A secreted factor NimrodB4 promotes the elimination of apoptotic corpses by phagocytes in Drosophila

Abstract

Programmed cell death plays a fundamental role in development and tissue homeostasis. Professional and non‐professional phagocytes achieve the proper recognition, uptake, and degradation of apoptotic cells, a process called efferocytosis. Failure in efferocytosis leads to autoimmune and neurodegenerative diseases. In Drosophila, two transmembrane proteins of the Nimrod family, Draper and SIMU, mediate the recognition and internalization of apoptotic corpses. Beyond this early step, little is known about how apoptotic cell degradation is regulated. Here, we study the function of a secreted member of the Nimrod family, NimB4, and reveal its crucial role in the clearance of apoptotic cells. We show that NimB4 is expressed by macrophages and glial cells, the two main types of phagocytes in Drosophila. Similar to draper mutants, NimB4 mutants accumulate apoptotic corpses during embryogenesis and in the larval brain. Our study points to the role of NimB4 in phagosome maturation, more specifically in the fusion between the phagosome and lysosomes. We propose that similar to bridging molecules, NimB4 binds to apoptotic corpses to engage a phagosome maturation program dedicated to efferocytosis.

This study suggests a role of the secreted protein NimB4 in phagosome maturation. Similar to bridging molecules, NimB4 binds to apoptotic corpses to engage a phagosome maturation program dedicated to efferocytosis.

Introduction

The clearance of apoptotic cells by phagocytes is an essential process during development and for the maintenance of tissue homeostasis (Arandjelovic & Ravichandran, 2015). Defective apoptotic cell clearance affects organ function and leads to the release of cytotoxic molecules and the development and progression of inflammatory and immune responses (Elliott & Ravichandran, 2010; Nagata et al, 2010; Poon et al, 2014). Recent studies also point to the importance of apoptotic cell clearance in the brain, as excessive or reduced phagocytosis can lead to neurodegeneration (Etchegaray et al, 2016; Purice et al, 2016; Hakim‐Mishnaevski et al, 2019; Hilu‐Dadia & Kurant, 2020). Efferocytosis, the phagocytosis of apoptotic cells, is a multi‐step process often mediated by professional phagocytes such as macrophages and glia. It begins with the identification of the target cells by receptors that recognize specific “eat‐me” signals that are exposed at the surface of apoptotic cells and distinguish them from living cells (Elliott & Ravichandran, 2016; Boada‐Romero et al, 2020). One of the best characterized “eat‐me” signals is phosphatidylserine (PS), a membrane phospholipid species usually found in the inner leaflet of the plasma membrane, but exposed on the surface of cells undergoing apoptosis (Segawa & Nagata, 2015; Nagata et al, 2016). Upon ligand recognition, phagocytic receptors engage downstream signaling pathways that initiate the uptake of the particle through an active and dynamic remodeling of the plasma membrane, which is mainly guided by the actin cytoskeleton (Pearson et al, 2003; Stuart & Ezekowitz, 2005; Melcarne et al, 2019a). Newly formed phagosomes undergo a maturation process to acquire digestive activity. The phagosome matures through fusion with endosomes and lysosomes, a process involving small GTPases of the Rab family (Kinchen & Ravichandran, 2010). Rab5 regulates the initial fusion events, by tethering early endosomes to the newly formed phagosome (Alvarez‐Dominguez et al, 1996; Duclos et al, 2000; Kinchen et al, 2008). As the phagosome matures, Rab5 is replaced by Rab7 in a process called Rab conversion (Kinchen & Ravichandran, 2010; Poteryaev et al, 2010; Yousefian et al, 2013). Once recruited, Rab7 is responsible for inducing fusion of the phagosome and the lysosome, initiating acidification and digestion of the corpse. A characteristic protein present on the phagolysosome membrane is lysosomal‐associated membrane protein 1 (Lamp1), which is needed for lysosome fusion with the phagosome (Huynh et al, 2007). Less is known about the final stage of phagosome resolution and how the phagosome is resorbed to allow the phagocyte to return to homeostasis. However, studies have shown that bacteria‐containing phagolysosomes in macrophages undergo fragmentation through vesicle budding and constriction (Levin et al, 2016; Botelho et al, 2020; preprint: Lancaster et al, 2020).

Recent studies in mammals have highlighted the importance of bridging molecules in the regulation of efferocytosis. Bridging molecules are secreted proteins that bind to the “eat‐me” signal to provide a link between apoptotic cells and phagocytic receptors and enhancing recognition and phagocytosis of apoptotic cells (Savill & Fadok, 2000; Ravichandran, 2003, 2011). In mammals, several bridging molecules have been identified, including milk fat globule‐epidermal growth factor 8 (MFG‐E8) and growth arrest‐specific 6 (Gas6) (Wu et al, 2006). Studies have shown that MFG‐E8 binds in a Ca²⁺‐dependent manner to PS exposed at the surface of apoptotic cells, engaging their uptake by integrins αvβ3/5 (Hanayama et al, 2002; Borisenko et al, 2004; Kusunoki et al, 2012). Moreover, the absence of MFG‐E8 in murine models causes a lupus‐like autoimmune disease due to defective clearance of apoptotic cells (Huang et al, 2017). In Caenorhabditis elegans (C. elegans), the secreted protein TTR‐52 mediates recognition of dying cells by linking PS with the phagocyte receptor CED‐1 (Wang et al, 2010). While these studies have highlighted the relevance of bridging molecules in efferocytosis, the precise function of these molecules remains poorly understood. Some studies point to a role during the early phase of phagocytosis, facilitating apoptotic cell uptake, while others suggest that bridging molecules orient the maturation of the phagosome (Peng & Elkon, 2011; Galvan et al, 2012). The mechanism underlying efferocytosis has been well characterized in Drosophila (Serizier & McCall, 2017; Davidson & Wood, 2020). In this insect, three phagocytic receptors, the α‐PS3/βν integrin heterodimer, Draper, and SIMU, have been identified for their role in the uptake of apoptotic cells (Freeman et al, 2003; Manaka et al, 2004a; Kurant et al, 2008; Nagaosa et al, 2011; Nonaka et al, 2013; Roddie et al, 2019). Studies have shown that the phagocytic ability of embryonic glia to engulf and degrade apoptotic neurons is determined by both SIMU and Draper. SIMU is required for recognition and engulfment of apoptotic neurons by glia, whereas Draper is mostly needed for their degradation (Kurant et al, 2008; Shklyar et al, 2014). Flies deficient for the two main receptors for apoptotic cells, Draper and SIMU, are viable but their lifespan is reduced and their central nervous system accumulates apoptotic bodies (Draper et al, 2014; Etchegaray et al, 2016). Although several engulfment receptors in Drosophila have been identified, little is known about their mechanism of action. So far, no bridging molecules have been identified in Drosophila.

In this study, we find that NimB4, a secreted protein of the Nimrod family, binds to apoptotic corpses in a PS‐dependent manner to promote phagocytosis of apoptotic cells by glia and macrophages (also called plasmatocytes). Macrophages from NimB4 mutants show defective phagocytosis of apoptotic cells ex vivo. Similar to draper mutants, NimB4 mutants accumulate apoptotic cells in both embryonic and larval phagocytes, and exhibit motor defect in larvae and reduced lifespan of adults. Further functional studies reveal the role of NimB4 in phagosome maturation, notably in the formation of the phagolysosome. Collectively, this study identifies the first secreted factor involved in the phagocytosis of apoptotic cells in Drosophila and suggests an evolutionarily conserved role of bridging molecules in efferocytosis.

Results

NimB4 is a secreted protein enriched in phagocytes

To identify potential bridging molecules, we searched for secreted proteins expressed in phagocytes that could promote the phagocytosis of apoptotic cells. We focused our attention on secreted members of the Nimrod superfamily with characteristic EGF‐like repeats (Bork et al, 1996; Callebaut et al, 2003; Kurucz et al, 2007), transmembrane Nimrod proteins have already been implicated in phagocytosis of bacteria (NimC1, Eater) or apoptotic corpses (SIMU, Draper), and in macrophages sessility and adhesion (Eater). In addition to these receptors, there are five secreted proteins of the Nimrod family called NimB1, B2, B3, B4, and B5 (Ju et al, 2006; Kurucz et al, 2007; Somogyi et al, 2008). Of these, only NimB5 has been characterized and has been shown to regulate peripheral hematopoiesis (Ramond et al, 2020). As bridging molecules are usually secreted by phagocytes, we monitored the expression of these five NimBs in macrophages of third‐instar larvae. Consistent with a recent transcriptome analysis (Ramond et al, 2020), RT–qPCR analysis showed that NimB4 was the most highly expressed NimB member in macrophages (Fig 1A). To track the expression of NimB4 at the subcellular level, we generated a transgenic fly line carrying a V5‐sGFP‐tagged NimB4 fusion under its own regulatory sequences, derived from the Dresden pFlyFos collection (Sarov et al, 2016). We used this reporter gene to follow NimB4 during embryonic and larval stages with a focus on glia and macrophages. At the embryonic stage, NimB4‐sGFP appeared to be strongly enriched exclusively in macrophages (Fig 1B), whereas at the larval stage NimB4‐sGFP was expressed in both macrophages and glial cells (Figs 1C and D, and EV1A). Live and fixed imaging of third‐instar larval macrophages showed that NimB4‐sGFP is present intracellularly. Live confocal imaging showed that NimB4‐sGFP localizes in small dots throughout the cytoplasm, suggesting a localization in the endosomal compartment (Fig 1C).

