASH to large Rab7-positive late endosomes and/or pre-lysosomes, we further analyzed a possible conserved role of WASH in lysosome function. Mature lysosomes in macrophage-like Drosophila S2R⁺ cells marked by a GFP fusion with the Lysosome-associated membrane protein 1 (Lamp1) were characterized by numerous F-actin patches visualized by a Lifeact–mCherry fusion (Fig. 6A; Movie 9). We also found WASH on the surface of these large lysosomes marked by either Lamp1 or Vha55, the V-ATPase regulatory B-subunit in fixed and living cells (Du et al., 2006; Fig. 6B,C; Movies 10, 11). Strikingly, loss of WASH function in macrophages resulted in enlarged Vha55-marked lysosomes (Fig. 6D,E; quantification in Fig. 6F), suggesting a conserved role in lysosomal function. Supporting this notion, WASH can physically associate with the V-ATPase, as found in co-immunoprecipitations with a Myc-tagged Vha55 construct (Fig. 6G).

Based on the functional and physical interactions between WASH and the V-ATPase subunit, we further examined possible defects in phagolysosome neutralization in wash mutant macrophages. These cells efficiently phagocytize Escherichia coli particles conjugated with a pH-dependent fluorescent dye (pHrodo) within a few minutes. The fluorescence intensity of the pHrodo dye increases as the pH of its surrounding becomes acidic. Once a phagocytized pHrodo-conjugated E. coli enters a lysosome, the acidic environment within the phagolysosome causes a dramatic increase in fluorescent emission (Fig. 7A′; Movie 12). These phagolysosomes showed prominent actin patches (arrowheads in Fig. 7A′; Movie 12). Strikingly, WASH–EGFP was found at phagocytized E. coli pHrodo particles upon acidification and accumulated together with F-actin in distinct patches on phagolysosomes (Fig. 7B,C; quantification in D; Movie 13). Loss of WASH function did not affect phagocytic uptake but rather resulted in a significant defect in lysosomal neutralization (Fig. 7E,F). The neutralization defect in wash mutant macrophages became most obvious 60 min after phagocytic uptake (Fig. 7E,F). From this time onwards, the fluorescent emission still increased continuously in wash mutants, whereas in wild-type cells, the fluorescence intensity ceased after one hour (Fig. 7E,F). Thus, these data strongly support a conserved role of WASH in the retrieval of the V-ATPase from lysosomes.

Similar to the phagocytic pathway, autophagy utilizes lysosomes to degrade cytoplasm, ubiquitylated proteins or non-functional organelles. Autophagy can occur as a direct transport of selected proteins across the lysosomal membrane (microautophagy) or as an engulfment of cytoplasm and organelles by double-membrane vesicles that later on fuse with lysosomes (macroautophagy; Xie and Klionsky, 2007). Under starvation, autophagy is upregulated by the cells in order to supply the organism with nutrients (Russell et al., 2014; Xie and Klionsky, 2007).