Figure 1. NimB4 is enriched in phagocytes.

1. RT–qPCR analysis of NimB1, NimB2, NimB3, NimB4, and NimB5 transcripts, normalized to RpL32, from carcass, macrophages, fat body, and gut of wild‐type (w¹¹¹⁸) L3 wandering larvae. Data are represented as mean ± SD from three independent experiments.
2. Representative confocal imaging of NimB4‐sGFP macrophages of embryo at stage 16; lateral (left) and ventral (right) views; tissues were stained with anti‐GFP (green, corresponding to NimB4‐sGFP) and anti‐SIMU antibodies (red, corresponding to macrophages). The arrows show the presence of NimB4‐sGFP inside the embryonic macrophages. Scale bars = 20 μm.
3. Representative live confocal imaging of NimB4‐sGFP macrophages of L3 wandering larvae at 30 and 60 min after dissection. Scale bars = 10 μm.
4. Representative confocal imaging of NimB4‐sGFP macrophages of L3 wandering larvae after fixation. Cells were stained with phalloidin (gray), DAPI (blue), and anti‐GFP (green, corresponding to NimB4‐sGFP). Scale bars = 10 μm.

Figure EV1. NimB4 is induced by clean injury.

• A
  Representative confocal images from confocal stacks of NimB4‐sGFP third‐instar larval brains. Tissues were stained with anti‐GFP (green, corresponding to NimB4‐sGFP) and anti‐Draper (red, corresponding to glial cells). The arrows indicate the presence of NimB4‐sGFP in the glial cells of third‐instar larval brain. Scale bars = 100 μm.
• B, C
  Quantification of the NimB4 transcripts, relative to RpL32 expression from macrophages (B) and fat body (C) extracts of wild‐type (w¹¹¹⁸ ) animals in unchallenged conditions (UC) and 1, 2, 6 h after clean injury. Statistics compare a challenged sample and its own unchallenged sample. Data are represented as mean ± SD from three independent experiments with five animals each (*P < 0.05 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
• D
  Western blot analysis of hemolymph from L3 larvae expressing NimB4‐RFP, NimB5‐RFP, or SP^vkg‐RFP from the fat body with the driver Lpp‐Gal4 driver. (NimB4: 77.4 kDa, NimB5: 62.3 kDa and RFP: 27 kDa, Tubulin: 55 kDa).
• E
  Representative images of L3 larvae overexpressing NimB4‐RFP, SP^vkg‐RFP (a secreted RFP), or CD8‐GFP in the fat body (CD8‐GFP is a membrane‐addressed GFP used to confirm that the Gal4 driver is not expressed in the nephrocytes) in the fat body. NimB4‐RFP and SP^vkg‐RFP accumulate in the pericardial nephrocytes indicating that both proteins are secreted into the hemolymph. CD8‐GFP is absent from the pericardial nephrocyte indicating that the Lpp driver is not expressed in this tissue. The arrows show the presence of RFP in the pericardial nephrocytes of the larvae. The brightness of the image is modified to improve visualization of the signal.
• F
  Representative images of L3 larvae overexpressing NimB4‐RFP, SP^vkg‐RFP, or CD8‐GFP in the hemocyte. NimB4‐RFP and SP^vkg‐RFP accumulate in the pericardial nephrocytes indicating that both proteins are secreted into the hemolymph. CD8‐GFP is absent from the pericardial nephrocyte indicating that the Hml driver is not expressed in nephrocytes. The arrows show the presence of RFP in the pericardial nephrocytes of the larvae. The brightness of the image is modified to improve visualization of the signal.

To investigate whether expression of NimB4 was modulated upon tissue damage, we collected larvae after clean injury with a thin needle in the anterior dorsal cuticle and monitored NimB4 expression in the two major immune‐responsive tissues, the fat body, and hemocytes. We observed that clean injury induces NimB4 expression in hemocytes with an acute phase profile 1 h following challenge (Fig EV1B). In contrast, the transcription of NimB4 in the fat body remained low and unchanged (Fig EV1C).

NimB4 binds to apoptotic cells

NimB4 is predicted to be a secreted protein due to the presence of a signal peptide (Somogyi et al, 2008). We generated a transgenic fly lines carrying a UAS‐NimB4‐RFP insertion and overexpressed NimB4‐RFP in the fat body using the fat body‐specific Gal4 driver Lpp. Western blot analysis of larval hemolymph extracts confirmed that NimB4‐RFP protein was enriched in the hemolymph, as expected for a secreted protein (Fig EV1D). Furthermore, a strong NimB4‐RFP signal was detected in the nephrocytes (a filtrating organ involved in the removal of hemolymph proteins) when NimB4‐RFP was expressed either from the fat body or the hemocytes (Fig EV1E and F) (Ivy et al, 2015; Troha et al, 2019).

MFG‐E8 binds PS exposed at the surface of apoptotic cells in a Ca²⁺‐dependent manner, engaging their uptake by the integrin α_vβ_3/5 phagocytic receptor (Borisenko et al, 2004). To explore if NimB4 could similarly bind to apoptotic cells, we incubated hemolymph containing NimB4‐RFP with healthy S2 cells, or with S2 cells undergoing apoptosis. We observed that NimB4‐RFP protein colocalized exclusively with the apoptotic cells, stained with carboxyfluorescein succinimidyl ester (CFSE; Figs 2B and D, and EV2A). The binding of NimB4 to apoptotic cells was specific as no binding was observed when labeled apoptotic corpses were incubated with a control secreted protein composed of the RFP protein fused to the signal peptide of the Viking protein (SP^vkg‐RFP) (Liu et al, 2017) (Fig 2A and D).

Figure 2. NimB4 binds to apoptotic cells ex vivo .

• A–C
  Representative confocal imaging of CFSE‐stained apoptotic bodies (green) incubated with secreted NimB4‐RFP (B, C) or SP^vkg‐RFP (A). Apoptotic bodies were pre‐incubated in absence (A, B) or presence (C) of Annexin V (25 μg/ml). Scale bar = 10 μm.
• D
  Quantification of the colocalization of NimB4‐RFP or SP^vkg‐RFP with the apoptotic bodies in presence or absence of Annexin V, as measured by Pearson's correlation coefficient. Values from at least four independent experiments are represented as mean ± SD (**P < 0.01 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).

Figure EV2. NimB4 binds apoptotic cells in vivo .

1. Representative confocal imaging of apoptotic bodies (blue) and S2 cells (blue) incubated with secreted NimB4‐RFP. The brightness of the image is modified to improve visualization of the signal. Scale bar = 10 μm.
2. Representative confocal imaging of imaginal wing disks from NimB4‐sGFP third‐instar larvae expressing or not the pro‐apoptotic gene in the imaginal wing disk (Ap > UAS‐reaper NimB4‐sGFP). Imaginal wing disks were stained with DAPI (blue). The binding of NimB4‐sGFP is delimited inside the dashed line corresponding to apterous‐Gal4 expression. Scale bar = 10 μm.
3. Representative confocal imaging of apoptotic bodies (blue) derived from Lpp > NimB4‐RFP or Lpp > SP^vkg‐RFP third‐instar larvae after clean injury. The brightness of the image is modified to improve visualization of the signal. Scale bar = 10 μm.
4. Quantification of the colocalization of NimB4‐RFP or SP^vkg‐RFP with the apoptotic bodies, as measured by Pearson's correlation coefficient. Values from at least four independent experiments are represented as mean ± SD (*P < 0.05, by Mann–Whitney test).

We then repeated the same experiment, but pre‐treated apoptotic cells with Annexin V, a protein that coats apoptotic cells by binding to PS (Engeland et al, 1998; Janko et al, 2013). Pre‐incubation of apoptotic corpses with Annexin V strongly reduced the binding of NimB4‐RFP to the apoptotic cells, supporting the notion that NimB4 binds to apoptotic corpses in a PS‐dependent manner (Fig 2C and D).

Next, we investigated if NimB4 can bind to apoptotic cells in vivo. For this, we induced apoptosis by expressing the apoptosis activator reaper in the imaginal wing disk using the driver Apterous‐Gal4. After induction of apoptosis, we observed the binding of NimB4‐sGFP to the region of the wing disk that expressed Apterous‐Gal4 (Fig EV2B). Additionally, we observed the binding of NimB4 to apoptotic cells released in the hemolymph after clean injury. No binding was observed between apoptotic corpses and the secreted protein SPvkg‐RFP (Fig EV2C and D). Collectively, our studies revealed that NimB4 has key features of a bridging molecule in that it is secreted and binds to apoptotic corpses in a PS‐dependent manner.

NimB4‐deficient animals accumulate apoptotic corpses during development

To investigate the role of NimB4 in vivo, we generated a null mutation in the NimB4 gene by CRISPR‐Cas9, referred to as NimB4^sk2 . The NimB4^sk2 mutant has a 14bp deletion in the second exon, inducing a stop codon at amino acid 113 (Fig EV3A–C). We isogenized the NimB4^sk2 mutation by backcrossing into the w¹¹¹⁸ DrosDel background. NimB4^sk2 homozygous flies were viable and did not show any striking morphological defect. We investigated the role of NimB4 in developmentally regulated apoptosis in the embryo and the larval brain (Kurant, 2011; Zheng et al, 2017). Stage 16 embryos were co‐stained with an anti‐SIMU antibody that marks macrophages, and anti‐Dcp‐1 that marks apoptotic cells by staining activated Death caspase‐1 (Dcp‐1). Macrophages of NimB4^sk2 mutants were normally localized throughout the embryo (Fig 3A). Interestingly, we noted an increased level of Dcp‐1‐positive apoptotic cells in NimB4^sk2 embryos. Embryonic macrophages of NimB4^sk2 but not wild‐type embryos appeared highly vacuolated and full of apoptotic corpses, a phenotype reminiscent of the draper^▵5 mutant (Kurant et al, 2008) (Fig 3A and B). This finding suggests that NimB4^sk2 phagocytes can engulf apoptotic corpses but are deficient in their degradation. H99 embryos do not show any developmentally induced apoptosis as they carry a small deficiency that removes the three pro‐apoptotic genes head involution defective (hid), reaper, and grim (Chen et al, 1996). We recombined the H99 deficiency and NimB4^sk2 to confirm that the vacuolization of NimB4^sk2 embryonic macrophages is indeed caused by the accumulation of apoptotic corpses. Consistent with this notion, NimB4‐deficient macrophages in the apoptosis‐null H99 mutant background did not show any vacuoles or Dcp‐1 staining (Fig 3A). This demonstrates that the vacuolization of NimB4^sk2 embryonic macrophages is linked to the accumulation of apoptotic corpses inside the cell.

Figure EV3. Generation of NimB4‐deficient flies by CRISPR‐Cas9 system.

1. Representation of the Nimrod gene locus on the 2^nd chromosome (cytology: 34E5). The NimB4^sk2 mutation is a 14 bp deletion in the first exon. Scheme adapted from Ramond et al, (2020).
2. The mutation of NimB4 gene induces a frameshift leading to a premature stop codon.
3. Experimental design of qPCR assay to validate NimB4^sk2 mutants. Scheme adapted from Ramond et al (2020).
4. Quantification of the distance (in cm) traveled in 1 min by wild‐type (w¹¹¹⁸), NimB4^sk2, draper^▵5, and NimC1¹ larvae (n = 60 per genotype). The result shown is a mean of three independent experiments containing two cohorts each (*P < 0.05, **P < 0.01, ***P < 0.001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
5. Forty days old wild‐type (w¹¹¹⁸), draper^▵5 , and NimB4^sk2 flies were tested using a standard climbing assay (mean ± SD), (n = 100 flies for each genotype), (*P < 0.05, ** P < 0.01, by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
6. Lifespan of NimB4^sk2 (red) and wild‐type (w¹¹¹⁸ , black) males kept on standard food at 25°C. The result shown is a mean of three independent experiments containing two cohorts of 20 male flies each (****P < 0.0001; log‐rank test).

Figure 3. The NimB4^sk2 mutant is defective in cell corpse clearance.

1. Representative confocal imaging of wild‐type (w¹¹¹⁸), NimB4^sk2 , and apoptosis‐null NimB4^sk2; Df(3L) H99 embryonic macrophages at stage 16 of development; ventral and lateral view; Tissues were co‐stained with Dcp‐1 (red, corresponding to apoptotic corpses) and anti‐SIMU (green, corresponding to macrophages) antibodies. The arrows show the presence of apoptotic cells inside embryonic macrophages. Scale bars = 20 μm.
2. Quantification of caspase‐positive particles (anti‐Dcp‐1, red) within macrophages (anti‐SIMU, green). Values from at least five independent experiments are represented as mean ± SD total volume of caspase‐positive particles (**P < 0.01, by Mann–Whitney test).
3. Representative projections from confocal stacks of entire third‐instar wild‐type (w¹¹¹⁸) and NimB4^sk2 larval brains stained with anti‐Draper (green) and anti‐Dcp‐1 (red). The arrows show the presence of apoptotic cells in the central area of NimB4^sk2 larval brain. Scale bars = 100 μm.
4. Quantification of caspase‐positive particles in the central brain area of larval brain. Values from at least five independent experiments are represented as mean ± SD total volume of caspase‐positive particles (**P < 0.01, by Mann–Whitney test).

Moreover, we observed that the brain of NimB4^sk2 larvae also displayed a greater amount of apoptotic cells than wild‐type larvae (Fig 3C and D). These data demonstrate that NimB4 is required for proper clearance of apoptotic cells during embryonic and larval brain development.

Defective clearance of apoptotic cells may lead to lethality and organ dysfunction (Henson & Hume, 2006). Similarly to draper^▵5 mutant, NimB4^sk2 third‐instar larvae displayed a crawling defect, with a drastic reduction of linear movement (Fuentes‐Medel et al, 2009) (Fig EV3D). In contrast, no crawling defect was observed in NimC1¹ deficient larvae lacking a phagocytic receptor involved in bacterial uptake (Melcarne et al, 2019b), indicating that this phenotype is specific to draper^▵5 and NimB4^sk2 mutants. At the adult stage, the draper^▵5 flies have a shortened lifespan and an age‐dependent locomotor defect (Draper et al, 2014).

A single‐cell transcriptomic atlas of the aging Drosophila brain (Davie et al, 2018) revealed that NimB4 is also expressed in the adult glial cells, suggesting its involvement in the phagocytosis. To evaluate the importance of NimB4 during aging, we monitored two phenotypes usually associated with neurodegeneration in NimB4^sk2 adult flies and found that similar to the draper^▵5 mutant, NimB4^sk2 flies displayed a climbing defect (suggesting declining motor activity), and reduced life span (Fig EV3E and F).

NimB4 enhances efferocytosis

Having shown that NimB4 is required for the elimination of apoptotic cells, we then investigated at which step of the efferocytosis process NimB4 is required, using macrophages of third‐instar larvae. Since NimB4 is secreted and binds to apoptotic corpses, we analyzed whether NimB4 contributes to the initial phase of uptake. To this end, we incubated macrophages and apoptotic cells on ice to inhibit the engulfment process, without altering binding to the phagocytic cell (Pearson et al, 2003). Macrophages of NimB4^sk2 larvae bound apoptotic corpses to the same extent as wild‐type cells (Fig 4A and B). Conversely, the number of apoptotic corpses bound to the cell membrane of draper^▵5 macrophages was reduced. Scanning electron microscopy experiments confirmed that apoptotic corpses bind to NimB4^sk2 but not draper^▵5 macrophages (Fig 4C). This indicates that, unlike the Draper receptor, NimB4 is not involved in the initial phagocytic step of binding.

Figure 4. Phagocytosis of apoptotic cells is reduced in NimB4^sk2 mutant.

1. Representative confocal imaging of wild‐type (w¹¹¹⁸), NimB4^sk2 , and draper^▵5 macrophages incubated with fluorescently labeled apoptotic cells (red, CellTrace™ Red) on ice for 1 h and stained with Alexa Fluor™ 488 phalloidin (green). Scale bar: 10 μm.
2. Quantification of the number of apoptotic cells (red) binding to the macrophages (green). Values from three independent experiments are represented as mean ± SD (***P < 0.001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
3. Representative Scanning Electron Microscopy images of spread macrophages extracted from wild‐type (w¹¹¹⁸), NimB4^sk2 , and draper^▵5 L3 wandering larvae and incubated 30 min with apoptotic cells (artificially colored in red) at room temperature. The arrows show the binding of apoptotic cells to the macrophages. Scale bar: 1 μm.
4. Ex vivo phagocytosis assay using Alexa555 fluorescent apoptotic bodies. Wild‐type, NimB4^sk2, NimB4 genomic rescue (NimB4^sk2, [NimB4]), draper^▵5 and NimC1¹, eater¹ macrophages from L3 wandering were incubated with Alexa555 fluorescent apoptotic bodies for 30, 60, or 120 min at room temperature. Phagocytosis was quantified by flow cytometry. Data are represented as mean ± SD from three independent experiments (*P < 0.05, **P < 0.01, ****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests ns: not significant).
5. Ex vivo phagocytosis assay using AlexFluor488 Staphylococcus aureus Bioparticles. Wild‐type, NimB4^sk2 , and NimC1¹, eater¹ macrophages from L3 wandering were incubated with bioparticles for 30, 60, or 120 min at room temperature. Phagocytosis was quantified by flow cytometry. Data are represented as mean ± SD from three independent experiments (****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).

We then investigated whether NimB4 is required for the subsequent step of phagocytosis, that is the engulfment of bound apoptotic corpses. For this, we tested the ability of NimB4^sk2 larval macrophages to internalize fluorescently labeled apoptotic bodies using an ex vivo phagocytic assay. Macrophages from wild‐type, NimB4^sk2 , draper^▵5 , and the double mutant NimC1¹, eater¹ were incubated for 30 min, 1 h, or 2 h with Alexa555 fluorescent apoptotic bodies, and flow cytometry was used to measure their phagocytic index. We found that, at 30 min, only the double mutant NimC1¹, eater¹ exhibited a reduced phagocytosis. At 1 h, we observed reduced phagocytosis in both NimC1¹, eater^,1 , and draper^▵5 mutants. Interestingly, the NimB4^sk2 mutant exhibited reduced phagocytosis of apoptotic corpses at a later time point (2 h) compared to the wild type (Fig 4D). Use of a genomic fragment encompassing the NimB4 gene rescued this decreased phagocytosis of the NimB4^sk2 mutant at 2 h (Fig 4D). Of note, NimC1¹; eater¹ mutants were unable to phagocytose apoptotic cells at all time points, indicating that the role of these two receptors is not restricted to bacterial phagocytosis.