The cellular function of WASH in lysosomal neutralization might not be restricted to the digestion of phagocytic cells such as macrophages but could also be important for starvation-induced autophagy in the Drosophila fat body, a nutrient storage organ analogous to the vertebrate liver. This tissue has a strong, well-characterized autophagic response to starvation, which flies use to mobilize nutrients and promote their survival. We used Lysotracker Red staining for the detection of lysosomes and autophagosomes. Fat bodies from well-fed wild-type third-instar larvae remained essentially negative for Lysotracker (Fig. 8A–C). In contrast, Lysotracker staining in the fat body of fed wash mutant animals was already increased (Fig. 8D–F). However, in response to amino acid starvation (sugar-only diet), we observed a more dramatic induction of enlarged punctate Lysotracker staining in mutant cells compared to in wild type (compare Fig. 8I,L). Induced autolysosomes in the wash mutant fat body seem to be functional given that they also showed an increased Magic Red staining, a marker for active cathepsin B, suggesting that the pH within the lysosomal lumen is more acidic promoting its hydrolytic activity during starvation-induced autophagy (Fig. 8M,N). Interestingly, WASH localizes to actin-enriched vesicular structures in the fat body cells, similar to what was observed in macrophages (Fig. 8O). These fat body structures were also marked by an Atg8–GFP transgene (Fig. 8P), a key autophagic protein that associates with autophagosomes (Shpilka et al., 2011). To further test an increased induction of functional autophagic vesicles, we analyzed Atg8 lipidation by immunoblotting of fat body lysates. Lipidation of Atg8 drives its association with the autophagosome and serves as an indicator of autophagy induction (Shpilka et al., 2011). The non-processed form (often referred to as Atg8-I) can easily be distinguished from its membrane-associated faster migrating active form (Atg8-II). Consistent with our finding that lysosomal acidity was increased in the wash mutant fat body, immunoblot analysis of Atg8 showed that the amount of both the non-processed and lipidated form was increased (Fig. 8Q). Thus, loss of WASH function increases starvation-induced autophagy. Finally, we examined possible changes in the life span of wash mutant flies under amino acid starvation. Under these conditions, wash mutants displayed an accelerated lethality compared to control flies (Fig. 8R). Increased lethality of starved wash mutant flies could be rescued by expression of the WASH–EGFP transgene in the mutant background (Fig. 8R). Thus, increased acidification of autolysosomes might shorten life span rather than promote longevity, similar to results of a recent study with WASH conditional knockout (KO) mice (Xia et al., 2013).

Taken together, our data highlight a conserved role for WASH in the endocytic sorting and recycling of membrane proteins like integrins and the V-ATPase. A loss of WASH function therefore leads to defects in cell spreading, cell motility and lysosomal neutralization. Lysosome neutralization defects in particular might contribute to a decreased life expectancy under amino acid starvation because nutrients are stuck in lysosomes and cannot nourish the starving cells or organism.

Drosophila wash was originally identified as an essential gene required for oogenesis and larval development (Linardopoulou et al., 2007; Liu et al., 2009). Our data clearly show that both larval lethality and all morphological defects of developing eggs described so far are not due to the loss of WASH function but are rather caused by a second-hit mutation. The generation of viable wash mutant flies bearing the same chromosomal deletion finally confirmed that WASH is dispensable for early Drosophila development and oogenesis. The defects in oogenesis in trans-heterozygous washΔ185/wimp mutant flies observed by Liu et al. are probably due to the wimp mutation used given that wimp is a change-of-function mutation of an RNA polymerase II subunit and reduces maternal function of a variety of genes, including, but not limited to, wash (Liu et al., 2009; Parkhurst and Ish-Horowicz, 1991). Despite the apparent lack of any obvious phenotype during oogenesis, WASH seems to be expressed in Drosophila ovaries. Thus, redundant or compensatory pathways might mask WASH function in developing eggs. Previous expression data in wild-type flies revealed an accumulation at the oocyte cortex in addition to an overall cytoplasmic expression in all nurse and somatic follicle cells throughout oogenesis (Rodriguez-Mesa et al., 2012). The protein-A-affinity-purified monoclonal antibody (mAb P3H3; Rodriguez-Mesa et al., 2012) used in this study specifically recognizes WASH in immunoblots (Fig. 1B–D). However, in immunostainings of developing egg chambers, we found no significant differences in the expression pattern and signal intensity between wild type and mutants.