To test if the role of NimB4 is specific to efferocytosis, we performed the same experiment with fluorescent AlexFluor488 Staphylococcus aureus Bioparticles. Interestingly, NimB4^sk2 macrophages retained the ability to phagocytose S. aureus bacteria at 2 h similar to the wild‐type macrophages (Fig 4E). We included macrophages from NimC1¹ and eater¹ larvae that show impaired phagocytosis of bacteria as a positive control (Melcarne et al, 2019b). We conclude that NimB4 is not specifically required for the binding of apoptotic corpses but enhances phagocytosis of apoptotic cells.

Loss of NimB4 inhibits phagosome maturation

The accumulation of apoptotic corpses in embryonic NimB4^sk2 macrophages in vivo indicates impaired phagocytosis in these cells, while a reduction in phagocytosis only at late time points suggests a defect in phagosome maturation. We therefore explored if the loss of NimB4 alters the phagosome maturation process. In these experiments, we included as controls draper^▵5 and croquemort^▵ mutants previously shown to have defective phagosome maturation (Kurant et al, 2008; Han et al, 2014). The intracellular vesicles were analyzed using the fluorochrome LysoTracker Red, which fluoresces in acidic compartments. We observed that NimB4^sk2 , draper^▵5, and croquemort^▵ but not eater¹ or NimC1¹ macrophages contained numerous and enlarged acidic vesicles compared to wild‐type macrophages (Figs 5A and B, and EV4A). To study possible NimB4 interactions with phagocytic receptors, we measured LysoTracker signal in macrophages double mutants for NimB4^sk2 and draper^▵5, croquemort^▵, eater¹, or NimC1¹ (Fig EV4A). We observed an increased LysoTracker signal in the NimB4^sk2; draper^▵5 mutant suggesting that Draper and NimB4 additively contribute to phagosome maturation. The NimB4^sk, croquemort^▵ double mutants did not show any increased LysoTracker signal compared to the single mutants. This result suggests that Croquemort and NimB4 might work together in the phagosome maturation process.

Figure 5. NimB4^sk2 mutant macrophages have an increased number of vesicles.

1. Representative fluorescence microscopy images of wild‐type (w¹¹¹⁸), NimB4^sk2 , draper^▵5 , eater¹ , and croquemort^▵ third‐instar larvae macrophages stained with LysoTracker Red (live imaging). Overlay of fluorescence and differential interference contrast microscopy (DIC). Scale bar = 10 μm.
2. Mean fluorescence intensity after staining with LysoTracker Red (live confocal imaging). Values from at least three independent experiments are represented as mean ± SD (**P < 0.01, ****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests).
3. Representative transmission electron micrographs of macrophages from wild‐type (top, w¹¹¹⁸) and NimB4^sk2 (bottom) L3 wandering larvae. Scale bar: 2 μm.
4. Quantification of the number of vesicles per macrophages in the electron micrographs. Values from at least three independent experiments are represented as mean ± SD (****P < 0.0001 by Mann–Whitney test).
5. Representative confocal imaging of Rab7^EYFP immunostaining in wild‐type (w¹¹¹⁸) and NimB4^sk2 macrophages. Tissues were stained with anti‐GFP (red, Rab7), counterstained with phalloidin (gray) and DAPI (blue). The arrows indicate the enlarged vesicles are decorated with Rab7^EYFP in the NimB4^sk2 macrophages. Scale bar = 10 μm.
6. Quantification of the diameter of the vesicles. Values from at least three independent experiments are represented as mean ± SD (****P < 0.0001 by Mann–Whitney test).

Figure EV4. Genomic rescue of the NimB4^sk2 mutant.

1. Mean fluorescence intensity after staining with LysoTracker Red (confocal live imaging). Values from at least three independent experiments are represented as mean ± SD (*P < 0.05, ***P < 0.001 ****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
2. Representative fluorescence microscopy images of wild‐type (w¹¹¹⁸ ), NimB4^sk2, and NimB4 genomic rescue (NimB4^sk2, [NimB4]) macrophages stained with LysoTracker Red probe (live imaging). Lower panels show an overlay of fluorescence and DIC. Scale bar = 10 μm.
3. Mean fluorescence intensity after staining with LysoTracker Red (confocal live imaging). Values from at least three independent experiments are represented as mean ± SD (****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
4. Representative fluorescence microscopy images of wild‐type macrophages (HmlΔ>+) and macrophages expressing NimB4‐RNAi (HmlΔ > NimB4‐IR) stained with LysoTracker Red probe (live imaging). Lower panels show an overlay of fluorescence and DIC. Scale bar = 10 μm.
5. Mean fluorescence intensity after staining with LysoTracker Red (confocal live imaging). Values from at least three independent experiments are represented as mean ± SD (****P < 0.0001 by Mann–Whitney test).
6. Representative transmission electron micrographs of macrophages from control macrophages (HmlΔ>+) and macrophages expressing NimB4‐RNAi (HmlΔ > NimB4‐IR) from L3 wandering larvae. Scale bar: 2 μm.
7. Quantification of the number of vacuoles per macrophages. Values from at least three independent experiments are represented as mean ± SD (**P < 0.01, by Mann–Whitney test).
8. Representative fluorescence microscopy images of macrophages from wild‐type (HmlΔ^ts >+) or macrophages expressing NimB4‐IR (HmlΔ^ts  > NimB4‐IR) using a temperature‐inducible macrophage driver HmlΔ^ts to restrict the expression of NimB4‐IR after the second‐instar larval (L2) stage. HmlΔ^ts  > KK NimB4‐IR (29°C) larvae developed at 29°C whereas HmlΔ^ts  > KK NimB4‐IR (18–29°C) larvae were shifted from 18 to 29°C at the second‐instar larval stage. Macrophages were stained with LysoTracker Red probe. Lower panels show an overlay of fluorescence and DIC. Scale bar = 10 μm.
9. Mean fluorescence intensity after staining with LysoTracker Red (confocal live imaging). Values from at least three independent experiments are represented as mean ± SD (****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).

The accumulation of intracellular vesicles in NimB4^sk2 macrophages was rescued by complementing the mutant with a transgene containing a wild‐type copy of NimB4 gene (Fig EV4B and C). In addition, silencing NimB4 using RNAi with the Hml‐Gal4 macrophage driver reproduced this phenotype, confirming that the accumulation of intracellular vesicles was indeed caused by the inactivation of NimB4 in macrophages (Fig EV4D and E). We next analyzed the subcellular morphology of NimB4^sk2 intracellular vesicles by transmission electron microscopy. NimB4^sk2 but not wild‐type macrophages were filled with large intracellular vesicles that occupied most of the cell volume. These vesicles were surrounded by a single lipid bilayer and had a clear lumen (Fig 5C and D). A similar vesicle accumulation was also observed upon silencing of NimB4 by RNAi (Fig EV4F and G). To confirm that this phenotype was not linked to a defect due to loss of NimB4 during development, we used a temperature‐inducible macrophages driver (HmlΔ^ts) to express NimB4‐RNAi after the second‐instar larval (L2) stage. LysoTracker staining showed that hemocytes in which NimB4 was inactivated after the L2 stage also showed accumulation of acidic vesicles (Fig EV4H and I). Therefore, we conclude that the accumulation of acidic vesicles in NimB4 macrophages is not caused by a defect at the early stages of development.

NimB4 plays a crucial role in phagosome–lysosomes fusion

Phagosome maturation culminates in the fusion of the phagosome with lysosomes, leading to the formation of an acidic phagolysosome. The formation of phagolysosomes is a necessary step to ensure efficient digestion of apoptotic corpses (Zhou & Yu, 2008). During phagosome maturation, the transition from early phagosome to late phagosome is marked by the conversion of Rab5 to Rab7. In addition, the phagosome acquires Lamp1, which is required for phagolysosome fusion (Huynh et al, 2007). Our observation that accumulated vesicles in NimB4^sk2 macrophages are LysoTracker‐positive and have therefore initiated acidification suggests that NimB4 is required at a later stage of phagosome maturation. To determine more precisely the step at which maturation of NimB4^sk2 phagosomes is blocked, we tested whether Rab7, a marker of late phagosomes, was enriched on the vesicular membrane of NimB4^sk2 macrophages. We observed that the enlarged vesicles of NimB4^sk2 macrophages were positive for Rab7^EYFP, confirming that phagosomes are blocked at a late stage of maturation (Fig 5E and F). To determine whether fusion between the phagosome and lysosomes requires NimB4, we analyzed the fate of wild‐type, NimB4^sk2 , and draper^▵5 macrophages stained with the phagosome marker LysoTracker and the lysosome marker Lamp1‐mcherry. In wild‐type macrophages, these markers colocalized, indicating correct formation of the phagolysosomes. In contrast, we found no colocalization of LysoTracker and Lamp1‐mcherry in the NimB4^{sk 2} and draper^▵5 macrophages (Fig 6A and B). The Lamp1‐mcherry lysosomal signals appeared as small punctate structures between the LysoTracker‐positive vacuoles. This suggests that a loss of NimB4 or draper impairs the fusion rather than the clustering of phagosomes with lysosomes (Fig 6A and B). An increased Lamp1‐mcherry fluorescent signal in NimB4^sk2 mutant macrophages indicated that the blockage of phagolysosome fusion was not due to a reduced number of lysosomes in the NimB4^sk2 mutant. Interestingly, we observed a similar increase in the number of Lamp1‐mcherry‐positive structures when wild‐type macrophages were pre‐treated with chloroquine, a drug that inhibits endosomal acidification, which is used here to block phagosome maturation (Fig EV5A and B).