We observed a specific and prominent expression in Drosophila macrophages and the fat body that was completely lost in wash mutants. Our data strongly support a conserved role of Drosophila WASH as an endosome-specific NPF in regulating integrin trafficking and lysosome function. Important regulators of endosomal trafficking and recycling are Rab GTPases (Jordens et al., 2005; Wandinger-Ness and Zerial, 2014). Internalized integrin receptors enter early endosomes and can either be sorted by a ‘fast’ Rab4-mediated recycling pathway or transit through a ‘slow’ Rab11-positive recycling pathway back to the plasma membrane (Bretscher, 1992; Rainero and Norman, 2013). Our data further support previous findings of a WASH function in regulating late endosomal integrin recycling in mammalian cell culture. WASH significantly colocalized with Rab4- and Rab7-marked endosomes in macrophages. However, we only found a significant difference in βPS-integrin signals in Rab7-marked endosomes in wash mutant macrophages. Thus, fast Rab4-mediated recycling of integrins might either be independent of WASH or might be masked by redundant functions of other WASP-protein family members. In summary, reduced focal adhesions, cell spreading and cell migration defects in wash mutant macrophages are most likely to be caused by an impaired recycling of βPS-integrin.

In wild type, receptors that are not recycled remain within early endosomes as they undergo early-to-late endosomal maturation, a step that requires the exchange of Rab5 for Rab7. Rab7 is known to recruit the retromer on endosomal membrane, and it is required for the transfer of cargo from the late endosomes to the lysosome (Priya et al., 2015; Pryor and Luzio, 2009). A conserved function of WASH in retromer-dependent endocytic recycling of the luminal protein Serpentine has already been found during Drosophila trachea development (Dong et al., 2013). In macrophages, WASH also associates with large circular Rab7-positive late endosomes and mature lysosomes. We found prominent actin patches that depend on WASH function on both endosomal compartments. Such localized actin patches are thought to define discrete endosomal membrane domains that facilitate protein sorting and vesicle fission (Derivery and Gautreau, 2010; Derivery et al., 2012, 2009; Seaman et al., 2013). Carnell and colleagues described the first mechanistic model of how WASH-dependent actin polymerization directly controls protein sorting in Dictyostelium lysosomes (Carnell et al., 2011). In Dictyostelium, WASH physically associates with the V-ATPase, a key H⁺ pump that is crucial for the establishment and maintenance of the acidic pH in endosomes and lysosomes (Maxson and Grinstein, 2014). WASH directly drives the removal of V-ATPase from lysosomes and its recycling to small vesicles. Once the V-ATPase has been recycled, the lysosome becomes neutralized. Without WASH-driven actin polymerization, lysosomes never recycle V-ATPase into small vesicles and therefore do not neutralize to form post-lysosomes. Consequently, loss of WASH results in an insufficient retrieval of the V-ATPase and subsequently in an accumulation of undigested phagocytosed material that cannot be excreted by Dictyostelium (Carnell et al., 2011). In this study, we have provided the first evidence that this function of WASH is not unique for Dictyostelium and that WASH is likely to have a conserved role in lysosome neutralization and V-ATPase recycling in higher eukaryotes such as Drosophila. Drosophila WASH binds to Vha55, the V-ATPase regulatory B-subunit, and loss of WASH function strongly affects phagolysosomal neutralization in macrophages.

Increased lysosome acidification is expected to have global effects not only on phagocytic digestion but also on autophagy, an evolutionarily conserved process that mediates the degradation of intracellular materials in lysosomes. Supporting this notion, we observed an increased starvation-induced acidification of autolysosomes that significantly reduced the life span of mutant flies. A similar, but more severe phenotype has recently been reported in wash knockout mice (Xia et al., 2014). WASH deficiency results in extensive autophagy that most likely causes embryonic lethality (Xia et al., 2014). Thus, WASH function inhibits rather than stimulates autophagy, as previously proposed based on RNAi cell culture experiments (Zavodszky et al., 2014a). The mechanisms promoting autophagy in mice and flies, however, seem to be different. Increased autolysosomal acidification seems to promote autophagy in wash mutant flies, whereas in mice, WASH might directly inhibit the formation of autophagosomes before they fuse with lysosomes to form functional autolysosomes. This direct effect of WASH on autophagosome formation is independent of its function as an NPF but seems to be mediated by inhibiting ubiquitylation of Beclin1/Atg6 (Xia et al., 2014). This tumor suppressor is part of a lipid kinase complex that mediates the initial stages of autophagosome biogenesis by recruiting other Atg proteins (Levine et al., 2015; Wirawan et al., 2012). However, Beclin1/Atg6 proteins are known to exert numerous non-autophagic functions, including protein sorting, and endocytic and phagosomal maturation (Levine et al., 2015; Wirawan et al., 2012). In flies, Atg6 is also required for multiple vesicle trafficking pathways and hematopoiesis (Shravage et al., 2013). Like in mammals, Drosophila Atg6 also directly interacts with the lipid kinase Vps34, and co-expression of Atg6 and Vsp34 is sufficient to induce autophagy. Consistent with its role as a tumor suppressor, loss of Atg6 function causes defects in blood cell differentiation, resulting in an overproduction of macrophages (Shravage et al., 2013). Given the inhibitory function of WASH on Beclin1/Atg6, we overexpressed WASH in flies. However, neither macrophage-specific nor ubiquitous overexpression of WASH resulted in an increased number of macrophages or the formation of melanotic blood cell masses (data not shown). Thus, gain of WASH function cannot phenocopy atg6 mutants, suggesting distinct or more complex regulatory mechanisms in flies compared to in mammals.