Figure 6. Loss of NimB4 blocks the phagosome maturation process by impairing the phagosome‐lysosome fusion.

1. Representative confocal imaging of localization of Lamp1‐mcherry and LysoTracker Green in wild‐type (w¹¹¹⁸), NimB4 ^sk2, and draper^▵5 macrophages (live imaging). The arrows indicate the colocalization (w¹¹¹⁸ ) or the clustering (NimB4^sk2 and draper^▵5 ) of Lamp1‐mcherry and LysoTracker Green signals. Scale bar: 10 μm.
2. Quantification of the colocalization of Lamp1‐mcherry with LysoTracker Green, as measured by Pearson's correlation coefficient between the two signals. Values from at least five independent experiments are represented as mean ± SD (*P < 0.05 by ANOVA test followed by post hoc Dunnett's multiple comparison tests).
3. Ex vivo phagocytosis assay using apoptotic cells labeled with pHrodo™ Red. Wild‐type (w¹¹¹⁸ ), NimB4^sk2, and draper^▵5 macrophages from L3 wandering larvae were incubated with pHrodo™ Red apoptotic cells for 60 min at room temperature. Phagocytosis was quantified by flow cytometry. Data are represented as mean ± SD from three independent experiments (*P < 0.05 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).
4. Ex vivo phagocytosis assay using pHrodo™ Red Staphylococcus aureus Bioparticles™ conjugates. Wild‐type, NimB4^sk2 and NimC1¹, eater¹ HmlΔ‐Gal4 > UAS‐GFP macrophages from L3 wandering larvae were incubated with pHrodo™ Red S. aureus Bioparticles™ for 30, 60, or 120 min at room temperature. Phagocytosis was quantified by flow cytometry. Data are represented as mean ± SD from three independent experiments (*P < 0.05 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).

Figure EV5. Elevated Lamp1 levels in the NimB4^SK2 macrophages.

1. Confocal imaging of wild‐type and NimB4 ^sk2 larval macrophages expressing the reporter Lamp1‐mcherry. Tissues were stained with an anti‐mCherry antibody (red, corresponding to the Lamp1 signal) and DAPI (blue). Wild‐type macrophages (w¹¹¹⁸ ) were incubated in the presence or in the absence of Chloroquine (50 μg/ml) for 40 min. Scale bar = 10 μm.
2. Mean fluorescence intensity after staining with LysoTracker Red. Values from at least three independent experiments are represented as mean ± SD (***P < 0.001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).

The fusion of the phagosome with the lysosome allows the acidification of the phagosome. We therefore assessed the acidification of phagosomes in both NimB4^sk2, draper^▵5, and wild‐type macrophages. For this, we used apoptotic cells labeled with pHrodo‐succinimidyl ester (SE), a pH‐sensitive fluorescent dye. We then assessed the acidification of the phagosome measuring the mean fluorescence intensity at 1 h after incubation as the NimB4^sk2 mutant did not show phagocytosis defect at this time point. We observed by flow cytometry a reduced fluorescence intensity in the NimB4^sk2 and the draper^▵5 mutant compared to wild type consistent with defective acidification of the phagosome (Fig 6C).

We repeated the same experiment with pHrodo bioparticles labeled with S. aureus as the NimB4^sk2 mutant previously showed no defect in the phagocytosis of S. aureus Alexa Fluor 488 BioParticles (Fig 6D). We observed by flow cytometry a reduced fluorescence intensity of pHrodo‐labeled S. aureus bioparticles in the NimB4^{sk 2} mutant macrophages compared to wild type (Fig 6D) consistent with defective acidification of the phagosome.

It has been shown that overexpression of Rab7 efficiently promotes the phagosome‐lysosome fusion (Harrison et al, 2003). We therefore investigated whether increased expression of Rab7 could rescue the phagosome‐lysosome fusion defect of NimB4^{sk 2} macrophages. Overexpression of Rab7, but not Rab5, strongly reduced the number of accumulated vacuoles in NimB4^{sk 2} macrophages, showing that promoting the phagosome‐lysosome fusion is sufficient to rescue the NimB4^{sk 2} cells (Fig 7A–C). In this experiment, we found that the overexpression of Rab5 in wild‐type macrophage has a pronounced defect on its own as previously described (Bucci et al, 1992). We conclude that NimB4 is required for the late stage of phagosome maturation and more specifically, for the fusion of the phagosome and lysosome.

Figure 7. The increased expression of RAB7 rescues the phagosomes accumulation in NimB4^sk2 macrophages.

• A–B
  Representative fluorescence microscopy images of wild‐type (A) or NimB4^sk2 (B) macrophages expressing Rab5 (middle panel) or Rab7 (right panel) driven by HmlΔ‐Gal4 and stained with the LysoTracker Red (live imaging). Overlay of fluorescence and DIC. Scale bar = 10 μm.
• C
  Quantification of the mean fluorescence intensity of LysoTracker in macrophages from larvae (confocal live imaging). Values from at least three independent experiments are represented as mean ± SD (**P < 0.01, ****P < 0.0001 by ANOVA test followed by post hoc Dunnett's multiple comparison tests. ns: not significant).

Discussion

The clearance of apoptotic cells by phagocytes is a critical event during the development of all multicellular organisms. Failure in this process can lead to autoimmune or neurodegenerative diseases. In this study we identified a secreted protein of the Nimrod family, NimB4, that binds to apoptotic corpses in a PS‐dependent manner to promote efferocytosis. Similar to mutants of the phagocytic receptor Draper, the NimB4^sk2 mutant accumulates apoptotic cells in both embryonic and larval macrophages. Further functional studies reveal its role in phagosome maturation, notably the formation of the phagolysosome. Our study identifies the first secreted factor involved in the phagocytosis of apoptotic cells in Drosophila, resembling in many respects the bridging molecules found in mammals.

NimB4 is a secreted factor that promotes the phagocytosis of apoptotic cells

NimB4 was initially identified as a secreted member of the Nimrod family of unknown function. Our results show that NimB4 is expressed in macrophages and glial cells, and that its expression is induced upon injury, a stimulus associated with increased apoptosis (Steller, 2008). In mammals, the expression of the gene encoding the bridging molecule MFG‐E8 is regulated by a “find me factor” released by apoptotic cells (Miksa et al, 2007). Further study should identify the signaling pathway regulating NimB4 upon injury and test whether Drosophila “find me factor” are involved in this process (Ravichandran, 2010, 2011).

We also observed that NimB4 is secreted by macrophages and that it is likely associated with the endosomal compartments intracellularly. NimB4 mutants display a shorter lifespan and a locomotor defect that can be causally linked to defective apoptotic cell clearance during development and aging. These phenotypes are very similar to those previously described for draper mutants, albeit slightly weaker (Kurant et al, 2008; Draper et al, 2014; Etchegaray et al, 2016).

Our data also indicate that NimB4 binds to apoptotic corpses in a PS‐dependent manner. As a member of the Nimrod family, NimB4 shares homology with SIMU and Draper, two phagocytic receptors that also interact with PS (Shklyar et al, 2013a; Tung et al, 2013). This suggests that the common EGF/Nimrod domain found in Nimrod family members may contribute to binding PS. However, it is unclear whether NimB4 binds to apoptotic cells alone and interacts directly with receptors involved in efferocytosis, or whether other co‐factors are required. It is worth noting that the Drosophila genome encodes several other secreted NimB proteins such as NimB1, NimB2, and NimB3 that are not yet characterized (Somogyi et al, 2008). It is tempting to speculate that some of these proteins could work synergistically with NimB4 in the process of efferocytosis. The similarities between the NimB4^sk2 and draper^▵5 mutant phenotypes suggest that both proteins function in a similar efferocytosis pathway. In this study, we did not find any evidence that NimB4 interacts with Draper. Future studies should address whether NimB4 establishes a direct bridge between apoptotic corpses and phagocytic receptors.

NimB4 is specifically required for the phagocytosis of apoptotic cells

NimB4 is required for the phagocytosis of apoptotic cells but not bacteria. This indicates that it is not a core component of the phagocytic machinery. This result is in line with many studies that show that the phagocytic process is multi‐faceted and is tailored to the nature of the ingested particle. For instance, studies have shown that phagosomes containing apoptotic cells mature faster than those containing opsonized viable cells (Erwig et al, 2006). These observations indicate that phagocytic targets can differentially affect the maturation rate, perhaps through the recruitment of different phagocytic receptors. In contrast to NimB4, Draper contributes to the phagocytosis of both bacteria (S. aureus) and apoptotic cells (Manaka et al, 2004b; Shiratsuchi et al, 2012), pointing to more versatile functions. In this study, we incidentally showed that NimC1 and Eater, two immune receptors implicated in the phagocytosis of microbes, are also required for the efferocytosis process. In contrast, the role of SIMU seems to be restricted to apoptotic cell clearance during embryonic and early adult phase development (Kurant et al, 2008). These authors speculated that the presence of SIMU reinforces uptake of apoptotic cells at critical developmental stages characterized by massive apoptosis. We hypothesize that, like SIMU, NimB4 may have a more specific function related to a particular efferocytosis program.

The secreted NimB4 participates in phagosome maturation

Importantly, our study shows that secreted NimB4 binds to apoptotic cells and is an essential component of the machinery that promotes efferocytosis. Thus, as observed in mammals and C. elegans, our study reveals that secreted factors contribute to the apoptotic cell clearance in Drosophila. However, NimB4 is not required for macrophages binding to apoptotic bodies. Instead, apoptotic corpse uptake is reduced only at late time points. We hypothesize that this phenotype is a secondary consequence of a defect in phagosome maturation. The accumulation of immature phagosomes that are not properly eliminated would indirectly impair the uptake of new apoptotic corpses. Efferocytosis would then be impaired only at late time points when the accumulation of phagosomes reaches a threshold preventing further phagocytosis. Phagosome maturation is punctuated by two main events: the Rab5‐Rab7 conversion and the fusion between the phagosome and the lysosome (Vieira et al, 2002; Kinchen & Ravichandran, 2008; Akbar et al, 2011; Pauwels et al, 2017; Pradhan et al, 2019). Interestingly, overexpression of Rab7 rescues the phagosome maturation blockage of NimB4 mutant. Our study suggests that NimB4 is required late in the process, at the step of phagosome‐lysosome fusion. Interestingly, the receptors Draper and Croquemort, initially thought to be involved in the early phase of phagocytosis, were later shown to be critical for phagosome maturation (Kurant et al, 2008; Han et al, 2014). How those receptors and NimB4 contribute to the maturation of phagosome is currently unknown. We speculate that the binding of NimB4 to apoptotic cells could orient the maturation process; that is, apoptotic corpses may require specific “digest me” signals similar to the “eat‐me” signals that are required for uptake. Several mechanisms could explain how a secreted factor can impact phagosome maturation. Binding of NimB4 to apoptotic cells could cluster phagocytic receptors or activate a specific phagocytic receptor that initiates a dedicated digestion program affecting the degradation speed of ingested apoptotic cells. Alternatively, NimB4 could promote phagocytosis and lysosome maturation independently by interacting with different partners in the phagocytic process.

Is NimB4 a bridging molecule?

Bridging molecules recognize “eat‐me” signals on apoptotic cells facilitating their uptake by phagocytic receptors (Ravichandran, 2010). In mammals, several soluble bridging molecules such as MFG‐E8, Gas6, and protein S mediate recognition of apoptotic cells by cross‐linking the PS “eat‐me” signal with specific phagocytic receptors. For instance, MFG‐E8 binds to the integrin receptor of phagocytic cells (Hanayama et al, 2002; Akakura et al, 2004; Nandrot et al, 2007). In C. elegans, the protein TTR‐52 bridges the surface‐exposed PS on apoptotic cells and the CED‐1 receptor on phagocytes, which mediates recognition and engulfment of apoptotic cells (Wang et al, 2010). While the initial concept viewed bridging molecules as opsonins dedicated to the uptake of apoptotic cells, recent studies suggest greater complexity. Similar to NimB4, subsequent studies showed that MFG‐E8 is required in the maturation phase of phagocytosis as well as in the recognition step. While apoptotic cells ingested by wild‐type dendritic cells rapidly fused with lysosomes, in dendritic cells deficient for MFG‐E8 smaller fragments of apoptotic cells persisted in endosomes (Peng & Elkon, 2011). NimB4 shares key features of bridging molecules, in that it is secreted by phagocytic cells, binds to apoptotic cells in the PS‐dependent manner, and is specifically required for efficient efferocytosis.

Although this study has not conclusively demonstrated that NimB4 interacts with phagocytic receptors as mammalian bridging molecules do (Boada‐Romero et al, 2020), our results reveal the importance of secreted factors that enhance efferocytosis and reinforce the notion that “bridging molecules” could play a broader role than initially thought, in orienting phagosome maturation. Future studies should characterize how secreted factors promote and direct the process of efferocytosis, and Drosophila offers a powerful system to address these questions.

Materials and Methods

Reagents and Tools table

Reagent/Resource	Reference or Source	Identifier or Catalog Number
Experimental models
DrosDel w1118 isogenic (wild type)	Gift from Luis Teixeria (Pais, 2018)	N/A
w1118; NimB4 sk2	This study	N/A
wiso; NimB4sk2 (isogenized in the DrosDel background)	This study	N/A
w; UAS‐NimB4 IR	VDRC	106392
w, UAS‐NimB4	This study	N/A
w; UAS‐NimB4‐RFP	This study	N/A
w;; UAS‐NimB4‐HA	This study	N/A
w; NimB4 sk2, UAS‐NimB4‐RFP	This study	N/A
w; NimB4 sk2; UAS‐NimB4‐HA	This study	N/A
w; NimB4 sk2, [NimB4]	This study	N/A
w;; NimB4‐sGFP‐V5 (Injection in VK33 attP docking site)	This study	pFLYFOS 8882336809946705
w1118;; eater1	Bretscher et al (2015)	N/A
w1118; NimC11	Melcarne et al (2019b)	N/A
draperΔ5	Gift from M. Freeman (Freeman, 2015)	N/A
; croquemort ▵	Gift from Yuh‐Nung jang (Han et al, 2014)	N/A
w1118; NimC11; eater1	Melcarne et al (2019b)	N/A
w; HmlΔ‐Gal4, UAS‐GFP;	Sinenko and Mathey‐Prevot (2004)	N/A
w;; HmlΔ‐Gal4	Sinenko and Mathey‐Prevot (2004)	N/A
; Tub‐gal80ts, UAS‐GFP; HmlΔ‐Gal4	This study	N/A
w; HmlΔ ‐Gal4, UAS‐GFP, NimB4 sk2	This study	N/A
w1118; NimC1, HmlΔGal4,UAS‐GFP;eater1	Melcarne et al (2019b)	N/A
w; NimB4sk2; HmlΔ‐Gal4	This study	N/A
; Tub‐gal80ts,UAS‐GFP; Lpp‐Gal4/TM3 (Referred to as Lppts )	Brankatschk et al (2018)	N/A
;; Lpp‐Gal4/TM3	N/A	N/A
; NimB4sk2; Lpp‐Gal4/TM3	This study	N/A
; Act5c‐gal4/cyoGFP	Bretscher et al (2015)	N/A
; UAS‐SPvkg‐RFP.3.1; (30 first amino acids of Viking protein followed by RFP)	Gift from Pastor‐Parja I (Liu et al, 2017)	N/A
;; Df(3L)H99, kniri‐1 pp /TM3, Sb1	DGGR	106395
; NimB4sk2 ; Df(3L)H99, kniri‐1 pp /TM3, Sb1	This study	N/A
; tubGal80ts; repoGal4	Hakim‐Mishnaevski et al (2019)	N/A
w1118;; TI{TI}Rab7[EYFP]	BDSC	62545
w1118; NimB4sk2; TI{TI}Rab7[EYFP]	This study	N/A
yw; UAS‐Rab5‐YFP	Bloomington	9771
yw; UAS‐Rab5‐YFP, NimB4sk2	This study	N/A
; UAS‐venus‐rab7	Gift from Brian McCabe	N/A
; UAS‐venus‐rab7, NimB4 sk2	This study	N/A
;; Lamp1–mCherry	Gift from Gábor Juhász (Lőw et al, 2013)	N/A
; NimB4sk2; Lamp1–mCherry	This study	N/A
Iso; iso; TM2/TM6	Gift from Luis Teixeira	N/A
FM7; iso; iso	Gift from Luis Teixeira	N/A
Iso; Gla/cyo; iso	Gift from Luis Teixeira	N/A
Recombinant DNA
pENTR/D‐TOPO	Invitrogen	K240020
pTW	DGRC	1129
pTWR	DGRC	1136
pOT2‐CG16873 Full length cDNA clone	DGRC	IP09831
pCaSpeR4	Our collection	N/A
BACR09N24	BACPAC Resources Center	RP98‐9N24
Antibodies
Chicken anti‐GFP	Abcam	Cat# ab13970
Rabbit monoclonal anti‐SIMU antibody	Shklyar et al (2013b)	N/A
Mouse monoclonal anti‐Draper antibody	Kurant et al (2008)	N/A
Rabbit anti Dcp‐1 antibody	Cell Signaling	Cat # 9578
Mouse monoclonal anti mCherry antibody	Invitrogen	Cat #M11217
Goat anti‐Mouse IgG (H+L) Secondary Antibody, HRP	Jackson ImmunoResearch labs	Cat# 31430
Goat anti‐Chiken IgY (H+L) Secondary antibody, Alexa Fluor 488	Thermo Fisher Scientific	Cat# A11039
Goat anti‐Mouse IgG (H+L) Highly Cross‐adsorbed Secondary Antibody, Alexa Fluor 555	Thermo Fisher Scientific	Cat# A21424
Goat anti‐Rabbit IgG (H+L) Highly Cross‐adsorbed Secondary Antibody, Alexa Fluor 555	Thermo Fisher Scientific	Cat# A32731
Oligonucleotides and sequence‐based reagents
NimB4 F deletion screen: CGA TCT CTG TGC CTC CAC TC	Microsynth	N/A
NimB4 R deletion screen CTT GCA ACA AAC CTC GAT CA	Microsynth	N/A
NimB4 deletion screen CCC AAC GGT TTG TCC GGA ACA TC	Microsynth	N/A
NimB1 qRT F CGG CCA AAG TGT GAG AGA TT	Microsynth	N/A
NimB1 qRT R TAT CGT CAC AGC TTC CGT TC	Microsynth	N/A
NimB2 qRT F GAG TGT CTG CCG AAG TGT GA	Microsynth	N/A
NimB2 qRT R TCA CAT ATC CGG TCT TGC AG	Microsynth	N/A
NimB3 qRT F TCC CAA CTC CAG AAA TCG TC	Microsynth	N/A
NimB3 qRT R AGC AGT CCT CCG AGC AAA T	Microsynth	N/A
NimB4 qRT F TTG TGC TCA ACT ACC GCA AC	Microsynth	N/A
NimB4 qRT R CGT CCA GCT CGT ATC CCT TA	Microsynth	N/A
NimB5 qRT F CGT AAC GAC AAC GGT GAC TG	Microsynth	N/A
NimB5 qRT R GTC TCG TCC AGC TTG TAG CC	Microsynth	N/A
RpL32 qRT F GAC GCT TCA AGG GAC AGT ATC TG	Microsynth	N/A
RpL32 qRT R AAA CGC GGT TCT GCA TGA G	Microsynth	N/A
NimB4_gateway cloning F CAC CAT GTC AAC AAT ACT GCG A	Microsynth	N/A
NimB4_gateway cloning R CAA CCA CCA CAT ATC GCT GAG	Microsynth	N/A
Chemicals, enzyme and other reagents
PrimeScript RT	TAKARA	Cat# RR037B
CFSE (5(6)‐CFDA/SE	Thermo Fisher Scientific	Cat# C34554
CellTracker™ Red CMTPX dye	Thermo Fisher Scientific	Cat# C34572
pHrodo™ Red, succinimidyl ester	Invitrogen	Cat# P3660
Triton‐X‐114 (10% solution)	Thermo Scientific	Cat# 28332
Complete protease inhibitor	Sigma	11697498001
4’,6‐Diamidino‐2‐phenylindole dihydrochloride (DAPI)	Sigma	Cat# D9542
AlexaFluor 488 Phalloidin	Life Technologies	Cat# A22287
AlexaFluor 555 Phalloidin	Life Technologies	Cat# A34055
LysoTracker® Red DND‐99	Thermo Fisher Scientific	Cat# L7528
LysoTracker® Green DND‐26	Thermo Fisher Scientific	Cat# L7526
phenylmethylsulfonyl fluoride	Sigma	P7626
AlexFluor™ 488 S. aureus Bioparticles™	Thermo Fisher Scientific	Cat# s23371
pHrodo™ Red S. aureus Bioparticles™	Thermo Fisher Scientific	Cat# A10010
Software
Prism 5	GraphPad Prism	N/A
Fiji 2.1.0/1.53c	Image J	N/A
Others
Tecan Infinite M200	Tecan	N/A
Leica SP8 IN1	Leica	N/A
The Zeiss AxioImager Z1	Zeiss	N/A