All crosses were performed at 25°C unless indicated otherwise. washΔ185 flies were obtained from Susan Parkhurst (Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA) (Linardopoulou et al., 2007). Lethal mutations on the washΔ185 chromosome were removed by multiple outcrossings to wild type (w1118). The following strains were used: pUASt-Vha55-EGFP (Davies et al., 1996); hmlΔ-Gal4 (Sinenko and Mathey-Prevot, 2004); pUASt-EGFP-FAT (Sander et al., 2013), and UASt-Lifeact–EGFP, UAS-Lifeact-RFP, Df(2R), P{UASp-YFP.Rab4}, P{UAS-Rab7.GFP} from the Bloomington Stock Center (IDs 26500, 58362, 9767, 24616 and 42706, respectively); RNAi lines from the Vienna Drosophila RNAi Center (VDRC; IDs: 24642, 39769, 24643, 107954, 31840, 23832, 110316 and 318285). Transgenic pUASt-WASH-EGFP flies were generated using ΦC31-mediated transgenesis (M{3xP3-RFP.attP}ZH-86F; Bischof et al., 2007).

Drosophila S2R⁺ cells were propagated in 1× Schneider's Drosophila medium, as described previously (Stephan et al., 2008). S2R⁺ cells were transfected as described previously (Bogdan et al., 2004).

SIM images were taken with an ELYRA S.1 microscope (CellObserver SD, 63×1.4 oil-immersion objective; Zeiss) using software ZEN 2010 D (Zeiss). Immunostaining and image acquisition were performed as previously described (Brinkmann et al., 2016; Zobel and Bogdan, 2013). Primary antibodies against the following proteins were used as follows: WASH [P3H3 from Developmental Studies Hybridoma Bank (DSHB), purified by protein-A affinity chromatography, 1:100], βPS-integrin (CF.6G11 from DSHB, 1:10); Rab7 (1:3000; Tanaka and Nakamura, 2008).

Live imaging of macrophage cultures was performed using a CellObserver SD spinning disc microscope (Zeiss), as reported previously (Sander et al., 2013).

Co-immunoprecipitation with WASH–EGFP and Vha55–Myc were performed as previously described (Fricke et al., 2009). Western blots were stained with primary antibodies against Myc (9E10 from DSHB, 1:10) and WASH (P3H3 from DSHB, 1:100).

Single female (1 day old) wild-type and washΔ185* mutant flies were crossed with four male (1 day old) wild-type flies and left on Drosophila standard food for 7 days. 14 days after the crosses, obtained offspring were counted for each cross. The Mann–Whitney test was used for statistical analysis; P-value<0.2121 (two-tailed).