Methods and Protocols

Drosophila rearing conditions

All Drosophila stocks were maintained at 25°C on standard fly medium consisting of 6% cornmeal, 6% yeast, 0.6% agar, 0.1% fruit juice (consisting of 50% grape juice and 50% multifruit juice), supplemented with 10.6 g/l moldex and 4.9 ml/l propionic acid. Third‐instar (L3) wandering larvae were selected at 110–120 h AEL.

Mutant and transgenic lines generation

NimB4^sk2 flies were generated using the CRISPR/Cas9 technique as previously described (Kondo & Ueda, 2013). Briefly, a transgenic fly line expressing Cas9 protein using the germline‐specific nanos promoter was crossed to a line expressing a custom guide RNA (gRNA). The cross‐produces offspring with an active Cas9–gRNA complex specifically in germ cells, which cleaves and mutates the genomic target site. The following gRNA sequence was used: GGTTTGTCGCTGCCCCGGAG. To avoid any background effects, we introgressed NimB4^sk2 mutant into the w¹¹¹⁸ DrosDel isogenic background for seven generations. For complementation studies, the genomic region of NimB4 was amplified from genomic DNA including 2 kb upstream and 2 kb downstream of the NimB4 coding sequence. Gibson assembly was used to clone the fragment into a pUAST‐attB‐GFP‐V5‐His backbone (Rauskolb et al, 2011). The NimB5 and NimB3 genes present in the upstream and downstream sequence of NimB4, respectively, were inactivated by directed mutagenesis. Mutagenic primers were designed to delete the G of the ATG start codon of NimB3 and to add GA in position 44 of NimB5 to generate non‐sense coding sequences. The plasmid was then microinjected in the VK33 attP embryos. For overexpression studies, the genomic region from the 5′UTR to the stop codon of the intron‐less NimB4 gene was amplified from BACR09N24 and cloned into the pDONR207 Gateway vector (Invitrogen, Carlsbad, CA, USA) and subcloned in the pTW (Drosophila Genomics Resource Center plasmid) transgenesis vector and used to generate transgenic UAS‐NimB4 flies. For the UAS‐NimB4‐RFP, the NimB4 cDNA sequence without STOP codon was cloned into the entry vector pENTR/D‐ Topo (Invitrogen) and subsequently shuttled into the RFP expression vectors pTWR (C‐terminal RFP tag), obtained from the DGRC Drosophila Gateway vector collection. Plasmids were injected either at the Fly facility platform of Clermont‐Ferrand (France) or by BestGene Inc. (Chino Hills, CA, USA).

RT–qPCR experiments

For quantification of mRNA, whole third‐instar larvae (n = 8) or dissected tissues (n = 20–40) were isolated by TRIzol reagent and dissolved in RNase‐free water. 500ng total RNA was then reverse‐transcribed in 10 ml reaction volume using PrimeScript RT (TAKARA) and a mixture of oligo‐dT and random hexamer primers. Quantitative PCR was performed on cDNA samples on a LightCycler 480 (Roche) in 96‐well plates using the LightCycler 480 SYBR Green I master mix (Roche Diagnostics, Basel, Switzerland). Expression values were normalized to that of RpL32.

Western blot

Hemolymph samples were collected as follows: Forty L3 larvae were bled on a glass slide on ice; hemolymph was recovered by dissection, mixed with 10 μl of PBS supplemented with complete protease inhibitor solution (Roche) and 1mM phenylmethylsulfonyl fluoride (Sigma) and N‐Phenylthiourea (Sigma), and then centrifuged for 10 min at 1,000 g, 4°C. This was followed by a second centrifugation 5 min 10,000 g. Protein concentration of the samples was determined by BCA assay, and 40 μg of protein extract was separated on a 4–12% acrylamide precast Novex NuPage gel (Invitrogen) under reducing conditions and transferred to membranes (Invitrogen iBlot 2). After blocking in 5% non‐fat dry milk in PBS containing 0.1% for 1 h, membranes were incubated at 4°C overnight with a mouse anti‐RFP antibody (Abcam) in a 1:1,000 dilution. Anti‐mouse‐HRP secondary antibody (Jackson ImmunoResearch) in a 1:15,000 dilution was incubated for 45 min at room temperature. Bound antibody was detected using ECL (GE Healthcare) according to the manufacturer's instructions. Membranes were imaged on a ChemiDoc XRS+ (Bio‐Rad).

Apoptotic cells preparation

The S2 cells were cultured in Schneider's insect medium (Sigma‐Aldrich) containing 10% FBS (Gibco™), penicillin (Sigma‐Aldrich), and streptomycin (Sigma‐Aldrich) at a concentration of 100 U/ml. To induce apoptosis, cycloheximide (CHX, Sigma‐Aldrich) was added at a final concentration of 50 μg/ml. 24 h after the cycloheximide treatment, the cells were isolated and removed by pelleting with centrifugation at 400 g for 5 min at 4°C. In order to stain the apoptotic bodies, CFSE (5(6)‐CFDA/SE, Molecular Probes™, or CellTracker™ Red CMTPX dye was added to the supernatant harvested from the cells at a final concentration of 5 μM and incubated 15 min at room temperature in the dark.

pHrodo staining of apoptotic cells

Apoptotic cells were labeled with the pH‐sensitive stain pHrodo™ Red, succinimidyl ester (Invitrogen), which is nonfluorescent at neutral pH and emits a strong red fluorescence (532 nm) in an acidic environment (pH 4–6). After the induction of apoptosis by cycloheximide treatment, apoptotic cells were washed twice with PBS and resuspended in PBS at 10⁶ cells/ml. 1 µl of 1 mg/ml pHrodo‐SE (stock solution in DMSO) was added to 50 ml of cell suspension. After incubation for 30 min at RT, cells were washed twice with PBS and resuspended in PBS.

Immunohistochemistry

For immunofluorescence, L3 larvae were dissected into 150 μl PBS pH 7.4, and macrophages were allowed to adhere on a glass slide for 40 min and fixed for 10 min in PBS containing 4% paraformaldehyde. Larval tissues were dissected in PBS and fixed for at least 1 h at room temperature in 4% paraformaldehyde in PBS. For immunostaining, fixed tissues were subsequently rinsed in PBS + 0.1% Triton X‐100 (PBT), permeabilized, and blocked in PBT + 2% bovine serum albumin (BSA) for 1 h and incubated with primary antibodies in PBT + 2% BSA overnight at 4°C. After 1 h washing, secondary antibodies and DAPI were applied at room temperature for 2 h. Primary antibodies used are as follows: chicken anti‐GFP (Abcam, 1:1,000), rabbit anti‐SIMU 1:100 (Shklyar et al, 2013b)), mouse anti‐Draper (Developmental Studies Hybridoma Bank, 1:100), rabbit anti‐Dcp‐1(cell Signaling, 1:100), Mouse anti‐mCherry (Invitrogen, 1:1,000). Alexa Fluor 488 and Alexa Fluor 555‐conjugated secondary antibodies (Life technologies, 1:100) were used. For the chloroquine experiment, macrophages were dissected into 150 μl PBS pH 7.4. with 50 μg/ml of chloroquine and allowed to adhere on a glass slide in this solution for 40 min. Cells were then fixed with PFA 4% and immunostained as described above. Finally, cells were stained with 1/15,000 dilution of DAPI (Sigma‐Aldrich) and mounted in Dako fluorescence media. Imaginal wing disks were dissected in PBS and then fixed for 15 min at room temperature in 4% PFA in PBS. Samples were rinsed twice with PBS, stained with 1/15,000 dilution of DAPI (Sigma‐Aldrich), and mounted in Dako fluorescence media.

Macrophages LysoTracker red staining

Macrophages were allowed to adhere on slides for 45 min and then incubated with 1 μM LysoTracker® Red DND‐99 (Invitrogen™, L7528) or LysoTracker® Green DND‐26 (Invitrogen™, L7526) in PBS for 1 min at RT. The samples were washed twice in PBS and mounted for immediate observations under fluorescence or confocal microscope.

Scanning electron microscopy

Samples for SEM were prepared as follows. Six wandering third‐instar larvae were bled into 50 μl of Schneider's insect medium (Sigma‐Aldrich) containing 1 μM phenylthiourea (PTU; Sigma‐Aldrich). The collected hemolymph was incubated on a glass coverslip for 30 min with apoptotic cells, before being fixed for 1 h with 1.25% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4. Samples were then washed in cacodylate buffer (0.1 M, pH 7.4), fixed again in 0.2% osmium tetroxide in the same washing buffer, and then dehydrated in graded alcohol series. Samples underwent critical point drying and Au/Pd coating (4 nm). Scanning electron micrographs were taken with a field emission scanning electron microscope Merlin, (Zeiss, Oerzen, Embsen, Germany).

Transmission electron microscopy

Third‐instar wandering larvae were bled in 50 μl of Schneider's insect medium (Sigma‐Aldrich) containing 1 μM phenylthiourea (PTU; Sigma‐Aldrich). The collected hemolymph was incubated on a glass coverslip for 1 h before being fixed for 2 h with 2% paraformaldehyde + 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4. Samples were then washed in cacodylate buffer (0.1 M, pH 7.4), fixed again in 1% osmium tetroxide and potassium ferrocyanide 1.5% in cacodylate buffer. After several washes in distilled water, samples were stained in 1% uranyl acetate in water, washed again, and then dehydrated in graded alcohol series (50, 70, 90, 95, 100%). Embedding was performed first in 1: 1 Hard EPON and ethanol 100%, and afterward in pure EPON, before being embedded on coated glass slides and placed at 60°C overnight. Images were acquired with a FEI Tecnai Spirit 120 kV (FEI Company, Eagle, The Netherlands).

Binding assay with apoptotic cells

L3 wandering larvae were bled into Schneider's insect medium (Sigma‐Aldrich) on a previously chilled glass slide. After larva bleeding, macrophages and apoptotic cells (CellTracker™ Red CMTPX Dye) were incubated directly on the prechilled slide, in cold Schneider's medium, on ice for 60 min. After fixation in 4% paraformaldehyde PBS, Alexa Fluor488‐phalloidin staining (Molecular Probes™) was performed. Finally, cells were stained with 1/15,000 dilution of DAPI (Sigma‐Aldrich) Fluorescent preparations were mounted in Dako fluorescence media.

Ex vivo larval macrophage phagocytosis assay

Ex vivo phagocytosis assays of red apoptotic cells, Alexa Fluor™ 488 S. aureus Bioparticles™ or pHrodo™ Red S. aureus Bioparticles™ conjugate for Phagocytosis (Invitrogen) were performed as follows. Five L3 wandering larvae carrying the HmlΔgal4,UAS‐GFP macrophages marker were bled into 150 μl of Schneider's insect medium (Sigma‐Aldrich) containing 1 μM phenylthiourea (PTU; Sigma‐Aldrich). The macrophage suspension was then transferred to 1.5 ml low binding tubes (LoBind, Eppendorf, Hamburg, Germany). The samples were incubated, respectively, with 2 × 10⁷ Bioparticles®‐Texas Red® Conjugate from S. aureus Wood (Invitrogen), 1 × 10⁶ Red‐labeled apoptotic cells or 10⁵ pHrodo™ Red S. aureus Bioparticles™ for 30, 60, or 120 min to enable phagocytosis, and then placed on ice in order to stop the reaction. Phagocytosis was quantified using a flow cytometer (BD Accuri C6 flow cytometer, Becton Dickinson biosciences, Franklin Lakes, NJ, USA) in order to measure the fraction of cells phagocytosing, and their fluorescent intensity. 75 μl volume was read in ultra‐low attachment 96‐well flat‐bottom plates (Costar no. 3474, Corning, Midland, NY, USA) at medium speed (35 μl/min). In a first step, macrophages were identified using the HmlΔgal4UASGFP live staining. The fluorescence intensity of single macrophages was measured in the green channel with 488nm laser and 530/30 standard filter. The Red signal of apoptotic cells, Alexa Fluor™ 488 S. aureus Bioparticles™ (Invitrogen), or pHrodo™ Red S. aureus Bioparticles™, indicative of macrophages with effective phagocytosis, was monitored with 488 nm laser and 585/40 standard filter. At least 2,000 cells per genotype and per assay were analyzed. Results are an average of three independent experiments.

The phagocytic index was calculated as follows:

$$\mathrm{Fraction}\mathrm{of}\mathrm{hemocytes}\mathrm{phagocytosing}(f)=\frac{[\mathrm{number}\mathrm{of}\mathrm{hemocytes}\mathrm{in}\mathrm{fluorescence}\mathrm{positive}\mathrm{gate}]}{[\mathrm{total}\mathrm{number}\mathrm{of}\mathrm{hemocytes}]}$$

$$\mathrm{Phagocytic}\mathrm{index}\left(\mathrm{PI}\right)=\left[\mathrm{Mean}\mathrm{fluorescence}\mathrm{intensity}\mathrm{of}\mathrm{hemocytes}\mathrm{in}\mathrm{fluorescence}\mathrm{positive}\mathrm{gate}\right]\times f.$$

Image analysis and quantification

All images used for quantification were captured with a Zeiss LSM700 microscope, and all analyses were performed using ImageJ. For quantification of the fluorescence signal intensity, the fluorescent images were first converted to 8‐bit images, and the total intensity value with an identical threshold was captured and measured with ImageJ. The freehand selection tool in ImageJ was used to capture and measure the area of the macrophages. Colocalization analysis was done with the ImageJ plugin “Just another Colocalization Plugin” after channel splitting and background subtraction. Rr (Pearson's correlation coefficient), Ch1:Ch2 ratios, M1 and M2 (Manders' colocalization coefficient for channel 1 and 2) were tabulated for each image.

Crawling assay

For the crawling assay, wandering third‐instar larvae were used. Each larvae was taken out of the wall of the vial using a paint brush and placed onto a 10 cm Petri dish plate containing 1.5% agar. A transparent, 1.5‐cm‐wide plastic ring was placed on the outer rim of the agar to prevent the animal from crawling to the edge of the plate. The animal was then left on the plate for at least 1 min to acclimate to the media. Total larval movement was followed for 1 min at 25°C.

Lifespan and behavioral assays

Lifespan experiments were repeated independently at least three times using two cohorts of 20 male flies per genotype/treatment each time. Freshly emerged flies were allowed to mate for 2 days at room temperature and sorted according to sex and genotype. Experiments were performed at 25°C, and flies were flipped to fresh vials every other day using standard medium. For climbing assays, flies were gently tapped to the bottom of a tube and filmed with a digital camera. The percentage of flies climbing above 7 cm within 10 s was calculated.

Statistical tests

Each experiment was repeated independently a minimum of three times (unless otherwise indicated), error bars represent the standard error of the mean of replicate experiments. Data were analyzed using appropriate statistical tests as indicated in figure legends using the GraphPad Prism software. Significance tests were performed using the Mann–Whitney test. For experiments with more than two conditions, significance was tested using ANOVA test followed by post hoc Dunnett's multiple comparison tests. P values of < 0.05 = *, < 0.01 = **, and < 0.001 = ***.

Author contributions

BP, SR, and BL conceived and designed the experiments and wrote the paper. BP, SR, KH, FM, ER, RH, MP, SK performed the experiments. BP, SR, BL, EK analyzed the data.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Review Process File

Expanded View Figures PDF

EMBO reports (2021) 22: e52262.

Data availability

No data were deposited in public databases.