For immunohistochemistry, female wild-type and washΔ185* mutant flies were kept on Drosophila standard food supplemented with fresh yeast for 1.5 days before the ovaries were dissected. Isolated ovarioles were fixed for 15 min with 4% paraformaldehyde (Sigma Aldrich) in PBS, permeabilized for 1 h in 0.1% Triton X-100 (Invitrogen) in PBS, blocked for 30 min in 3% BSA in PBS and subsequently stained with phalloidin–Alexa-Fluor-488 and DAPI for 2 h. The E. coli particle pHrodo uptake assay, Lysotracker and Magic Red stainings were performed according to the manufacturers’ instructions (Molecular Probes, USA).

A total number of 50 wild-type and 47 washΔ185* mutant macrophages were imaged, and all Vha55-positive vesicles in a 0.5-µm stack were measured using the AxioVision SE64 Rel 4.9 software. The Mann–Whitney test was used for statistics; P-value<0.0001 (two-tailed). To quantify the number of actin spots at vesicular structures per cell, a total number of 80 wild-type, 95 washΔ185* and 88 hmlΔ-Gal4>wash-RNAi macrophages expressing UASt-Lifeact–EGFP were imaged. The Mann–Whitney test was used for statistical analysis; both times a P-value<0.0001 was obtained (two-tailed).

To quantify the βPS-integrin immunostaining signal intensity in Rab7-positive vesicles, a total number of six wild-type and six washΔ185* mutant macrophages were imaged using SIM. 12 vesicles in wild type and 19 in washΔ185* were analyzed using FIJI. The t-test was used for statistical analysis; P-value<0.047. To quantify βPS-integrin immunostaining signal intensity and distribution, a total number of 23 wild-type and 23 washΔ185* mutant macrophages were imaged using SIM. A diametrical line was drawn across each cell, thereby determining the quantification section to be used. To overcome differences in length, each line was rescaled by setting the beginning point to 0 and the endpoint to 1. In this rescaled line, the position of each pixel was also rescaled between 0 and 1. To have relevant data for statistical purposes, the rescaled line was divided into 100 equidistant slices. The immunostaining signals within each slice were pooled, and the mean intensity was calculated. The shading represents s.e.m. For focal adhesion quantification, a total number of 75 wild-type and 53 washΔ185* mutant macrophages that expressed UASt-EGFP-FAT were imaged using spinning disc microscopy and then quantified (1727 in wild type and 1346 in washΔ185*) using the ZEN lite 2012 software. The Mann–Whitney test was used for statistical analysis; P-value<0.0001 (two-tailed).

Live imaging of macrophages in prepupae and in the pupal wing was performed as reported recently (Sander et al., 2013; Brinkmann et al., 2016). Number of flies used for two-dimensional migration: WT (n=68), washΔ185* (n=143), hmlΔ-Gal4>mys-RNAi (n=64). The Mann–Whitney test was used for statistical analysis; both times a P-value<0.0001 (two-tailed) was obtained. The number of flies used for three-dimensional wound ablation: WT (n=18), washΔ185* (n=22), hmlΔ-Gal4>mys-RNAi (n=10).

100 wild-type and 100 washΔ185* mutant stage-10 egg chambers were stained for F-actin (phalloidin) and nuclei (DAPI). For scoring the migration of border cell clusters, the egg chambers were divided into four regions: 75–100% migration, 50–75% migration, 25–50% migration and 0–25% migration. The Wilcoxon matched-pairs signed-rank test was used for statistical analysis; P-value >0.9999 (two-tailed).

We thank K. Brinkmann and K. Willrodt for performing initial RNAi experiments; J. A. Dow (Institute of Molecular, Cell & Systems Biology, College of Medical, Veterinary & Life Sciences, University of Glasgow, Glasgow, UK), S. Önel (Philipps-Universität Marburg, FB Biologie, Entwicklungsbiologie, Germany) and S. Parkhurst for sharing plasmids and flies. The anti-Rab7 antibody was kindly provided by Thomas Klein (Institute of Genetics, Heinrich-Heine-Universität Düsseldorf,Germany). We also thank C. Klämbt for critical reading of the manuscript, and the Bloomington Stock Center and VDRC for fly stocks.